Can the triac be skipped for light loads?

[djulien]

The MOC3023 specs say that the output side is good up to 400V and 1A. A single 100 ct. mini-light string draws 1/3 A, and a 50 ct. only draws 1/6 A. So, could I just skip the triac and connect the output side of the opto in series with a mini-light string? Or is the cold resistance of the mini-lights so low that this would exceed the 1A thru the opto output for the first instant that it is turned on? (Is there a way to measure that?)

The MOC3023 datasheet says that the "bilateral switch performs the functions of a triac", which implies that this might be possible with a light enough load. I realize that this arrangement would give up the latching feature of the triac, and that I would need to use PWM to keep the opto turned on long enough to light up the mini-lights (unless a latching opto were used).

thanks

don

[P. Short]

There was a lot of discussion about this on computerchristmas.com two or three years ago. One of the specs strongly implies that you cannot do this. However, someone (Ernie Horning, I think) spoke with one of the manufacturors, did some testing, and came to the conclusion that it would work.

[djulien]

There was a lot of discussion about this on computerchristmas.com two or three years ago. One of the specs strongly implies that you cannot do this. However, someone (Ernie Horning, I think) spoke with one of the manufacturors, did some testing, and came to the conclusion that it would work.

Phil, thanks for the lead. What keywords should I search for? I tried a couple of searches and found a lot of triac and opto discussion, but not that specific subject.

If the cold current thru an incandescent is 10x the normal current, which would put it over the 1A limit, should I go ahead and put a resistor in series, or can the opto handle a big spike as long as it is short enough?

don

[awhaley]

This is interesting. If you play with it and make it work, let us all know!

Low cost, single string per channel ssrs and controllers could be interesting. :) I'm imagining things like a Renard-8 with onboard optos built into each light arch, so it has its own low cost control onboard.... Or a 32 channel Mega-tree controller with onboard optos switching single strings....


Art

[P. Short]

Here are some threads that touch on that topic..there may be others as well.

http://computerchristmas.com/ForumBoard/read.php?f=4&i=4904&t=4904

http://computerchristmas.com/ForumBoard/read.php?f=4&i=4187&t=4187

http://computerchristmas.com/ForumBoard/read.php?f=4&i=4355&t=4355

[awhaley]

As always Phil, you're full of useful info. :)

The threads there are talking about how the MOC 30XX is rated around 60ma and they're all talking about using it to drive a single c7...

But I'm seeing the same thing that Don did...

http://www.isocom.com/datasheets/db91045.pdf

says forward current is 1A peak... Is there someone more familiar with triac and opto specs that can tell if we're reading the right things?

Thanks!

Art

[P. Short]

AFAIK the maximum current for the input diode is around 50-60 mA for all of these devices, and the peak current for the output diac/triac is the 1A number.

[RavingLunatic]

Art,

That datasheet leaves alot to be desired. If you check the one put out by Fairchild you will see that the peak is shown at a 1ms PW at 120Hz. Not a very useful amount of time on for our needs.

The max continous current shows as 60mA

Also their datasheet contains the very clear note:

Note: This optoisolator should not be used to drive a load directly. It is intended to be a trigger device only.



http://www.fairchildsemi.com/ds/MO%2FMOC3023M.pdf

[P. Short]

I'm not advocating that it be used this way, merely pointing out that this has been discussed before. YMMV.

[awhaley]

Thanks Raving! That is a far superior data sheet to the one I was looking at.

I think it's an interesting enough idea that one of us should blow up so optotriacs to see if it really could work. I've got a couple of projects ahead of this on the table.... but some night when PIC assembly is making me want to Rotate my Head Right through the Carry Bit, I may just decide that potentially smoking an opto sounds relaxing. :) If anyone gives it a try first, I'd love to hear.

Art

[djulien]

Thanks, Phil and Raving, for those links! Do you know if anyone ever built a display board that way? Do you know if that forum is used anymore? It seems like it pre-dates DIYC.

don

[P. Short]

AFAIK the output of the MOC302x latches on, just like a triac. So you don't give up this feature.

There are also some fairly inexpensive triacs out there in TO92 and SOT223 cases. They are usually rated for either 0.8A or 1A, although for power dissipation reasons I wouldn't drive them anywhere near that current. They end up costing $0.20 or less in 100 quantities. The next choice up the price scale would be a 2N6075B or similar part.

[awhaley]

I was sending in a mouser order today, so I added a couple of VO 3053 opto triacs to my order... it's a $.88 opto that's rated to 1 amp continuous... At this point it's no real cost savings over a 50 cent opto and a 30 cent 1amp triac, but it would save parts (triac and gate resistors), which might matter to somebody in an 'all onboard' design...

This chip also carries the mysterious advice on the data sheet that it should only be used to trigger triacs, not to drive a load directly... but it's worth a buck sixty to give it a couple of tries in my opinion, since I'm already paying the shipping. Whenever I get around to plugging it in, I'll let you all know how it smells.... ;)

Art

[awhaley]

I just got an email from the nice folks at Vishay, suggesting the part

http://www.vishay.com/docs/81842/vo3526.pdf

for consideration. It's a 16 pin DIP package (why I can't tell... there are 4 unconnected pins... they may have just needed the package size for heat dissipation?) that includes BOTH an opto, and a power triac. It's designed to switch loads, and rated up to 1A.

It appears to be quite new, and I can't find them for sale anywhere yet... so I'll keep my eye on it. If the price winds up right, they could be a cool low power SSR option.

Art

[djulien]

AFAIK the output of the MOC302x latches on, just like a triac. So you don't give up this feature.

I wondered about that. At first I thought it should, since it is basically a triac, but that led me to the question of why PWM is needed in some cases (ie, driving a 120 VAC relay coil or LED strings) - if the opto is latched, it should keep the triac on (or keep trying to turn it on again) even if the triac does not have enough load to stay on, rather than the triac turning off again. Are we not running enough current thru the opto to latch it in the standard SSR circuit?

don

[djulien]

I was sending in a mouser order today, so I added a couple of VO 3053 opto triacs to my order... it's a $.88 opto that's rated to 1 amp continuous...

Is there a typo on that spec sheet? "Peak repetitive surge current" says 1A for pulse width 100 msec, 120 pulses per sec (which would be all the time), but on-state rms current only says 100 mA.


but it would save parts (triac and gate resistors), which might matter to somebody in an 'all onboard' design...

I'm hoping to find something that will drive 170 mA continuous (mini-lights) at 24 V so I can make a 1K pixel "mega-triks" - not all pixels on at the same time, of course :)

don

[RPM]

Another thing to consider is that even though the specs say it can handle 1 amp surge, it also will state for how long, which is in microseconds.

Also, there will be a certain amount of "on resistance" which will cause the opto to heat up... this is the same thing that causes a triac to heat up as current is drawn through it. Too much heat will damage the opto, much in the same way it can damage a triac. The MOC3023 data sheet shows a total power dissipation of 330mw, which is derated according to ambient temperature, and as the temperature goes up the current drive capability goes down considerably.

An incandescent lamp will draw much more current when first turned on than the measured continous on current will be, so this needs to be factored in as well.

This being said, it would probably work to just use the opto output if running LED lights, since they draw so little current and don't have the "inrush current" effect of incandescents, but I would be careful trying them for incandescents... those little boxes fit lots of smoke in them ;)

Just my .02

RPM

[djulien]

those little boxes fit lots of smoke in them ;)

Hmm, that's an interesting display idea - "smoke-triks". I could put a bunch of then on a board, and then turn them on in various sequences to let out puffs of smoke. Kinda like water jets, but smoke instead. :) I would have to be careful not to use them all before the sequence is over, though.

From the links that Phil provided, it sounds like C7s will actually work, but if I want to use mini-lights, I'll need to use something like the VO3526 part that Art found. The cheaper triacs Phil referred to would reduce the cost, but I am also hoping to reduce the parts count, in order to simplify construction.

don

[ErnieHorning]

I did some quick testing at room temperature. Using the wonders of $10,000 equipment I can give you a couple of thousand words relatively quick.

I took these at 60 mA, 100 mA and 150 mA continuous resistive current and exactly 120VAC RMS. I applied exactly 5 mA to the input diode and checked whether I could change the state of the DIAC. I waited approximately one minute before checking or taking any measurements even though the temperature stabilized in about 10 seconds. It appeared as though the maximum current was 125mA before latching up and staying on.

Using a sample of one, it appears that the safe upper limit would be 100 mA. I didn’t leave it sit for more than a minute so I don’t know if 100 mA would work for an extended time. Based solely on this limited test, it appears that you maybe be able to control 50 count ES mini’s which draw 100mA. You could also split 100 count ES if the effort was worth it to you.

[Warlock]

Great work Ernie...What about driving a set of leds that would be about .33 amps for any extended time frame..Sorry I do not know the mv values...But a 70 count of leds is only .33 of an amp.If you only use 70 per channel would it work for leds to latch ???If it does then it might be a solution to shortening leds that we all can benifit from.
Thanks again for your work on this idea..
Joe..

[djulien]

Using the wonders of $10,000 equipment I can give you a couple of thousand words relatively quick.

I took these at 60 mA, 100 mA and 150 mA continuous resistive current and exactly 120VAC RMS. I applied exactly 5 mA to the input diode and checked whether I could change the state of the DIAC. I waited approximately one minute before checking or taking any measurements even though the temperature stabilized in about 10 seconds. It appeared as though the maximum current was 125mA before latching up and staying on.

Using a sample of one, it appears that the safe upper limit would be 100 mA. I didn’t leave it sit for more than a minute so I don’t know if 100 mA would work for an extended time. Based solely on this limited test, it appears that you maybe be able to control 50 count ES mini’s which draw 100mA. You could also split 100 count ES if the effort was worth it to you.

Ernie, that is some neat equipment you have there!

In your experience, is there much variation between manufacturers (or do they all come from the same place anyway)?


What about driving a set of leds that would be about .33 amps for any extended time frame..

Joe, what kind of LEDs are those? I thought LEDs draw a lot less (or are you refering to a super-flux LED)?

don

[Warlock]

Sorry you are correct it is .033 amp per 70 led string..If you could go around the triac in an ssr then it would make it easier to do my ideas...

[RPM]

If you are trying to save space you can always use something like this:
http://mouser.com/Search/ProductDetail.aspx?qs=CX134%252bdLMDF4X9k4CH7R5g%3 d%3d

It's basically an opto and triac rated for 2 amps in one package. It's a little pricey, but in quantity it may be worth a look, that's if you're trying to keep the parts count down.

RPM

[ErnieHorning]

Sorry you are correct it is .033 amp per 70 led string..If you could go around the triac in an ssr then it would make it easier to do my ideas...Well you’d certainly be able to run at least three strings of LED’s though I’m not sure how this would make anything easier.

[Warlock]

Hi Ernie...I was refering to only using 1 string per channel and getting it to latch.I've heard of too many people having issues with leds not latching cause of the load being to low.So if you can get around the triac it would make it easier if you went 1 string per channel..Then a 128 board with dimming wouldn't be nearly as costly and it limits the parts count.
With that idea in mind what would the max load be if you went this way as far as amperage ???

[djulien]

If you are trying to save space you can always use something like this:
http://mouser.com/Search/ProductDetail.aspx?qs=CX134%252bdLMDF4X9k4CH7R5g%3 d%3d

I like the compact package, but I was not so concerned about the actual space (although that is always a consideration). I was actually trying to save $ *and* reduce the number of components to be connected - I realize that these are opposing goals. :(


The MOC3023 data sheet shows a total power dissipation of 330mw, which is derated according to ambient temperature, and as the temperature goes up the current drive capability goes down considerably.

Regarding the peak pulse current vs. constant RMS current, I'm trying to get a clearer picture of the curve between those 2 points for a given pulse width or duty cycle. I haven't found that curve on the spec sheets, but there are some others related to RMS current. Is the peak pulse curve likely to have the same shape as the "RMS on-state current" curve? For example, I want to estimate the max pulse current if I use a pulse width of, say, 8 msec 120x/sec (ie, 50% of the AC cycle) instead of the 1 msec shown on the spec sheet. I'm hoping that the max pulse current at such a high duty cycle is significantly higher than just the RMS max current.

EDIT: I found a "Surge peak on-state current versus number of cycles" chart in the Triac specs. It is reasonable to assume that the curve for the opto is similar, since the output side is basically the same as a triac?

Sorry to keep asking so many questions about this.

don

[g2ktcf]

gents...beadboard the thing and test it! The parts are not that expensive if you smoke them.

[ErnieHorning]

The parts are not that expensive if you smoke them.I didn’t try a gross overload but I did go up to 300 mA. It got hot but it wasn’t in the smoke range. It was dang hot but it was still in the touching range. Doing nothing but dropping the load current back below the latch up current would allow it to turn off and operate correctly again.

Maybe I should crank up the current until it blows up, just for the curiosity factor and deviant enjoyment. Then we’ll know what it takes to destroy it beyond recovery and what the recovery headroom is.

The one thing that I couldn’t test, but suspected, is well its up at the higher current, it seemed as though the turn off time was longer. Kind of like a temporary latch up that cleared itself quickly.

[djulien]

gents...beadboard the thing and test it! The parts are not that expensive if you smoke them.

I am planning to do that. I agree that the cost of a few parts is not significant. I was just trying to get a better understanding of the parameters involved before I order more parts, due to the turn-around times.

don

[djulien]

I found a spare opto, so I ran some tests today - it works!

I am getting 185 mA at 28 VAC thru the MOC3023 with only a slight warmth, barely detectable. I tried 5 types of tests: constant on at 100%, on/off blinking at 40 Hz (25 msec interval in Vixen), .9 sec on/.1 sec off, ramp up for 3/4 sec, and constant on at 50%. I am assuming that the constant-on cases are the highest stress on the opto, since it needs to turn on and conduct for each AC cycle.

Also, since the opto was actually able to turn the lights on and off for all the tests that try to do so, I am also assuming that this means that the latching is working correctly (since I am using non-PWM firmware).

Since Ernie's temperature tests stabilized after 10 seconds, I only ran each test for a few minutes, although I did end up running a longer constant-on test for an hour while I read some forum posts.

[UPDATE: the test has run for over 2 hours, and the opto is still just slightly warm, so I'll consider it to have passed the test.]

My target application is 10 ct. mini-light strings - I want finer granularity than a 50 ct. string, more diffusion than a single C7/C9, and lower cost than a super-flux LED - so I reduced the voltage from 120 VAC down to 24 VAC to match the bulb count reduction (I'm actually using a 24 V transformer, but it seems to be a little off). I'm guessing that the opto temperature would be higher with 120 VAC, since there would be greater power dissipation.

I measured current by putting a 1 ohm resistor in series with the mini-lights and then measuring the voltage drop across it. I think I read that technique in one of Xmus' docs.

So, it looks like I can use a MOC3023 + 1 resistor on the input side as a smaller and cheaper SSR for lighter loads in this specific scenario (spec sheet warnings duly noted). I may be able to hit my target of $0.50 per charlieplexed pixel afterall.

Thank you all for the helpful info, links, spec sheets and part#s, especially Ernie for running the opto thru that fancy test machine!

don

[ErnieHorning]

I tested my sample of one at 24 volts AC and found that the maximum current was 280 mA.

[djulien]

I tested my sample of one at 24 volts AC and found that the maximum current was 280 mA.

Very good. That means there is some safety room for a 10 ct. mini-light string running at 170 mA, and even more for LEDs.

don

[djulien]

Since I hadn't fried the opto yet, I tried another test at 120V (155 mA). Once the opto turned on the lights, they stayed on and wouldn't turn off until I disconnected power.

Is this probably because the inrush current * voltage was too high? The opto was not warmer than room temp, so I don't think it was a heat problem.

If so, then is the way to resolve this to put a resistor in series to reduce the inrush loading?

don

[ErnieHorning]

Since I hadn't fried the opto yet, I tried another test at 120V (155 mA). Once the opto turned on the lights, they stayed on and wouldn't turn off until I disconnected power.

Is this probably because the inrush current * voltage was too high? The opto was not warmer than room temp, so I don't think it was a heat problem.

If so, then is the way to resolve this to put a resistor in series to reduce the inrush loading?

don
Referring to my original post (http://www.doityourselfchristmas.com/forums/showpost.php?p=56886&postcount=19) that I measured at 120 volts AC:


It appeared as though the maximum current was 125mA before latching up and staying on.You’re over the maximum current. If you used 200 mA ES 100 count mini lights and then cut them in half so that you had the two separate strings, you’d only have 100 mA. These opto’s should be able to work. They also have the higher in-current, just like a normal TRIAC.

[Warlock]

So is someone going to post the latching ablities of this idea and how they ramp as well so we can see if this is an option for onboard ssr's for leapers,mega trees with single string control ?? If it does work this opens up a whole new area for how one can attach leds and mini's directly to the optio instead of the triac...Also to if someone is willing to show their design for this it would be greatly appreciated as I'd love to try other ideas using this method like (shorter strings without a resistor to make up the difference)...I see my star design coming to life if this works..Might be the solution we have been searching for..

[djulien]

Referring to my original post (http://www.doityourselfchristmas.com/forums/showpost.php?p=56886&postcount=19) that I measured at 120 volts AC:

You’re over the maximum current. If you used 200 mA ES 100 count mini lights and then cut them in half so that you had the two separate strings, you’d only have 100 mA. These opto’s should be able to work. They also have the higher in-current, just like a normal TRIAC.

Ernie, yes I realize that, but *all* of my test cases were over the max current (approx numbers are shown below), and they all worked at lower voltages (8VAC, 16VAC and 24VAC), but not at 120 VAC. So it seems like the voltage (or maybe power) comes into play somehow. I thought it would be the power, except I could not feel any warmth on the opt at 120VAC, so that leads me to think that it is voltage-related, although this could be another bad assumption on my part. The spec sheet says max power dissipation of 300 mW, but at > 150 mA I was over that limit for many of my test cases, so they should not have worked.

I did not have ES minis on hand, so I used regular minis. I measured the voltage drop across a 1 Ω resistor connected in series to get the current. Unless I measured incorrectly, the minis I used had these characteristics:
2.7 VAC average per bulb draws ~ 195 mA (very bright, 27.4 VAC across 10 bulbs)
2.4 VAC average draws ~ 175 mA (full brightness)
1.8 VAC draws ~ 155 mA (nearly full brightness)
1.2 VAC draws ~ 125 mA (noticeably dim)
1 VAC draws ~ 110 mA (barely lit)

don

[Warlock]

What amperage are you using for your tests ??? If you lower that maybe the voltage won't be an issue...Also to the voltage issue might be true but how will you know unless you lowr the amperage and see if it still operates that way ??? Maybe you could run that test and see..
Just an idea as there has to be a difference with 120 v at 15 amps and 120 v at 2 amps..Might not hit it as hard and cause the spike that it seems unwilling to release..

[djulien]

So is someone going to post the latching ablities of this idea and how they ramp as well so we can see if this is an option for onboard ssr's for leapers,mega trees with single string control ?? If it does work this opens up a whole new area for how one can attach leds and mini's directly to the optio instead of the triac...

Warlock, I do not know enough about the underlying electronics theory and components to be able to write up general-purpose guidelines, but I do plan to write up something within the context of the project I am working on (if it actually works), and will include schematics, firmware, and various other relevant info (which would probably be a brief summary of this thread and another one about opto timing (http://www.doityourselfchristmas.com/forums/showthread.php?t=6346)). I don't want to write up something that I have not actually tested and run successfully through a season, but OTOH I agree with you that it does open up some interesting possibilies if it works, and that it would be helpful to other people to have an earlier write up (especially early in the year), so this creates a dilema for me.

I found it interesting/amusing that this is not a new idea. Phil pointed me to some links (earlier in this thread) that showed that other people were planning to do this a few years ago, but I didn't find examples of it actually having been done, so I'm thinking that there is still some kind of road-block to this approach. Or, maybe since I am not strictly following the manufacturer's guidelines for this application, there was some variation in the results for other people or in other contexts. But this is a DIY forum, so I suppose that a simple YMMV warning should be sufficient.


Also to if someone is willing to show their design for this it would be greatly appreciated as I'd love to try other ideas using this method like (shorter strings without a resistor to make up the difference)...I see my star design coming to life if this works..Might be the solution we have been searching for..

In lieu of a write-up at this time, here is a summary of the general idea I am experimenting with. There are actually several projects, but the first one is that I am wanting to build a low-cost "mega-triks" - a large sign-board maybe 30' x 8' consisting of about 1000 individually addressable pixels that are each 6" x 6". Nothing new there - this can already be done using the other projects described in this forum. But for me it would be cost-prohibitive, so in the spirit of DIY, I'm trying to lower the cost per pixel (to about 1/10). This leads me to try using the MOC3023 as the SSR and fractional mini-light strings for the lights. The latching/turn-on feature of triacs when used with AC also means that control signals are inherently latched, which then means that a simple charlie-plexing controller can be used to drive this thing rather than a regular row/column scan. With only one opto LED turned on at a time, power drivers are not needed on the controller. I think the controller can simply be a pair of PICs and an RS232/485 driver (for Renard), making the controller very simple - if I use PICs that have 16 I/O pins, that would be 32 x 31 = 980 pixels. Now there are some limitations to this approach - PIC bandwidth and memory size, and opto turn-on times will limit the number of pixels that can be on at one time, but that limitation existed due to power consumption of the pixels anyway. I am only wanting to do sparsely populated graphics or text, so most of the pixels will be off anyway.

So, there are 2 parts to this project: cheaper/larger pixels, and simpler controller. Either of these parts could be used separately - for example, the MOC-as-SSR approach could be used in place of existing AC SSRs for cases where the AC load is light, or the charlie-plexing controller could be used to turn on regular AC SSRs in scenarios where relatively few SSRs within a cluster need to be on at the same time. I suppose it might also work for LED strings, but I have not given consideration to that since they don't seem to meet my other requirements related to diffusion and viewing angle and distance.

I think your idea of shorter strings without a make-up resistor will definitely work for mini-lights as long as you adjust the AC voltage accordingly (approx 2 - 2.5 VAC per bulb for the 100 ct mini-lights, 6V per bulb for the 20 ct mini-lights, etc), and should also work for LEDs as long as you stay within the max current, although I think I read in one of the posts that a resistor is always needed in series with the LEDs.

Sorry to cause such a long, drawn out discussion. Hopefully some of this info is useful to you, even though it is not in final form yet.

don

[Warlock]

Thank-you for the very useful info Don..My theory tho is there always has to be a way of doing it.Just a matter of figuring it out..If you guys test this and all of us think about it (with the more experenced voices being the ones of reason) then I'm sure we all will work out the kinks.As far as the leds always needing a resistor not always the case.If you are limiting the voltage with a resistor off the opto anyway what is the need for the one that is on the leds in the first place ??? In theory there isn't a need for one only the one that is the connecting point...Basically that was what I was asking..If it could work as it does in theory as not all things do..
Also to if you were to do a shorter count in leds for a design animation then would it still follow the same rules and would the brightness be the same as with say a full string ???? I know of no one who has tried this (shocked that the crazy light lady hasn't either as I was reading her site last night)..But it would be great info for people who need only a small number of leds on at that specific time to make things easier..
Your mega tricks design sounds great and it should work but if you were to also try the super leds on the market they would also work(less voltage across each give you greatly flexablity to add more) then this approach means you can build more with the same voltage and amperage..Think about it if a string of 100 mini's pulls 0.66 amps at 120 volts and 70 leds only pulls 0.033 amps per string..Guess which wins that battle of keep hydro off your back??? Leds do everytime...The issues with your angles and visablity at a distance can easily be overcome by the different leds that are on the market..It all comes down to playing aorund with what does what in the real world as opposed to what is written by the manufactures of this stuff..Someone has to play around and see what specs are really true and which are false and replace the false with real world specs...Maybe the manufactures can help us with this if we ask them to do alot of testing for us with certain guidelines that we can understand they use...Only way I can see this really coming into it's own..
Again sorry for the novel but I think in the end with all the ideas it will really be fruitful ...
Again thanks to all who are testing these theeories and ideas..We all learn from this info..
Joe..

[P. Short]

If all you are interested in is dimming those 1000 lights to perhaps 8 levels of dimming, there are other ways to do it that are perhaps cheaper and do not involve driving chips in non-standard ways.

First of all, the best price that I see at mouser for MOC3023 optos in quantities of 1000 is $0.267 each, which extends to $267.

Another way to do it would be to use 1000 MAC997A6G triacs, which seem to cost about $0.126 each in that quantity (or $126). These parts are good for 0.8A at 400V, which should be good for your app. Then you need to worry about how to drive these parts. There are various ways of doing this, but my estimate is that it will take about $0.04 per output, or around $40. Next, you will need optoisolators, I think that you could get by with less than 40 4N35 chips for around $0.30 each, or $12.00 extended. Lastly, you need an isolated power supply for those parts, I think that a VBSD1-S5-S5-SIP (from Digikey) should work fine, costing about $5.00. The total parts here (exclusive of the PCBs, connectors, etc) would come to around $183 (plus tax, shipping and handling). The cost of the PICs is somewhat minimal compared with all of the other costs, so I'm not adding it in.

The one thing that I don't like about this approach is that there are logic chips at AC line potential, quite a hazard for the newbie. It would be best to add in an isolation transformer when initially testing this setup.

Edit - performance considerations. What I'm figuring is that there will be around 8 dimmer levels (per your earlier commentary), which translates to a slot width of about 1 mS each (8 mS / 8 levels). Each of these slots will be further divided into 32 sub-slots of 30 uS or so each. The 4N35 optos should be good for this, although you will probably need to add a 100K or so base-emitter resistor to get this speed.

[djulien]

Joe,


My theory tho is there always has to be a way of doing it.Just a matter of figuring it out..

I agree - that is why I haven't given up yet. :)


As far as the leds always needing a resistor not always the case.If you are limiting the voltage with a resistor off the opto anyway what is the need for the one that is on the leds in the first place ??? In theory there isn't a need for one only the one that is the connecting point...

Sorry, I didn't say it well, or misunderstood you. I had in mind raw LEDs - if there is already a resistor in series with the LEDs as part of an LED string, then I agree that you don't need another one.


Also to if you were to do a shorter count in leds for a design animation then would it still follow the same rules and would the brightness be the same as with say a full string ????

From what I've read, the brightness of LEDs depends mostly on the average current thru them, and the effective viewing angle. A shorter LED string would give off less total light, but at any given place along it it could have the same brightness as the original string, so I think it comes down to viewing again.

It sounds like most people cut the LED strings at parallel-circuit boundaries that preserve the 120 V circuit, and some cut them at other lengths and adjust the series resistor so that 120 V can still be used, but I thought I've read that someone cut them at different places and then used lower voltage. Maybe I misread it, but that should certainly work as long as the supply voltage is higher than the total voltage drop of the LEDs in series and there is a current-limiting resistor in there somewhere.


Your mega tricks design sounds great and it should work but if you were to also try the super leds on the market they would also work(less voltage across each give you greatly flexablity to add more) then this approach means you can build more with the same voltage and amperage..Think about it if a string of 100 mini's pulls 0.66 amps at 120 volts and 70 leds only pulls 0.033 amps per string..Guess which wins that battle of keep hydro off your back??? Leds do everytime...The issues with your angles and visablity at a distance can easily be overcome by the different leds that are on the market..

I thought about using the LEDs, I agree that they use less power and would work and I think I will use them for some of the variant projects, but they were not the most attractive choice for this particular project, because the super LEDs draw enough current to need some additional driver components, raising today's per-pixel price substantially. There was another consideration which I did not state earlier - a super LED would be a point-source of light, whereas a partial mini-light string allows the light source to be spread out more, which helps a lot with the diffusion and viewing angle problems with LEDs. I am actually wanting multiple point-sources of light with air between them for each pixel, so that I can safely look right at them and still be able to see objects behind the display (just a style issue).


It all comes down to playing aorund with what does what in the real world as opposed to what is written by the manufactures of this stuff..Someone has to play around and see what specs are really true and which are false and replace the false with real world specs...

Good point - the MOC spec was rather misleading in a few areas.

don

[Warlock]

Quote:
Originally Posted by Warlock View Post
Also to if you were to do a shorter count in leds for a design animation then would it still follow the same rules and would the brightness be the same as with say a full string ????
From what I've read, the brightness of LEDs depends mostly on the average current thru them, and the effective viewing angle. A shorter LED string would give off less total light, but at any given place along it it could have the same brightness as the original string, so I think it comes down to viewing again.

It sounds like most people cut the LED strings at parallel-circuit boundaries that preserve the 120 V circuit, and some cut them at other lengths and adjust the series resistor so that 120 V can still be used, but I thought I've read that someone cut them at different places and then used lower voltage. Maybe I misread it, but that should certainly work as long as the supply voltage is higher than the total voltage drop of the LEDs in series and there is a current-limiting resistor in there somewhere.

True which is the point I'm making your basically stuck at a certain point with your design.For example if you wanted to do a shooting star effect then say 1 star only needs 10 leds...The next needs 15 leds and the final in a 3 star design needs 20 leds...Ok if you do the resistor calculations then it works out and the brightness is even.But if you could hook it up to the opto and have it blinky flashy then all is better.All your asking for at that point is on and off not a fade or anything else..Unless like me you wanna push it even further into the simple question of leaving people wondering how the heck did he do that ??? I love it when I do something and people are left with that question...The main reason for me looking into this idea any many others in the first place...But back to my point...Since you wanna be so exotic with the leds there is no reason it shouldn't work.Voltage across the led is constant and at max for the brightness of the led you are using.So the level of brightness shouldn't be changed but has anyone tried this yet ?? If so how did it work out ?? Are the calculations correct or do they require tinkering to get them just as bright as a full string of 70 leds..

As for your other point about viewing angles..I agree to a point only because from what I've red about it is solely dependany on the lens that is used to difuse that light.With a mini it is clear glass or painted glass but with an led it is a plastic of some sort.So diffusion properties are not as consistant as with glass.But these newer designs outta hong kong seem to be really sweet with the way they are produced now.They are clearer and more of a pure color from batch to batch with viewing angles being alot better..So it is worth another look..ONly real issue I see with the leds these days is that stupid flicker..Even off of 120 direct it still does it..Drives me nuts and really distracks the eye from the beauty of them...Anyway there are always trade offs with whatever you choose to do..It all boils down to which ones you can live with as some are simply inherent with the design of the product or the design of the effect you are trying to achieve..

I still want to stay with a renard design..I was simply looking and trying to explore another solution..An isolation transformer is the best way to go for constant power at whatever wattage you need which is what I was designing...But taking away the triac is another good way to go cause when with more sensative triacs they sometimes under smaller loads have a hard time opening...That is the real issue that is trying to be over come...As for me and my leapers I'm going direct off the opto for the best effect..Now if someone can tell me if they ramp without flickering that would be nice...

Thanks guys lots left to do on this one..
Joe..

P.S. Thank-you Phil I do realize there are other ways to do this kinda design and make this happen..I was simply trying to work within the renard designs.I wasn't trying to alter anything other then the triac for my purposes and also include a transformer for whittle down the direct 120 volt supply to a more managable and suitable level.
Thank-you again Phil for pointing out design issues that I am trying to overcome within the renard designs.Excuse me within the renard designs that I wish to use is more like it and ones I wish to design around the same protocol that is used now.

[djulien]

If all you are interested in is dimming those 1000 lights to perhaps 8 levels of dimming, there are other ways to do it that are perhaps cheaper and do not involve driving chips in non-standard ways.

Yes, 8 or even 4 levels. In addition to cost, I'm also trying to reduce the number of parts so that assembly time is not too bad (this is probably a contradictory goal to reducing the cost).


Then you need to worry about how to drive these parts. There are various ways of doing this, but my estimate is that it will take about $0.04 per output, or around $40. Next, you will need optoisolators, I think that you could get by with less than 40 4N35 chips for around $0.30 each, or $12.00 extended.

Is the idea to put a logic gate or diodes at each intersection to turn on the triac? This sounds interesting. I may be too influenced by the original SSR circuit.


The one thing that I don't like about this approach is that there are logic chips at AC line potential, quite a hazard for the newbie. It would be best to add in an isolation transformer when initially testing this setup.

Are the logic chips still after the optoisolators, so that the area at risk is confined?

don

[P. Short]

The logic would be after the opto-isolators. The reason for putting it there is that the optos are still relatively expensive compared with the logic.

[Warlock]

Here is a link I think you will find also interesting about the leds we are trying to play with.
http://www.lsdiodes.com/shop/index.php?main_page=page_4

It also brought to my mind after reading it why some flicker and also why some people have a hard time getting them to work exactly right with ac current...The Crazy Light Lady make sence with the way she explains it.At the request of her I will not repost what she said but feel free to check her site out to see the issues that still revolve around leds and dimming,fading etc...

http://www.crazylightlady.com

Enjoy..
Joe..

[djulien]

As for me and my leapers I'm going direct off the opto for the best effect..Now if someone can tell me if they ramp without flickering that would be nice.

I did do a ramp test at 24 VAC with 10 ct. mini-light and it looked okay, but I was mainly watching for smoke at the time and checking the opto's temp so I wasn't really focusing on whether there was flicker.

don

[Warlock]

Do you have leds and can test those for flicker ??? On a simply slow ramp to see plus try simply on and of..Say on for a 5 count then off for a 5 count and back on again with leds to see if you see any flickering...Mine flicker like a pig and can be seen in my videos yet to be aired on utube cause of them taking em all down so often..
Thanks.
Joe..

[awhaley]

You might want to take a look at the project at:

http://macetech.com/blog/node/36

It's a 1200ish node LED display (384 RGB pixels) done with 12 TLC5940 chips. Apparently one TLC5940 is bussed to the cathode of 6 LEDs, and the other side of each led goes to one of 6 io pins (through an appropriate transistor or mosfet, I hope!) to select which bank of LEDs you're sending pwm data to. A pretty elegant and cost effective solution, if you were thinking about using LEDs. (12 ~$4.00 chips, plus anode mosfets and a PIC...) I don't see why this circuit couldn't be used with output mosfets or similar drivers to allow it to switch 24V rectified DC loads if you really do want to go with 10 incandescent lamps per output....

Having looked at the BOM for the LEDtriks system once upon a time, this is probably similar to what it does... but with the addition of pwm.

Anyway. Just wanted to throw out another way of doing what you're thinking about... maybe with less money or parts.

Art

[Warlock]

I posted something simular a bit ago...It's on the boards here but for a video board that links to the video out on the video card.

[awhaley]

One other "Can we leave parts out of the SSR to make it cheaper and simpler" question.... And I can hear people cringing already...

What if you skipped the opto?

Microchip's app notes
TB094 and PICREF-4 both show circuits for doing AC lines dimming without isolation... in fact the circuits look a little scarily unisolated for my taste, with ZC detected by a pin connected directly to AC power via a resistor, the circuit powered off the AC line with only a few diode and resistors between the PIC and 120V (no transformer), and the output of the PIC going through a resistor and nothing else on its way to the gate of a Triac.

I'm sure the extra protection we build into all of our controllers is valuable and not to be avoided for most general purpose use, but in your case where you're talking about 24AC, so there's already some safety factor, what if you looked at just driving .20c 1A triacs directly(with a resistor)?

I've never used a system without isolation, so I can't speak to the drawbacks... you'd surely want to add isolation somewhere to protect the computer from the AC power if several things failed at once.... but it's something to look at.


Art

[Warlock]

The opto protects the computer.Not really the issue I'm refering to...I was simply asking the results if you left out the triac and were simply using 1 set of leds per channel..PLus would you need the extra resistor if you shortened the line..But it also rasises an interesting question about it..Using the PWM code it seems to work better but not flawless for leds.Still a few issues with dimming and flicker so I'm reading..So the issue now becomes either a: are we powering the boards wrong (meaning should we start looking at using dc power for the boards) or b:or would leaving the triac off the design and using 1 string per channel solve that problem ???

[P. Short]

If the 24VAC is already isolated (i.e. generated by using a regular isolated-winding transformer as opposed to an autotransformer or the like) then I don't see any issue with skipping the opto. But I do cringe at the idea of any detailed discussion of this in a forum where there are newbies present who have difficulty understanding what people are saying and have no experience whatsoever with electricity or electronics.

[Warlock]

Hey Phil the only reason I continued with this idea of powering lights off the opto was to give an idea to people after I found out it could be done...Firegod told me last year that was how he was running his so I already knew the answer but wanted one of the more trusted people to post it.Since Ernie did I kept it going simply to ask if you used only 1 string per channel of leds would it be possible to skip using the triac ?? I believe the answer is yes but since a few people were actually worried about smoking the opto they really didn;t wantch the ramp and fades to see if flicker was present..So I requested it as a favor..I plan on using that design for my leapers as my leapers for 2009 will be 8 channels each.6 foot mega tree 48 channels and 4 colors might go led tricks on that not sure yet.That is just a start of a packed 4 color 37,000 led light display using vixen as the control which is why I'm here asking for these tests to be done...Another reason is what others have told me they are experencing as well.If we can make sure it doesn't happen with a simple adjustment for a purpose isn;t that what we are all here for ?? So sorry no disrespect but that is why this thread has been kept going..As for the other part you are refering to I totally agree...Unless you get feed back from the more experenced people such as Phil and others then I wouldn't build anything..Just my opinion...I've built many a 12 volt toys over the years (sum 20 years) but never 120 volt ones...Also manybe a good reason to start with dc stuff..Get the total hang of this before stepping up to 120..Just another idea Phil for your entry level board..Sorry just thinking outloud again..lol..
Enjoy.
Joe..

[djulien]

You might want to take a look at the project at:

http://macetech.com/blog/node/36

It's a 1200ish node LED display (384 RGB pixels) done with 12 TLC5940 chips. Apparently one TLC5940 is bussed to the cathode of 6 LEDs, and the other side of each led goes to one of 6 io pins (through an appropriate transistor or mosfet, I hope!) to select which bank of LEDs you're sending pwm data to. A pretty elegant and cost effective solution, if you were thinking about using LEDs. (12 ~$4.00 chips, plus anode mosfets and a PIC...) I don't see why this circuit couldn't be used with output mosfets or similar drivers to allow it to switch 24V rectified DC loads if you really do want to go with 10 incandescent lamps per output....

Multiplexing and POV don't work too well with incandescent bulbs because they are slow to respond, but for LEDs that looks very cool!

don

[awhaley]

I can imagine that Don. I know PWM does in fact work with incandescent sources, but I hadn't thought about how fast the multiplexing would have to switch to keep from being detected.... My inclination is that at high intensity levels it might look better than at low, because the filament wouldn't have time to cool between cycles if it was very hot?

But you're right... if you want an IncandescentTriks, you probably don't want a multiplexed system... hmmmm. And stringing together 70 PWM driver chips seems a bit excessive... though an option. And I guess it's less than 1000 optos, isn't it? :)

Art

[djulien]

Do you have leds and can test those for flicker ??? On a simply slow ramp to see plus try simply on and of..Say on for a 5 count then off for a 5 count and back on again with leds to see if you see any flickering...Mine flicker like a pig and can be seen in my videos yet to be aired on utube cause of them taking em all down so often..
Thanks.
Joe..

I realize this response is a little late now, as the discussion has continued on, but I thought I'd post a test result for LEDs anyways. I tried a 25 ct C9 LED string, and it did not turn on at all when connected directly to the opto and 120 VAC. My PICs are flashed with non-PWM firmware, so I think the problem is probably that the LEDs do not draw enough current to keep the opto turned on. The box says that this string draws 20 mA. I don't know what number to compare that to on the opto spec sheet.

When the LED string is plugged directly into the outlet, I can see slight flicker thru my peripheral vision, so I expect that I would also see flicker if I managed to dim them. I guess this is not a good test case.

don

[ErnieHorning]

I can imagine that Don. I know PWM does in fact work with incandescent sources, but I hadn't thought about how fast the multiplexing would have to switch to keep from being detected.... My inclination is that at high intensity levels it might look better than at low, because the filament wouldn't have time to cool between cycles if it was very hot?PWM works for incandescent’s or LED. Multiplexing wont work for incandescent’s because they are too slow at turning on, which is why it works so swell for LED’s.

[Warlock]

Thanks Don that has helped alot...Can someone please test this theory and see if it works on boards with pwm flashed pics and see if it does the same thing ???
Thanks guys as always great work...It has so far gone exactly as I expected.No change as of yet..So lets see if it works with a flashed pic with PWM.If it does then we know it is something further down the line as different designs don't show it but some do...Also something to figure out as well..
Again Don thank-you for doing this..It really does mean alot :-).
Joe..

[ErnieHorning]

I tried a 25 ct C9 LED string, and it did not turn on at all when connected directly to the opto and 120 VAC. My PICs are flashed with non-PWM firmware, so I think the problem is probably that the LEDs do not draw enough current to keep the opto turned on.It Would be the same problem as with any TRAIC. You would need to use the PWM firmware.

[djulien]

I can imagine that Don. I know PWM does in fact work with incandescent sources, but I hadn't thought about how fast the multiplexing would have to switch to keep from being detected.... My inclination is that at high intensity levels it might look better than at low, because the filament wouldn't have time to cool between cycles if it was very hot?

Since incandescent bulbs are slow to respond, multiplexing can work as long as the control signal is latched. The triacs or optos do allow that to happen. I meant POV or multiplexing by itself without a latching device wouldn't work too well.


if you want an IncandescentTriks, you probably don't want a multiplexed system... hmmmm. And stringing together 70 PWM driver chips seems a bit excessive... though an option. And I guess it's less than 1000 optos, isn't it? :)

Actually, I'm open to using LEDs, but they seemed to cost more. I was considering using a MOC3023 at $0.30 + a 10 ct. mini-light string at $0.10 = $0.40 per pixel as the basis for comparing an LED solution to an incandescent solution.

If the PWM chips are $4 each, 70 would be $280, or $0.28 per pixel. I thought that LEDs that would be bright enough would be ~ $0.50 each, which brings it up to $0.78 per pixel, but maybe I didn't look at the right type or supplier.

don

[RavingLunatic]

When the LED string is plugged directly into the outlet, I can see slight flicker thru my peripheral vision, so I expect that I would also see flicker if I managed to dim them. I guess this is not a good test case.


I believe why you are seeing that flickering is that the LED string that you used is probably only half-wave rectified. I've seen other posts in the past with people who have experienced the same sensation of seeing flickering in their peripheral vision. But since it is happening so quickly most people never notice it.

I would guess that when you dim the half-wave recitifed LED string that the flickering may even appear to get worse since the LEDs are only being lit during half of the AC cycle and then you are cutting that half into pieces for the dimming.

So that is probably the answer that you are really seeking Joe. DC dimmers don't have the flicker because they don't have to worry about the AC cycle like AC dimmers do.

From the inputs that I got from our members about the problems that they experience this season, I would say that most of the people (here at DIYC) who experienced "flickering" issues (LED and mini) were from our friends "down under". And that is probably related to their power being at 50Hz, so missing half of an AC cycle would be more noticeable to the naked eye.

[Warlock]

I also read on some boards that some even noticed it on full wave leds more so on half wave...Either way it still happened...So what would the solution around this be ?? If lowering a bit of the power seemed to make it less noticable but dimming seemed to make it really noticable then doesn't it make sence to go to a dc board to drive a dc object and have it work without the flickering ??
Just an idea to toss out there..
What do you think RL ??

I also have some cheap leds for a bike light and they also flicker..Not recitified at all just hooked directly to a dc battery of low voltage..
Thanks Rl..

[djulien]

I believe why you are seeing that flickering is that the LED string that you used is probably only half-wave rectified. I've seen other posts in the past with people who have experienced the same sensation of seeing flickering in their peripheral vision. But since it is happening so quickly most people never notice it.

I would guess that when you dim the half-wave recitifed LED string that the flickering may even appear to get worse since the LEDs are only being lit during half of the AC cycle and then you are cutting that half into pieces for the dimming.

Yes, that makes sense.

I just connected a bridge in front of the LED string, and that cleared up the flicker (and the LEDs appear a little brighter), so that's an easy fix.

I know that the regular AC SSRs (MOC3023 + BTA08-400B) work okay with 120 V rectified AC, because I ran a bunch of them that way this past season.

So now I guess the question is how well the half-wave LEDs look on rectified AC when they are dimmed using the PWM firmware. I'm not set up for that currently; it will take me a while to try that.

don

[djulien]

I also have some cheap leds for a bike light and they also flicker..Not recitified at all just hooked directly to a dc battery of low voltage..

Hmmm, why would LEDs flicker when running from a DC battery, unless there is some kind of marginal voltage booster in there? Does it have a removal cover so you could look inside?

don

[RavingLunatic]

Joe,

I believe that the original Lynx that RJ designed was a DC dimmer at 120V and he eventually went away from that method due to rising costs of the components required to make it happen. It just wasn't as cost efficient as AC dimming was.

I believe that most folks don't want to perform surgery on their LED strands to convert them to be able to use a lower DC voltage. This would also require more channels or wiring them in parallel which would eventually negate the low current advantage of the full string using AC.


As they say "they are many ways to skin a cat" but sometimes you just have to wonder "why do I want to skin that cat?" :D

[Warlock]

Great work Don as always thank-you...

RL alright so dc boards are out as they are not as cost effective..Alright not a problem..So let me ask this why would it not happen when connected to the opto directly but a triac on the same board makes it happen ?? Any reasoning behind this ???
Thanks again RL I think we are all learning why this happens..
Joe..

[RavingLunatic]

RL alright so dc boards are out as they are not as cost effective..Alright not a problem..So let me ask this why would it not happen when connected to the opto directly but a triac on the same board makes it happen ?? Any reasoning behind this ???


Did I miss something in this thread????

When did it work differenly when connected to the opto versus the triac?

In a real world situation this could happen if the load was not enough to latch the triac but is enough to latch the opto. But that has less to do with what type of LED strand being used and more to do with the load being put on the opto/triac.

[Warlock]

Rl that is exactly what I'm trying to figure out..Don tried it and verses pluging it into the wall the flickering stopped and it was brighter when he bridged it on his boards...Makes a strong assumption that if the same was done with another board programmed with PWM it would work the same way but be able to dime effectivly..

[RavingLunatic]

Rl that is exactly what I'm trying to figure out..Don tried it and verses pluging it into the wall the flickering stopped and it was brighter when he bridged it on his boards...Makes a strong assumption that if the same was done with another board programmed with PWM it would work the same way but be able to dime effectivly..


I think you misunderstand what he did. He put a bridge rectifier in front of the LED strand to remove the flicker, basically applied DC to the LED strand instead of AC. The voltage is also probably greater than 115/120V and the current is probably also higher, thus brighter lights.

Has nothing to do with how PWM or non-PWM would affect the dimming when operating as AC.

[Warlock]

Rl I understand exactly what he did...What I'm saying is that if you do the same thing on a dimming board it gives a greater assumption that it would dim correctly without the flicker as well..He as I believe was also going from the opto as well and not from a triac.So it also proves my earlier assumption as well as what Firegod told me last year was 100% true..It can run off the triac but only in limited loads(or small loads) which would inturn meana higher channel count.But a small price to pay to do all this and get rid of that silly flickering problem...That has been driving me nuts for 2 months.

[djulien]

I think you misunderstand what he did. He put a bridge rectifier in front of the LED strand to remove the flicker, basically applied DC to the LED strand instead of AC. The voltage is also probably greater than 115/120V and the current is probably also higher, thus brighter lights.

RL, you are correct about how I connected the bridge. I think there is a subtle difference between constant DC and rectified AC. The bridge just swaps the polarity of half of the AC cycles, so that instead of getting a 60 Hz half-wave pulse, the LEDs are seeing a 120 Hz half-wave pulse. The voltage still goes up and down and the peak voltage is still the same, it's just always in the same direction instead of alternating + and -, so the LEDs turn on each time. The optos and triacs turn on and off as they should, because the supply voltage still drops to 0 120x/sec. I guess they would always be operating in 1 quadrant instead of 2, or something like that.

don

[awhaley]

The other thing to think about if you did go with LEDs, is that you COULD multiplex them, so instead of needing 70 PWM chips, you could get back to 12-16, which is a huge savings in control hardware, which could be applied to the LED purchase price.

I'm still not sure you'll beat .40c per pixel with LEDs in any shape or form that you're happy with though, but it's something to think about.

I do think that cutting up minilight strands for your pixels is rather ingenious btw, if we do hatch a cheap way to control them. :)

Art

[phillikl]

I hate to redirect the flow of traffic on this, but was doing some research for a possible approach of running multiple channels from 1 opto and ran across this.....

http://news.thomasnet.com/fullstory/823188


These seems to cover most of the points we have scratching our head about (power consumption, ampreage, speed, heat, cold, and moisture). The only thing I would be concerned with is the Low Voltage Protection from upstream of the SSR. From the article these will run about .94 for a 1amp chip.

[P. Short]

I hate to redirect the flow of traffic on this, but was doing some research for a possible approach of running multiple channels from 1 opto and ran across this.....

http://news.thomasnet.com/fullstory/823188


These seems to cover most of the points we have scratching our head about (power consumption, ampreage, speed, heat, cold, and moisture). The only thing I would be concerned with is the Low Voltage Protection from upstream of the SSR. From the article these will run about .94 for a 1amp chip.

Those parts seem to be intended for DC operation for driving transistors of one sort or another, not for use with AC loads. If that is what you want, they will be fine.

[djulien]

I hate to redirect the flow of traffic on this, but was doing some research for a possible approach of running multiple channels from 1 opto and ran across this.....

http://news.thomasnet.com/fullstory/823188

These seems to cover most of the points we have scratching our head about (power consumption, ampreage, speed, heat, cold, and moisture). The only thing I would be concerned with is the Low Voltage Protection from upstream of the SSR. From the article these will run about .94 for a 1amp chip.

phillikl, thanks for the link. That sounds like it nicely addresses the power issues.

Regarding the control signals, I think I'm still "stuck" on AC - I'm trying to use a lower refresh rate with charlie-plexing and non-PWM rather than a higher scan rate and PWM on the controller side, and then use the latching feature of optos or triacs with AC to dim or compensate for the POV effect. Maybe there's no point in doing that, but I got the impression that PWM devices cost more (and they are more versatile), plus I don't know how to use them well.

don

[djulien]

Here are some more incandescent test results (I'm trying to get a better picture of the actual operating characteristics for the opto, since I don't know much about the theory of operation or the real meaning of many of the technical parameters).

First test - I connected 2 50 ct. mini-lights in series at 120 VAC, to try to get the current within the opto's 100 mA spec limit (I didn't want to use a series resistor, because it sounded like it would need to handle heavy wattage):
- When intensity was set to 100% in Vixen, current was 115 mA (lights were dim, as expected); I didn't quite meet the opto's spec, but got close
- On/off in Vixen worked okay up to 3 seconds
- After 3 seconds, the opto locked up with the lights on, and I could no longer turn them off within Vixen, so I had to disconnect the power to turn the lights off. This sounds similar to the lock-up condition that Ernie described, but the opto did not feel hot or even significantly warm. I think that if the cold inrush current was too high, then I would not have been able to turn the lights on/off at all during the first 3 seconds. So I don't know why this happened.
- If I set the intensity to 25% in Vixen, the lights came on but would flicker, and could not be turned off again if I left them on more than 3 seconds. I'm not sure why there would be flicker in this case. Since the current is over the 100 mA limit, it seems like there would be enough holding current to keep the opto firmly turned on. The only other thing that comes to mind is that if the firmware is using 30 usec pulses (non-PWM), then my favorite chart on the MOC spec sheet indicates that the trigger current should be ~ 10 mA instead of 5 mA. I'm using the 680 Ω resistor with the opto, so the trigger current should be ~ (5-.7)/680 = 6.3 mA. Maybe I need to lower it to 430 or less? Is that related to this comment:


it might be desirable to increase the opto current for Renard when running in non-PWM mode.

This post (http://doityourselfchristmas.com/forums/showthread.php?p=9811) seems to be saying that I should be using a 300 Ω resistor with the opto instead of 680 Ω.

Second test - I connected 3 50 ct. mini-lights in series:
- With intensity set to 100% in Vixen, current was 97 mA, so this is within the opto's specs
- On/off in Vixen worked consistently (no 3 second limit)
- Dim at 25% and ramping worked consistently (no 3 second limit)

Third test - 2 50 ct. mini-lights again, but with ~ 340 Ω in front of the opto input (~13 mA):
- On/off in Vixen worked okay up to 30 seconds, lock-up after 30 sec. With more trigger current, it takes longer for lock-up; I'm still not sure why.
- If I set the intensity to 25% in Vixen, the lights came on dim but would flicker, and the opto locked up after 20 seconds.

4th test - 2 50 ct. mini-lights again, but with ~ 227 Ω in front of the opto input (~19 mA):
- On/off in Vixen worked okay up to 38 seconds, lock-up after 38 sec.
- If I set the intensity to 25% in Vixen, flicker for ~ 45 sec, then lock-up after 45 sec.

Okay, so I guess this means that the opto really is picky about that 100 mA limit for the output current, but only at higher voltages (my earlier tests at 8V, 16V and 24V worked). I suppose it could be related to dv/dt being too large when 120 VAC is used, except that it takes a while to happen. Maybe it is related to cumulative heat or power dissipation, except that the opto doesn't feel very warm on the outside. Maybe the junction itself is hot, but I just can't feel it on the outside.

And then for that question about opto input current, it seems to prolong the operability of the opto if the input current is higher, but then at some point it doesn't matter any more. According to the "Trigger current vs. Ambient temperature" chart, the trigger current goes down as the temp increases, not up, but the "Leakage current vs. Temperature" chart goes up, so maybe there is heat build-up inside that causes leakage, which syphons off enough trigger current so that even with more than enough trigger current available, the opto can't latch on any more? (If someone who understands how these devices work has comments, or pointers to somewhere I can read about it, I'd be interested in the real answer, even though now it's more a matter of curiosity than practicality, I guess).

I think I am concluding that the MOC3023 really shouldn't be used as a triac, at least not at 120 VAC
EDIT: ... under the conditions that I used. Under other conditions it does work (see later in this thread)

OTOH, for smaller voltages (which was my original target application), it did seem to work okay. I don't have a 2:1 transformer handy to be able to test at an intermediate voltage like 60 VAC.

I'm not sure if the flicker that Joe described is related to the flicker I saw in the tests above, or it was just the 60 Hz half-wave issue.

EDIT: I was thinking that the flickering was the result of some sort of "conflict" within the opto, where a voltage or current was struggling to be enough but was getting pulled down by something else within the opto. Now I think it might be simpler than that. A Wikipedia article (http://en.wikipedia.org/wiki/Thyristor) speaks of a minimum delay called "commutated turn off time" that is needed for the thyristor to "recover" before it can be turned on again, after it has been forced off. I'm not sure that is exactly what is happening, but the recovery time seems like a valid explanation - the flickering might simply be that the opto is ignoring several turn-on pulses from Vixen until it is ready to turn on again.

I'm try to gather up some parts to do LED tests next time, since there was some interest in those.

don

[awhaley]

Another cool little integrated SSR part to throw out...

PR36MF12NSZF

Mouser has a few of them at .70 cents (.63 if you take 50) It's a dip-8 package opto and power triac rated to .6A, which doesn't leave a LOT of room over a .45A string of minis, but might be worth testing. I'll throw a few on the next order I send in, but that might be a little while.... Just wanted to toss it on the table in case anyone is still looking for a low power SSR option....

Art

[djulien]

Another cool little integrated SSR part to throw out...

PR36MF12NSZF

Mouser has a few of them at .70 cents (.63 if you take 50) It's a dip-8 package opto and power triac rated to .6A, which doesn't leave a LOT of room over a .45A string of minis, but might be worth testing. I'll throw a few on the next order I send in, but that might be a little while.... Just wanted to toss it on the table in case anyone is still looking for a low power SSR option....

Art

Yes, I think I will also add some to my next order. It sounds like it will do the job.

Does the 25 mA holding current mean that it is more likely to work with non-PWM firmware for LEDs and relays? That would be nice in that it would simplify multi/charlie-plexing on the controller side.

don

[ErnieHorning]

I think I am concluding that the MOC3023 really shouldn't be used as a triac, at least not at 120 VAC. Okay, so the warning on the spec sheet was correct, but now I think I have a slighty better feel for why it can't be used.

I’m using a Fairchild MOC3023 at 120 VAC and they seem to work fine.

I paralleled to 6 watt bulbs giving a 97 mA load. I ran it for 10 minutes and it still could be turned off. I bumped the voltage up to 140 VAC moving the current to 106 mA. I ran it for 10 minutes and it still could be shut off. I found that it really only required 2mA to turn the DIAC on.

I brought it home and hooded it up to a Renard that I have bread boarded. I can dim just fine. No flicker or hesitation. I’ve left it on for a minute and the rapidly cycled it, looking for any hesitation.

I have it repeating a ramp up in one second and then fully on for 10 seconds. It’s been doing this for an hour so far.

Boy, you had me worried there. I’m in need for a lot of channels in a close space and was counting on this working. I wasn’t intending on charlieplexing, just individual channels, just like we’ve been doing. I have one application that requires 60mA and I’m pretty confident now, that it’s going to work. I have another application that requires 100mA per channel and though I hadn’t intended on it, because the testing that you had me curious about, it looks possible that this might work too.

[ErnieHorning]

I have it repeating a ramp up in one second and then fully on for 10 seconds. It’s been doing this for an hour so far.I let this run for over 4 hours last night. Temperature sensitivity is still unknown but this test was at approximately 70˚F.

I may build up a test fixture that simulates an SSR control pulse, synchronized to the line frequency. I think I can just use a 555 timer IC configured as a one-shot, triggered by a zero-cross pulse and control the pulse delay with a variable voltage. I should be able to drive at least five opto’s at the same time. I don’t think I have enough light bulbs to test any more anyway.

Depending on my ambition, I may write a quick program to control an HP data acquisition system. This way I can send out the pulses and check if the light turned on or off with an opto sensor. It’s also possible then to add a thermo couple and pitch this in a thermo chamber at work (if its not being used for legitimate work). These chambers can go well beyond anything you encounter anywhere on the planet.

[djulien]

I paralleled to 6 watt bulbs giving a 97 mA load. I ran it for 10 minutes and it still could be turned off. I bumped the voltage up to 140 VAC moving the current to 106 mA. I ran it for 10 minutes and it still could be shut off. I found that it really only required 2mA to turn the DIAC on.


I brought it home and hooded it up to a Renard that I have bread boarded. I can dim just fine. No flicker or hesitation. I’ve left it on for a minute and the rapidly cycled it, looking for any hesitation.

I have it repeating a ramp up in one second and then fully on for 10 seconds. It’s been doing this for an hour so far.

Boy, you had me worried there.

Sorry, I should have said "at significantly more than 100 mA", and/or my test methodology isn't as good as it should be. I'll edit that post to be more correct.

If it was working for you at 140V and 106 mA, then maybe I was just too far past the manufacturer's guidelines in my test, which makes sense that it would not work in my test case. But there's not that much different between 106 mA and 115 mA, so that seems like kind of a sharp curve there, or else it varies between batches or manufacturers.

In another thread you pointed out that the max reverse voltage for the opto's input varied from one manufacturer's spec to another (Fairchild and others were 3V, Lite-on was 6V). So, maybe Fairchild just makes a more capable device, or maybe the Lite-on device is less sensitive on the input side or something like that.


I found that it really only required 2mA to turn the DIAC on.

If that was from the Renard, is that PWM or non-PWM?

don

[djulien]

This way I can send out the pulses and check if the light turned on or off with an opto sensor. It’s also possible then to add a thermo couple and pitch this in a thermo chamber at work

I like that - the system can monitor whether the lights actually come on or not. I suppose that idea could be used when running a live display - use a Vixen trigger or something that checks the status of the lights and warns if something is wrong. That would also cover other types of problems such as GFCI monitoring, all with the same technique.


(if its not being used for legitimate work).

What is your definition of "legitimate"? ;)

don

[ErnieHorning]

But there's not that much different between 106 mA and 115 mA, so that seems like kind of a sharp curve there, or else it varies between batches or manufacturers.I believe I said that the maximum was 120 mA, which is 200% of what the manufacturer says is the absolute maximum. On a sample of one, I guess we can’t complain if no other chip can make it that far. As far as I know, I might have gotten the only one.


If that was from the Renard, is that PWM or non-PWM?It’s non-PWM running in a PIC10. I don’t know the exact width of the pulse, but there is no reason to believe its any different then the 3 uS in a PIC16F688.

[ErnieHorning]

I like that - the system can monitor whether the lights actually come on or not. I suppose that idea could be used when running a live display - use a Vixen trigger or something that checks the status of the lights and warns if something is wrong.I don’t think it would be feasible to monitor every light, but you might be able to check the current; it could be built into the SSR board. I don’t think it would be worth it to monitor on-the-fly, but you could run a diagnostic maybe once an hour. Cycle through each channel and verify that the current hasn’t increased or decreased from the last time it was scanned.

I’m not sure that it would be worth the extra hardware though. I’m not quite anal enough to have the need to know the exact second that a bulb burns out….. Yet.:wink:

[Warlock]

Well you wouldn't be able to monitor every light but you can using a buffer system monitor every channel of lights to see from inside if a channel goes bad..
Friends of mine do this and use this sort of system so I know it is possible..

[djulien]

Well you wouldn't be able to monitor every light but you can using a buffer system monitor every channel of lights to see from inside if a channel goes bad..
Friends of mine do this and use this sort of system so I know it is possible..

How about this - I saw a video on YouTube where someone taped some detectors to a computer monitor or TV screen to send signals to a monitoring device. Vixen could drive some servos in a pan-and-tilt stand under a video camera, so each bulb status would be displayed on the screen, then a trigger would fire if the light off the screen was below a configurable threshold? :)

Maybe that is too much camera movement and focusing to be reliable. I also found some open source motion detecting software a while back, so maybe it could be comparing an image of a correctly lit display to a video camera output, then sound an alarm if any of the bulbs were out? (I actually would like to use this software for security purposes at some point, but that project is way down on my current list).

don

[djulien]

I believe I said that the maximum was 120 mA, which is 200% of what the manufacturer says is the absolute maximum. On a sample of one, I guess we can’t complain if no other chip can make it that far. As far as I know, I might have gotten the only one.

I got confused - I was thinking of 100 mA as the peak constant output based on your earlier test. Too many numbers to keep track of.


It’s non-PWM running in a PIC10. I don’t know the exact width of the pulse, but there is no reason to believe its any different then the 3 uS in a PIC16F688.

Did you mean that you are running Renard from a PIC10xxx, or was that a separate test? If it was on the PIC10xxx, did you need to change much or any code to get it to run?

I also didn't quite understand the 3 uS - I thought the Renard firmware is using 30 usec?

don

[ErnieHorning]

Did you mean that you are running Renard from a PIC10xxx, or was that a separate test? If it was on the PIC10xxx, did you need to change much or any code to get it to run?it’s a side project that happens to be on a breadboard at the moment so I barrowed it. Phil wrote and doesn’t look anything like the released code since there is no UART, it had to emulated in software.


I also didn't quite understand the 3 uS - I thought the Renard firmware is using 30 usec?I missed a zero, though its actually 32.6 uS (120/255).

[djulien]

I did some tests with LED strings. Since I haven't decided how I will be using these LED strings yet, I didn't want to chop them up, so I just used the whole string as-is at 120V. All results below are for 120V.

I used a 25 ct. LED C9 string. When plugged directly into the outlet, it draws ~ 10 mA, and there was some flicker in my peripheral vision, and the waving tests showed flashing very clearly. It is obviously half-wave. When I connected a bridge in front of it, this pretty much eliminated any noticeable flickering (and less noticeable during the "wave test"), and the current consumption went up to ~ 20 mA. The info sheet says the LEDs are 20 mA, which they appear to be, even though they would only get an average of 10 mA the way they were manufactured (ie, half-wave).

For the first test, I connected the LED string in series with the MOC3023 output. The LEDs would not turn on from Vixen, as expected (the PICs are non-PWM). I used non-PWM firmware for these tests because I wanted to observe the opto's latching characteristics (PWM doesn't need the opto's latching feature since the opto is basically just being used as an isolator/driver). I'd like to use the latching feature in order to free up control signal bandwidth on the controller side, as well as reduce the average output current needed from the controller chips.

For the second test, I added the bridge in front of the opto + LED string, thinking that maybe the current drain would be enough to allow the opto to latch. It was not.

For the 3rd test, I removed the bridge and connected a 10K Ω 2W resistor in parallel across the LED string to add an additional load of 12 mA. The opto was now able to turn the LED string on/off, dim and ramp/fade smoothly from within Vixen. At 50% intensity, the current measured at 13 mA, and the LEDs still lit, so there's some inconsistency here.

For the 4th test, I added the bridge again. Total load was now ~ 35 mA (should have been 32 mA, but close enough). As expected, Vixen was able to turn the LED string on/off, dim/ramp, etc.

For the 5th test, I used a 47K 1/4 W resistor instead of the original 10K. This would add ~ 2 mA to the load instead of ~ 12 mA, for a total of ~ 27 mA (should have been 22 mA). The LED string turned on/off and dimmed correctly within Vixen.

So it looks like the latching current for the opto is ~ 22 mA, which is just a little too high for a bare full-wave LED string to trigger it. Adding a 10K 2 W resistor in parallel with the half-wave LED string compensates for this, although adding a little more power consumption.

Alternatively, running the LED string on rectified AC, the LEDs will be brighter with less flicker, and only a 47K 1/4 W resistor is needed in parallel.

Probably an alternate part, with a slightly lower latching current, would be more convenient for use this way (to try to avoid the need for an extra resistor, in the non-PWM case).

I used "flame-resistant" resistors for this test, in case my calculations were off. :) They seemed to be correct, though. I had figured that 120 V/10K = 12 mA, and 12 mA * 120 V = 1.4 W, so a 2W resistor was safe, and 120 V/47K ~= 2.1 mA, and 2.1 mA * 120 V ~= .25 W, so a 1/4 W resistor was safe (just barely).

I'm guessing that the results would be similar with lower voltage and fractional LED strings, although there would likely need to be some adjustment of the resistor values.

Joe, I'm not sure if this helps to answer your questions about using LEDs and flicker with the opto directly.

don

[Warlock]

Thanks Don simply brilliant work...I owe you one..
It answered everything I hoped it would.
Thanks again.
Joe

[djulien]

I’m in need for a lot of channels in a close space and was counting on this working. I wasn’t intending on charlieplexing, just individual channels, just like we’ve been doing. I have one application that requires 60mA and I’m pretty confident now, that it’s going to work. I have another application that requires 100mA per channel and though I hadn’t intended on it, because the testing that you had me curious about, it looks possible that this might work too.

Ernie, can you share any info about that project? Was it something like a C9-triks or C7-triks?

don

[budude]

So - I picked up some 50ct half-wave LEDs strings cheap today and am planning to build some fountain sticks with them. This sounds like I should be able to forgo the Triacs and drive these strings directly. If so that would be great as it would allow a much more compact setup. Thanks for all the leg work!

[ErnieHorning]

Ernie, can you share any info about that project?Specificaly, it's a cheap and fully addressable C7 string using the Renard protocol. I was hoping to have it running for this years display. I even showed of the 1" Sq controller board; all parts are surface mount. I just need to hack it into a couple C7 strings. Everythings ready, just need some time and space to do it.:mad:


So - I picked up some 50ct half-wave LEDs strings cheap today and am planning to build some fountain sticks with them. This sounds like I should be able to forgo the Triacs and drive these strings directly.This should work great for LED's.

You could also use the non-PWM renard and a very small power supply because the leakage current is a maximum of 100µA.

[Entropy]

It looks like 100 mA is about the maximum you'd safely want to drive continuously.

From the original IC's datasheet:
The voltage drop across the output can be as high as 3 volts at 100 mA. The maximum power dissipation is 300 mW. Any higher than 100 mA and you're at a high risk of exceeding 300 mW - Especially since you'll probably run into thermal runaway.

[djulien]

Specificaly, it's a cheap and fully addressable C7 string using the Renard protocol. I was hoping to have it running for this years display. I even showed of the 1" Sq controller board; all parts are surface mount. I just need to hack it into a couple C7 strings. Everythings ready, just need some time and space to do it.:mad:

Please post more details about it when it's ready!

don

[djulien]

So - I picked up some 50ct half-wave LEDs strings cheap today and am planning to build some fountain sticks with them. This sounds like I should be able to forgo the Triacs and drive these strings directly. If so that would be great as it would allow a much more compact setup.

Another nice thing about a half-wave LED application is that if you are planning to use the half-wave strings as-is (ie, not put additional diodes or a bridge in front of them to run them as full-wave), then you could go even more compact by using "SSR doubling". That is, you could connect 2 of the strings back-to-back (in parallel, oriented in opposite directions) to each opto instead of 1, and then use SSR-doubling firmware to send alternating channels to each SSR in the positive and negative AC half-cycles.

don

[P. Short]

Another nice thing about a half-wave LED application is that if you are planning to use the half-wave strings as-is (ie, not put additional diodes or a bridge in front of them to run them as full-wave), then you could go even more compact by using "SSR doubling". That is, you could connect 2 of the strings back-to-back (in parallel, oriented in opposite directions) to each opto instead of 1, and then use SSR-doubling firmware to send alternating channels to each SSR in the positive and negative AC half-cycles.

don

Unless the half-wave string is already two half-strings in parallel, oriented in opposite directions. There are some out there like that, I think.

[budude]

These strings are pretty cheap - I haven't pulled apart the plug ends yet but they look like regular mini ends (not the "super size" style) and there's no blobs so I need to figure out where the diode is.

[djulien]

Unless the half-wave string is already two half-strings in parallel, oriented in opposite directions. There are some out there like that, I think.

Thanks for catching that, Phil. I forgot to mention that.

Actually, in cases where someone wants to use the shorter strings and SSR doubling, those "twisted" strings make things even easier, since they are like 2 shorter strings that have already been connected in parallel in opposite directions.

don

[djulien]

Specificaly, it's a cheap and fully addressable C7 string using the Renard protocol.

What was your approximate final per-pixel cost on that (excluding the controller), if I may ask? Was it under $1.00?

don

[ErnieHorning]

What was your approximate final per-pixel cost on that (excluding the controller), if I may ask? Was it under $1.00?
Each pixel IS a controller. There's a PIC10F on each one. These didn't cost me much because I needed to buy the micro's and the rest. Using Mouser as a reference, to build 50 would cost $1.03 per pixel in parts. Higher quantities and using other suppliers would drop the cost more. This doesn't include the board cost but these should be pretty cheap. There are no holes and I'm using .31 thickness. It also doesn't include the cost of the lights but I only paid $1.50 for C7 strings when they went 90% last year.

Mouser has a limited supply of non-ROHS MOC3023's that would drop the pixel cost to 83¢ but this until they run out.

There's also one board at the beginning of the string that has a RS485 receiver and a small power supply to power it.

Power would come from a RenT.

I still consider this as being at the beta stage and feel that there is still room for cost reduction.

[djulien]

Each pixel IS a controller.

I should have worded my question differently; I meant excluding the Renard or whatever interface that sits between Vixen and the SSRs. Anyway, you answered the question that I meant to ask.


1" Sq controller board; all parts are surface mount. I just need to hack it into a couple C7 strings

Since the C7s are typically wired in parallel, would you run another supply wire along the string into the optos, and then a cat5 or something along the string to supply the data to the PICs?

(Sorry for all the questions. I'm also looking into turning a string into individually addressable "pixels", except I'm looking at driving the optos via chipiplexing from a shared Renard rather than using a dedicated uController for each pixel).

don

[ErnieHorning]

Since the C7s are typically wired in parallel, would you run another supply wire along the string into the optos, and then a cat5 or something along the string to supply the data to the PICs?
It's nothing more than a 1 channel Renard. I wanted something that was simple, cheap and easy to put together. I thought this was the best I could get and still be safe.
http://img341.imageshack.us/img341/3843/ren1.th.jpg (http://img341.imageshack.us/i/ren1.jpg/)

... I'm looking at driving the optos via chipiplexing from a shared Renard rather than using a dedicated uController for each pixel).I thought about doing this a couple of years ago but I figured it would be a wire nightmare. How thick will the wires be by the time you get to the 50th light?

[djulien]

How thick will the wires be by the time you get to the 50th light?

One cat5 would be enough for up to 56 monochrome pixels. I ran a cat5 along the icicles on my roof line and it worked pretty well, so I'm thinking it will also work well for a string of C7s. (I've read that data should not be run parallel to power, but I didn't have any issues with it for the relatively short runs that I used).

EDIT: Page 48-50 of the article in this link is what I'm currently hoping to do:
http://doityourselfchristmas.com/forums/showthread.php?t=10889

don

[ErnieHorning]

One cat5 would be enough for up to 56 monochrome pixels. I ran a cat5 along the icicles on my roof line and it worked pretty well, so I'm thinking it will also work well for a string of C7s. (I've read that data should not be run parallel to power, but I didn't have any issues with it for the relatively short runs that I used).You're not thinking of running 120 through Cat5... Are you?

The connectors add too much bulk so I'm not intending on using connectors, just soldering directly to the boards. I haven't found a case that I like so I'm thinking of just potting with epoxy.

[Brad Riley]

Doesn't the Triac also act as a second, or maybe the first devise isolating HV from the computer??

Seems like cheap insurance.


Brad

[djulien]

You're not thinking of running 120 through Cat5... Are you?

The connectors add too much bulk so I'm not intending on using connectors, just soldering directly to the boards. I haven't found a case that I like so I'm thinking of just potting with epoxy.

No, I was not going to run 120V thru the cat5, just the control signals. I have read in other posts that 120 V cannot safely be run thru cat5 due to inadequate insulation and limited current capacity.

I was intending to run a cat5 carrying the control signals from the Renard thru the cat5, and then pairs of wires would be connected to the input side of the optos, more or less like regular SSRs.

Yeah, the connectors seem to be the tricky part in this scenario. I was hoping that you had figured out a slick way to do that. I'm experimenting with a few ideas there.


Doesn't the Triac also act as a second, or maybe the first devise isolating HV from the computer??

I am not too familiar with the detailed triac characteristics, but I don't think there is any isolation in there, given that the gate connects thru a series resistor to one of the high voltage terminals.

don