Grinch design flaw? - Nope design is good

[scorpia]

UPDTATE: Looks like the Design is ok. i have added this in the first post to make sure people dont get the wrong idea with the following questions i have asked.

I am trying to learn more about the chip that is used in the grinch and understand how it works , since the design didnt follow what seems to be the manufaturers recommended design i asked the question,

I have left the original message below for history's sake.
---------------------------------------------------------------------------------

Umm,

i was looking at using a different chip in the grinch and i was wondering what is the logic with tieing the Rset pin (23) to ground?

Isnt this supposed to have a current limiting resistor installed? I cant see any reference in the data sheet of what happens with a 0Ohm resistor.

at very least i would think that the chips are drawing the maximum current that the chip can produce. (90ma?) so would add quite a large load to the power supply.

I guess im wondering why it is tied to ground and not via a resistor like the data sheet suggests?

Peter

[Wayne J]

Here is a quote from RJ's How To, he tells about it.


I was asked to explain the operation some so I added this.

The reason we can get away with one chip is because of two factors. the 595 chips output power when they go high (vixen says turn the light on) This is the opposite of what the ssr design we use needs to turn on it needs to be sinked (hooked to Ground) so we need a way to invert this. You could redesign the ssrs to work either way but I needed this design to work with ssr's we have. The 2803 chips does this for us. the other reason is if the 595 did put the correct signal out it could not handle the current of running all the ssrs at one time. so the 2803 is a power chip and provides us with enough current to do the job.

The 5027 on the other hand does sink the outputs when the port is set high. And it has transistors built into it to provide far more than we need to drive the ssr's. This combines both chips into one and makes it the ultimate 595 replacement. In fact its 16 bit so it replaces 2 595's. Your ssr's can be just as far away. There is nothing giving up in the design other than led's, cost, size. The leds could be added but the cost would increase. In fact a clever person could add the Led's with out the need for the resistor array as the 5027 can provide constant current by setting one resistor between all the R-EXT pins and ground. I have it directly grounded to have the circuit provide how ever much current the ssr needs (up to the chips max of 90ma) so other designed ssr's with less sensitive triacs will still work. But if you used leds that match up well you could set the current add the led with no load resistor and let the chip do the work. This was outside the scope of the project as I intended it to be as cheap as possible. It was my hope that this controller could be built for under $.25 a channel or 64 channels for $15. I believe it is possible with a coop order. With the coop that just ran getting the ssr's to around $1.80 a channel we would be close to offering a DIY'er a $2.00 a channel setup without enclosures. The Current 595 design runs close to $3.00 a channel.

[RJ]

Umm,

i was looking at using a different chip in the grinch and i was wondering what is the logic with tieing the Rset pin (23) to ground?

Isnt this supposed to have a current limiting resistor installed? I cant see any reference in the data sheet of what happens with a 0Ohm resistor.

at very least i would think that the chips are drawing the maximum current that the chip can produce. (90ma?) so would add quite a large load to the power supply.

I guess im wondering why it is tied to ground and not via a resistor like the data sheet suggests?

Peter

Because the Current is limited at the SSR via the resistors. Since the voltage is 5v and there is a set resistence in the SSR limiting the current to what we need there is no need to have the Chip limit current under it's maximum. The power supply is only going to provide what is requested of the SSR. This allows flexibilty also as some SSR design need more current than others to operate. This allows their designers to set the current up to the limit via resistor.

This we limit the part count and cost(I know it's cent's) and the Grinch was designed to be the simpliest, cheapest, fastest to build, way to blink 64 channels of lights.

RJ

[Wayne J]

LOL... posted at the same time. ;)

[RJ]

LOL... posted at the same time. ;)

Hey at least we gave hime the same info. It's only a problem when the response disagree! LOL

RJ

[wjohn]

yep, 90 mA is fine, the SSRs will only draw what the are designed to sink (5-15mA), and if you want to terminate LED directly to the GRINCH (aka strobes) the IC will current limit at 90mA. Most of the high power 150-250,00 mcd LEDs need ~100mA anyway.

THe LEDTRIKs, is different, and it has current limiting resisitors set for the MBI/Allegros at 20mA, as the LEDs in the displays will be driven on more consistantly, and there is no inherent current limiting in the 768 LEDs that make up the display. the Value I selected was 1 k, to give 20 mA max (I fwd rating of the LED)

[scorpia]

thats basicly what i thought, but i still dont see how the 0 ohm resistor limits the chip to its max current draw and then allows the SSR's to only pull what they need (say 20ma.)

i thought the chip was a constant current device.

Looks like i need to have another read over the data sheets to get my head around the idea.

i might try modifying one of the chips on the grinch board to limit to about 50ma and see if the maxim chip works. the maxim might not be compatable.

Peter

[wjohn]

Peter,

Constant Current is the Maximum the IC will deliver, from a given supply (5 VDC in this case). If you place a short between the terminals of a Constant current source, it will limit the current flow to 90 mA ( in this case). the equivilent value is around 56 ohms for 90 mA at 5 VDC.

If you place a 100 ohm load across a 90mA constant current source (from a 5 Volt supply), you will get a current of 50mA. Why? the potential of the supply is limited to 5 VDC and against a 100 ohm load, you cant force 90mA thru the load.

Change the supply voltage to 10VDC, and the math changes.

Do the same experiment with a 1Kohm load, and you will get 5 mA (about what the SSRs conduct from the GRINCH).

Therefore, Constant Current is the maximum current the supply can provide, not the amount it will supply, regardless of the resistive load connected.

The reverse here is a contsant Voltage regulated supply, they will provide their nominated voltage (up to the rated current). If the resistive load is less than the V/I of the supply, then either to supply will drop its voltage to match it maximum supply current, or kick in some protection (nominally current limiting) to stop the Power supply melting down.

regardless of Constant Current or Constant voltage, the other value (V or I) will be variable to the ratings of the supply.

hope that helps.

John.

[scorpia]

John,

i understand what you saying except that if you change the load and the current changes then its not a constant current chip.

If the chip is constant current then it should adjust the voltage to suit the load.

If the chips isnt a constant current chip and rset sets the maximum current then what you say is correct. i just thought i read that the chip was a constant current source.

As i said, back to the the datasheet :)

Peter

[P. Short]

The constant current is always limited by the available voltage. If the constant current source has an open output (i.e. infinite resistance), do you think that it is going to reach the specified current, or is it going to be zero?

Saying that something is 'constant current' is never really totally accurate, just as saying that a power supply is 'constant voltage' is never totally accurate (what if the output is shorted)?

--

PHil

[wjohn]

John,

i understand what you saying except that if you change the load and the current changes then its not a constant current chip.

If the chip is constant current then it should adjust the voltage to suit the load.

If the chips isnt a constant current chip and rset sets the maximum current then what you say is correct. i just thought i read that the chip was a constant current source.

As i said, back to the the datasheet :)

Peter

It is only Constant current, within the specs of the chip. In this case 5 V and 90 mA, and the Load Resistance can be between 0 and 56 ohms, and the IC can provide 90 mA. above 56 ohms, the current will drop IAW Ohms Law.

You cant fight ohms Law, it wins every time.

[scorpia]

Guys,

yes i understand what your saying. i think you have missed my point.

assuming you stay within the limits of the chips design.
how can a device supply a constant 90ma, and then have the output current vary depending on the load. that not a constant current thats a variable current. If it was constant current it would have to lower the voltage to compesate depending on the load. As you say ohms law is a bitch.

so either i missread the chip data sheet and its really not a constant current source, it is just current limited to a maximum. Or its a constant current source and the output voltage is varied to adjust to the load. or i missunderstand what a constant current source is. (the latter may be the case as im more used to set voltage devices.)

EDIT: i just reread john response and saying its only constant current when using a 0-56ohm load doesnt make the chip constant current in my book. it just means it limits the current to a maximum of 90ma.

so im gonna have a read and come back.

peter

[P. Short]

In normal operation (pin 23 or whatever not shorted out) the output will be a constant current as long as the resistance isn't too high. Suppose the voltage on the chip is 5V and it's programmed for 25 mA. Then, if the load is a resistance, the current through it will be that 25 mA so long as that resistance is less than 200 Ohm (approx). If the resistance is greater than that, the current will be determined by the value of that resistor (approx. 5V/R).

--

Phil

[Vadoo]

A clever person could add the Led's with out the need for the resistor array as the 5027 can provide constant current by setting one resistor between all the R-EXT pins and ground Apparantly I'm not as clever as I thought. I used the resistor array rather then just useing one resistor at the R-EXT pin. I will consider an edit in the thread I posted the design in when I get time. Which brings up another question...


The Current is limited at the SSR via the resistors How did the resistors end up at the SSR's rather then at the 5027 chip? That could save 7 resistors per 5027, or 28 resistors per Grinch.

[wjohn]

the SSR is a standard design, the current limiting resistor on the OPTO of the SSR is there to limit current flows into the OPTO.

The GRINCH is the ONLY controller that is Constant Current, the rest are not.

[wjohn]

Guys,



EDIT: i just reread john response and saying its only constant current when using a 0-56ohm load doesnt make the chip constant current in my book. it just means it limits the current to a maximum of 90ma.


peter

Peter, for the device to be constant current, by your definition, it would need to have infinite voltage; for a load of infinite resistance.

Ohm's Law is king. if you want to have a constant current (100mA) into a resistive load, and you are prepared to have that resistance of the load be any value from 1 ohm to 1 Meg Ohm, then the supply voltage across the resistave load must be able to ramp from 0.1V (1ohm) to 100,000 volts (1M ohm). This is where practical design kicks in. in the case of the 5027, the maximum practical voltage is 5 V, therefore the resistance can be between 0 and 56 ohms. The chip cant produce a voltage any greater than 5 Volt, so it is a constant current source between it limits of the data sheet.

You cant travel faster than the speed of light, and constant current souces can only work within their specs.

[Vadoo]

the SSR is a standard design, the current limiting resistor on the OPTO of the SSR is there to limit current flows into the OPTO.

The GRINCH is the ONLY controller that is Constant Current, the rest are not.

I was a little off on my math. It would save 60 resistors to have 4 resistors on the Grinch at each of the R-EXT inputs rather then resistors on the input side of the optos at the SSR's. I know its a standard design, and its very nice to have things standardized, but the idea of the Grinch was to make it as cheap as possible per channel. At $.05 a piece thats only $3 but its still a savings in money and a good bit of savings in build time. Would it be worth a redesign of the Grinch and a different set of SSR's specific to only one controll board?

[kmc123]

My opinion, for what it's worth is no - There is so much confusion today on RenT, RenC, etc... This would just extend confusion into the Grinch world for newbies imho...

[Vadoo]

My opinion, for what it's worth is no - There is so much confusion today on RenT, RenC, etc... This would just extend confusion into the Grinch world for newbies imho...

One of those it may be better but its too late to change it kind of things. I agree, unfortunately. Oh well, if it aint broke dont fix it:)

[wjohn]

Vadoo,

you can cut the track on the GRINCH board and fit 4 resistors the the controller, if needed ( a nice SMT resistor would look not look out of place. You can then fit 64 wire link jumpers to the 16 SSR boards.

Possible, but not that practical.

[RavingLunatic]

Vadoo - This is a case of what came first the chicken or the egg. In this case the SSR came first and used by Olsen and Renard systems before the Grinch was developed. So instead of creating another SSR design (with no real extra benefit) the Grinch was setup to work with existing SSRs.


You can always make the mod wjohn just suggested but for a savings of about .06 cents or less (depending on where you buy your resistors) for each SSR, is it really worth it?

[scorpia]

Guys,

i think that i have dragged this thread right off from what i intended it to be.

my original concern was that it might be possible that the MBi chips could possibly be running outside of there rated spec's since there is no limiting resistor as stated as needed in the data sheet. and possibly damaging the chips.

I understand that the standard SSR's that we use should never take the chips anywhere near there rated design as they are current limited. but with people hooking high powered LED's and other items directly to the grinch people might want to keep that in mind.

still i have learnt some more about electronics but im thinking about modifying my grinches to include a resistor for safety's sake. I figure it cant hurt.

Peter

[Vadoo]

Seems to make sense to try avoid running things at their limit. However that chip is imited by the ssr's. However if there was a short or something (outside of adding led's in which case it would take a redesign that should include additional resistors) it could damage the chip sinse it would have no resistance to keep the chip from reaching max specs, curent, voltage and such. Am I helping or confusing things more?

[RJ]

Seems to make sense to try avoid running things at their limit. However that chip is imited by the ssr's. However if there was a short or something (outside of adding led's in which case it would take a redesign that should include additional resistors) it could damage the chip sinse it would have no resistance to keep the chip from reaching max specs, curent, voltage and such. Am I helping or confusing things more?

OK,

You are confusing things more.

There is no Flaw in the design. The Chip's sense voltage is set internal and is limited to it's max normal operation range via this.

The resistor is to allow you to lower it for running LED's to something other than this. You can not overload the chip the way the grinch is setup. At least the way I set it up. If it has been changed then this is a different story.

By grounding the vref of the chip you will get it's maximum allowed continuous output. (notice I said continuous output) This means the chip can do this continuous with no ill effect. You can Short the output of a grinch as designed to +5 supply and go eat dinner and come back and unshort it and it will go back to working as if nothing has happened. I have done this intentionally so I know this to be true.

If you need a lower output, put a resistor on it.
If you want to play, put a resistor on it.

But since at least three people who do know what they are taking about have told you it is not needed for this application please do not continue to claim it is a flaw in the design.

This will start everyone who does not know better to doing unneed mods. I think the poor grinch design has been modded enough.

I am pretty sure John has sold hundreds of these boards and I do not think we have had any problem with that as of yet. Now port voltages, that's a whole other story.

RJ

[Vadoo]

OK,

You are confusing things more.

There is no Flaw in the design. The Chip's sense voltage is set internal and is limited to it's max normal operation range via this.

The resistor is to allow you to lower it for running LED's to something other than this. You can not overload the chip the way the grinch is setup. At least the way I set it up. If it has been changed then this is a different story.

By grounding the vref of the chip you will get it's maximum allowed continuous output. (notice I said continuous output) This means the chip can do this continuous with no ill effect. You can Short the output of a grinch as designed to +5 supply and go eat dinner and come back and unshort it and it will go back to working as if nothing has happened. I have done this intentionally so I know this to be true.

If you need a lower output, put a resistor on it.
If you want to play, put a resistor on it.

But since at least three people who do know what they are taking about have told you it is not needed for this application please do not continue to claim it is a flaw in the design.

This will start everyone who does not know better to doing unneed mods. I think the poor grinch design has been modded enough.

I am pretty sure John has sold hundreds of these boards and I do not think we have had any problem with that as of yet. Now port voltages, that's a whole other story.

RJ

I think i just got yelled at:( See what i get for commenting in the middle of heated arguments.

[wjohn]

yes, lot of GRINCHs have been built, and the only thing that still amazes me is that Allegro are still giving out samples of the IC to us.

I wold love to send them a Christmas card, but then they would stop the supply.

[RJ]

No,

At least I didn't think I was yelling at anyone. I was simply pointing out the lack of need of it. As has a number of people have already. And that if he wants to add them by all means go for it. On the other hand don't have others start modding thinking there are issues there are not.

RJ

[Macrosill]

This is what happens when some people come here and think they know more than the people desiging the stuff we use. Another example of loudmouths running around aqnd talking about something they do not know about!

There is no design flaw. RJ is a very compitent electronics designer. AN example is a while ago I asked him to help me design a tester while we were in the chat. In about 5 minutes there was a design on my desktop. It was not just some leds and a button. This thing was sequentially running through the leds and channels to test each one, one after the other. I was without words. There is nothing wrong with asking questions but lets not "attack" people by incinuating a flaw in their work, go about it the right way.

[ben]

This is what happens when some people come here and think they know more than the people desiging the stuff we use. Another example of loudmouths running around aqnd talking about something they do not know about!

There is no design flaw. RJ is a very compitent electronics designer. AN example is a while ago I asked him to help me design a tester while we were in the chat. In about 5 minutes there was a design on my desktop. It was not just some leds and a button. This thing was sequentially running through the leds and channels to test each one, one after the other. I was without words. There is nothing wrong with asking questions but lets not "attack" people by incinuating a flaw in their work, go about it the right way.

RJ is a very good engineer. I just want to note one thing, if there was a flaw, wouldn't the first couple of hundred boards had issues by now? Not to be rude but I am sure someone would have had a problem or picked up on it.

Cheers, Ben

[Vadoo]

He did title the thread "Grinch design flaw" with a question mark at the end. Still maybe not the greatest way to go about it. But he is trying to help, and he did help me. Some of you may have seen the board I designed under Grinch meets Olsen 595. Indicator leds, home etch file. It works very well. It seems to have gotten a decent response from people. However, I did screw up, and i didnt realize that untill scorpia started this thread and asked some questions. I guess I didnt really screw up, but there is a much better way I could have gone about designing the board. Unfortunatly I'm busy with a fairly involved project right now, and the design works just fine, so I wont be fixing it for a while. But in the future I hope to improve on it using the information scorpia pointed out.

Please read his edit on his original post before adding anything else to the conversation. It seems to be a good stopping point for this thread:)

p.s. Dont yell at me, I think the original Grinch is an awesome board!!!!! LOL

[scorpia]

Yes,

im sorry that me asking the question seem's to have gotten people upset. I never meant to get people offside. As i stated i just asked for some clarification because the design didnt follow what seemed to be the recommended config.

RJ as great and has pointed me to some more technical information offsite and that has really helped clarify how he came to the design.

I have learnt alot about both this design even alittle about forums (even though i have been on them for years). So please nobody take offense at any thing i said.

thanks again

Peter

[Wayne J]

Good, all things seem to be settled, so now is a great time to call this horse dead. ;)

[P. Short]

Good, all things seem to be settled, so now is a great time to call this horse dead. ;)

Not yet.

After doing some measurements in connection with a thread in another topic, it appears to me that this practice increases the current through the chip. With pin 23 open (not that I advocate this) the current through the chip (all outputs off) is about 7 mA and with it shorted to ground the current appears to be above 50 mA. If you read the specs for the chip, the current should be around 10-15 mA when using an appropriately sized resistor. And based on the measurements that I've taken, it appears that this current appears to be added onto the output currents when the load is non-zero. This places an extra load on the power supplies (current at full load is around 140-150 mA vs 100-110 mA. It may not seem like a lot, but it all adds up.

I don't think that it is worth changing the Grinch board at this time, but (in retrospect) it would have been better to have included the resistors when it was first designed.

--

Phil

[RJ]

Yes,

im sorry that me asking the question seem's to have gotten people upset. I never meant to get people offside. As i stated i just asked for some clarification because the design didnt follow what seemed to be the recommended config.

RJ as great and has pointed me to some more technical information offsite and that has really helped clarify how he came to the design.

I have learnt alot about both this design even alittle about forums (even though i have been on them for years). So please nobody take offense at any thing i said.

thanks again

Peter


Peter the extra current that Phil is posting about is the 4 ma extra per channel I spoke with you about via PM. So it is not in addition to what we talked about. It does not change anything with what you are doing. Add the resistor the same on the new chip as it is required.

RJ


Must be slow.