Using a ULN2803A instead of DCSSRs

[dlovely]

I have been looking at the Ren24LV but am worried about the draw of the Super Strips on it. Its said that it can take 500ma per channel. The spec sheet shows 400ma as a recommended but only on 1 channel. The spec says the chip can use parallel channels to distribute the load. I am thinking of now using the Ren64 and running each channel to two or four inputs of the ULN2803A as show in the pictures. That Dual ULN2803A would handle every channel RGB+W on at the same time.

Any thoughts? DCSSRs would also work, I was just looking at the cost difference between a ULN2803A and DCSSR.

[LabRat]

I have been looking at the Ren24LV but am worried about the draw of the Super Strips on it. Its said that it can take 500ma per channel. The spec sheet shows 400ma as a recommended but only on 1 channel. The spec says the chip can use parallel channels to distribute the load. I am thinking of now using the Ren64 and running each channel to two or four inputs of the ULN2803A as show in the pictures. That Dual ULN2803A would handle every channel RGB+W on at the same time.

Any thoughts? DCSSRs would also work, I was just looking at the cost difference between a ULN2803A and DCSSR.

Wouldn't it be easier to just double up the ULN's (piggy back) on the 24LV? This assumes that the traces on the board can handle the increased current. We would have to check with Frank as to what width those traces were when he did the layout.

[n1ist]

In addition to the 500mA max per driver (that's a max; they seem to suggest that 350mA is a good upper value), there's also a 1W per driver and 2.25W total for the chip. Piggybacking will parallel the drivers, but it makes it harder to get the heat out...
/mike

[dlovely]

I am looking at doing 12 Super Strips, so would need two 24LV's so it would be cheaper to go with a Ren64 at this point.

I am looking at building a second project that will require a lot of LEDs and would need to get a bunch of the ULN's for that anyhow. Seems just like a lot of overkill to use a DCSSR.

[budude]

I am looking at doing 12 Super Strips, so would need two 24LV's so it would be cheaper to go with a Ren64 at this point.

I am looking at building a second project that will require a lot of LEDs and would need to get a bunch of the ULN's for that anyhow. Seems just like a lot of overkill to use a DCSSR.

I'm in the same boat - I want lots of smaller LED based layouts and DCSSRs are a bit much for that. I'd be worried about quadrupling up the 2803s though - if they aren't matched up just right the one driver may try sinking most of the current and the magic smoke gets out. You might look at the ULN2074 also - it's good for 1.5A/output (with package limitations as well...).

[aussiephil]

One thing i am curious about is the cost per channel of these solutions including the actual controller.
Cheers
Aussiephil

[smartalec]

I'm in the same boat - I want lots of smaller LED based layouts and DCSSRs are a bit much for that.
You might look at the ULN2074 also - it's good for 1.5A/output (with package limitations as well...).

hmmm ULN2074 looks nice an powerfull for the leds, the only thing i saw from looking at the data sheet, is how a person hooks it up, what pins to what?
my electronics knowledge is very rusty

[dlovely]

hmmm ULN2074 looks nice an powerfull for the leds, the only thing i saw from looking at the data sheet, is how a person hooks it up, what pins to what?
my electronics knowledge is very rusty

See figure 18 in the spec.

http://www.st.com/stonline/books/pdf/docs/1534.pdf

Pins 1, 8, 9, 16 would go to the controller.

Nice find budude. Looks like it even supports a heat sink, tho not sure it would be needed. It would still have a cost advantage over DCSSRs, $2.31@10 qty.

[ctmal]

I was thinking about this solution before but decided against it strictly because(I think) there would be a somewhat substantial voltage drop across the IC(I was planning on 5v input). It's just something you would have to account for.

I do have a couple of the 2074's hooked at home(I'm using them in an LED sign). I can take some voltage measurements this weekend if anybody needs them.

[fathead45]

so you could run a cat5 cable out of the output of a ren64 to the uln and control the 4 colors?? would you be injecting power thru the uln?? im really starting to dig the idea.

[budude]

so you could run a cat5 cable out of the output of a ren64 to the uln and control the 4 colors?? would you be injecting power thru the uln?? im really starting to dig the idea.

Yes - you'd have to make some type of board to hold the chip and feed the external power in but that is not a big deal. You have to remember one thing though and that is the ULN2803 inverts the output so you have to modify the Ren code to use DC mode (which makes the PIC flip it's output). If you want to use AC SSRs, they'll have to connect to the output of the ULN2803 and not the Ren64 itself (this is how the Ren24LV is also...).

[dlovely]

so you could run a cat5 cable out of the output of a ren64 to the uln and control the 4 colors?? would you be injecting power thru the uln?? im really starting to dig the idea.

Thats what I want to do. This thread was also to gain an interest level. A simple PCB with two RJ45 jacks, a two pin set-screwed terminal for power and the IC.

[fathead45]

ok, thanks budude. that makes alot more sense. So if i dont run AC SSR's and run directly out of the output of the ren to the board with the uln i wont need any ssr's? i just want to make sure im thinking correctly, lol.

if so im really liking this idea b/c for very cheap i can control my RGB archs and not have to use alot of Ren24lv. i could run 2 archs off a ren64. hmmm to much to think about before preorder end, lol.

[dlovely]

ok, thanks budude. that makes alot more sense. So if i dont run AC SSR's and run directly out of the output of the ren to the board with the uln i wont need any ssr's? i just want to make sure im thinking correctly, lol.

if so im really liking this idea b/c for very cheap i can control my RGB archs and not have to use alot of Ren24lv. i could run 2 archs off a ren64. hmmm to much to think about before preorder end, lol.

Yep exactly, I am building 12 Super Strips. That would require 2 Ren24LV's and I would have to be really careful on how many were on at the same time. I can drive all 12 off of one Ren64 and it seems its not much more than the Ren24LV.

Ren24LV
Parts $28.50+$5.95 Ship
Board $14+$5.95 Ship
Total: ~$54.40

Ren64 (No Group buy currently)
Estimated at ~$42 (Parts only, need to add shipping and esd costs, ect) if they hit 50+ units + ~$5.95 Ship
Board $? and no shipping as it would bundle with the parts.

Post with the Ren64 estimates (http://doityourselfchristmas.com/forums/showpost.php?p=103235&postcount=19)

[budude]

Yep - basically you are creating a Ren64LV. You will need an external power supply to drive the LEDs and you'll need to put in current limiting resistors somewhere so the LEDs don't blow. dlovely's idea is a good simple approach to make the modification.

[dlovely]

Yep - basically you are creating a Ren64LV. You will need an external power supply to drive the LEDs and you'll need to put in current limiting resistors somewhere so the LEDs don't blow. dlovely's idea is a good simple approach to make the modification.

Probably not going to be enough interest to do a PCB group buy so I might just home etch the boards I need unless interest makes it worth it, tho I am not at 1yr so I can't host it anyhow.

So with the Ren64, for the PIC DC power code, I don't need the ZC signal if I am not planning on any AC SSRs? Want to run the Ren64 off the same 5V power supply. Can I still use PWM?

[DIY Guy]

Yep - basically you are creating a Ren64LV. You will need an external power supply to drive the LEDs and you'll need to put in current limiting resistors somewhere so the LEDs don't blow. dlovely's idea is a good simple approach to make the modification.

I think it's a great idea. In this setup would we only be using 5 of the 8 strands in the cat5? What's the feasibility of sending voltage down the other strands?

Tim

[dlovely]

I think it's a great idea. In this setup would we only be using 5 of the 8 strands in the cat5? What's the feasibility of sending voltage down the other strands?

Tim

You could, however its a 4 channel input. I was wanting to follow the current pinouts for the AC/DC SSR's for control. Then output to follow the Super Strip pinout so it could be in inline power injector for them.

I am not sure this fits our other application we discussed, but figured others would have interest in this Super Strips version.

[fathead45]

this would work great for my RGB Arch's, im gonna pick up some of these and try them.

[DIY Guy]

Yes, I was thinking the same pinout, but doing a breakout at each end.. instead of terminating those wires in the rj45 at the controller end, connect them to a voltage source.

Same thing at the remote end.

POE standard says 15.4W, so ~600mA@24v per strand; does that sound right?

[budude]

Probably not going to be enough interest to do a PCB group buy so I might just home etch the boards I need unless interest makes it worth it, tho I am not at 1yr so I can't host it anyhow.


Yep - I hear you - maybe just post a pdf of the etch pattern (if not the original files) so others can use it if they want. Home etch is a little extra work but when you want a unique board layout and you want it today - you can't beat it!


So with the Ren64, for the PIC DC power code, I don't need the ZC signal if I am not planning on any AC SSRs? Want to run the Ren64 off the same 5V power supply. Can I still use PWM?

Nope - no ZC needed - I believe there's a resistor to tie the ZC signal high on the Ren64 - leave that there but you won't need the ZC opto or AC in. With that said, it's not a huge savings so you may want to get the parts just in case you want to switch back to AC at some point... You should be able to power the Ren64 off the same supply - if you get a good well regulated one it should work fine. You would want to keep PWM enabled in the code.

I'm by no means a Ren code jockey but I believe the following changes are required to use PWM/DC/ULN2803s in the mix (from Renard-20071229.asm posted in Wiki):

;; PWM_build can be either 0 or 1
#define PWM_build 1 (change from 0)
;; DC_build can be either 0 or 1 (it should be 0 if PWM_build is 0)
#define DC_build 1 (change from 0)
;; CTR_LOCKOUT should be 1 for the DC build, can be a somewhat larger number for non-PWM or non-DC builds.
#define CTR_LOCKOUT 1 (change from 15)
;; For positive_true outputs, remove the following define (normally, this
;; should be left in place).
; #define OUTPUT_NEGATIVE_TRUE (comment out)

I'll find out this weekend actually as I'm doing a similar project but with the controllers on-board as well.

[budude]

Yes, I was thinking the same pinout, but doing a breakout at each end.. instead of terminating those wires in the rj45 at the controller end, connect them to a voltage source.

Same thing at the remote end.

POE standard says 15.4W, so ~600mA@24v per strand; does that sound right?

Actually, POE now supports 20W at 48V so at least ~400mA anyway. I'm running my SuperStrips just fine with Cat5 and it's 3x360mA (soon to be 4x) per pair (remember the current returns back on the commons). I would suggest using a standard length cable to all your endpoints since you will get different voltage drops, the LEDs will appear dimmer with longer cables so they will look inconsistent. I cut all my strip leads to 30' and adjusted the voltage at the end appropriately. Yes - you have to deal with lots of extra cable closer to distribution but it's worth it.

[P. Short]

There are a few discrete darlington transistors that look like they would work at the 400mA-500mA output level, such as the KSP13 and MPSA13. You would need base resistors as well...

[chilloutdocdoc]

Here's a very nifty chart for figuring out how much is too much.

Taken from: Here (http://www.datasheetcatalog.org/datasheets/105/366825_DS.pdf)

I would say it all depends on if you plan on using the super strips 100% or, adding them in for effects, and considering that dimming also has a play, 50% dimming automatically cuts it to 50% duty cycle.


Edit: Didn't see that there were 3 pages of discussion before I posted this.. OOPS!

[DIY Guy]

Are all outputs created equal? Or when using multiple outputs in parallel, will current have more of a tendency to rush through certain outputs on the uln?

Can we assume it will be divided equally?

[dlovely]

Are all outputs created equal? Or when using multiple outputs in parallel, will current have more of a tendency to rush through certain outputs on the uln?

Can we assume it will be divided equally?

Probably cannot assume it will be equally divided on multiple. The ULN2074 was suggested and has a higher limit so would not need multiple in parallel.

[dlovely]

This is kind of what I was thinking of making. Still need to learn how to use Eagle, This is the first project I have made using it.

Size is 2.3" x 0.75"

Dave

[budude]

Your output traces will probably need to be beefed up a bit - you can use THIS (http://www.circuitcalculator.com/wordpress/2006/01/31/pcb-trace-width-calculator/) to figure out how much.

[dlovely]

Your output traces will probably need to be beefed up a bit - you can use THIS (http://www.circuitcalculator.com/wordpress/2006/01/31/pcb-trace-width-calculator/) to figure out how much.

Yeah figured that, Need to learn Eagle a bit more. I also could not find the right terminal so just used one I found in the lib. So need to verify that will work too. Also want to change the RJ45 from top plug to angle ones.

[fathead45]

i like it. man i really need to get home etching started, lol

[P. Short]

The ULN2064B might be a more useful part...it is similar, but has internal catch diodes for use with inductive loads.

[P. Short]

And, to play the devil's advocate, do you really need the higher peak current capabilities of the ULN2074? If you are looking for a higher continuous current capability, the ULN2074 is not any better than the ULN2803 (with doubled outputs). The continuous current capabilities of both parts is limited by the power dissipation to more or less the same value, and the ULN2803A is a lot cheaper and easier to find.

[christmas-light]

i like it. man i really need to get home etching started, lol

How do you want to do that :confused:

I can see that it is doubel layer pcb

[budude]

And, to play the devil's advocate, do you really need the higher peak current capabilities of the ULN2074? If you are looking for a higher continuous current capability, the ULN2074 is not any better than the ULN2803 (with doubled outputs). The continuous current capabilities of both parts is limited by the power dissipation to more or less the same value, and the ULN2803A is a lot cheaper and easier to find.

True - but I would think that you might get more life out of the 2074 than doubled (or quadrupled!) 2803's. I only pointed to the 2074 since it is used on the LEDTriks and seems to work fairly well. The datasheet I found for the 2074 showed the entire "family" of parts including the 2064 you mentioned as well. I didn't look at pricing - but yes - the 2074 is twice the price of of a 2803 so it may not be a good way to go if saving pennies.

[DIY Guy]

What about Phil's mention of the KSP13? At .07/ea and .09/ea for resistors, it's a total of $1.28 vs $2.41 for the ULN2074.

If you wanted to bump up the current capacity, perhaps the MPSA14D26Z at .09/ea with a current handling of 1.2A?

It may also shrink the size of the board.

Tim

[smartalec]

I've thought about using the ULN series of chips, but there seems to be hassles with how much current it can handle.

Thats why i'm thinking of my old circuit with its outputs, eg 2 transistors
it would probably work out like $1 a channel, so the ULN might be cheaper but i feel transistors can handle the current an heat better.
http://doityourselfchristmas.com/forums/showthread.php?t=10967
thread of my progress using transistors.

thats my 2c worth.

[aususer]

Venturing in here looking for a way to link my (yet to be delivered/built) PLCC6 RGB strings to my REN64. (been leaning towards aussiephils method)
I see that I could either make a mosfet-based driver circuit - or persue this procees...
Both logical.

re: the input logic: I had a thought - for the addition of a <$1 why not just add an logic inverter (ie. CD4069UBC) to the inputs and avoid the firmware change/specificity?
The inverter should flip the input logic, and therefore avoid having to allocate specific firmware/pic. You could even mix channels on a REN64 (as I want to do I only want 24 channels initially - the rest are SSRez's)

Would that work?

[smartalec]

i used a 4049 in my test on the breadboard, but i dont think it will be fast enough, where the 74HC14 would be faster switching.
an then run that into the ULN2003


Venturing in here looking for a way to link my (yet to be delivered/built) PLCC6 RGB strings to my REN64. (been leaning towards aussiephils method)
I see that I could either make a mosfet-based driver circuit - or persue this procees...
Both logical.

re: the input logic: I had a thought - for the addition of a <$1 why not just add an logic inverter (ie. CD4069UBC) to the inputs and avoid the firmware change/specificity?
The inverter should flip the input logic, and therefore avoid having to allocate specific firmware/pic. You could even mix channels on a REN64 (as I want to do I only want 24 channels initially - the rest are SSRez's)

Would that work?

[P. Short]

How fast do you think it needs to be? The spec for the cd4069ubc show sub-microseconds delay...which I think should be fine. You also might want to consider an 8-channel buffer part, like a 74hc240 or 74hc540.

[smartalec]

i did'nt know there was 8channel ones, that makes it even better for my original design for my arch's, thanks for that p.short


How fast do you think it needs to be? The spec for the cd4069ubc show sub-microseconds delay...which I think should be fine. You also might want to consider an 8-channel buffer part, like a 74hc240 or 74hc540.