LED PWM Driver

[anearey]

Hello, I've been trying to design a driver for an LED panel with approximately 700 LEDs on it. I've drawn out a PCB, based on a PWM circuit I found on the internet, which I altered a bit to fit my purpose more. I plan to mirror and transfer to a piece of copper clad FR4, via the "Toner Transfer" method. I've already tried one PCB, but that design did not work for some reason. I think I may have ended up oversimplifying it to try and save space. Here is the new PCB layout that I've designed. Does this look like it should work?

Component specs are as follows:

C1-----------0.1uF Ceramic Cap
R1-----------4.7kOhm .25W Resistor
SD (2x)----1N4148 Switching Diodes
DIP555-----NE555 Timer
IRF530-----14A Power MOSFET
VR1-3-------Leads for 10k Linear Pot

W2 and W8 will connect by wire to their respective pair

 

[(*steve*)]

Perhaps if you can show us the circuit, and how you plan to connect the 700 LEDs.

Have you read the sticky post on driving LEDs?

 

[anearey]

Hi steve,

You've actually helped me in the past with the same project, and I now am coming back here for help with my dimmer circuit haha.

Here was the previous thread I started:
https://www.electronicspoint.com/led-panel-circuit-t226268.html

 

[(*steve*)]

Cool, so the panel is going to be well behaved

Can you point me to the PWM circuit? I presume you are using a 555 and a pair of steering diodes to get variable mark space ratio between 0 and 100% (or thereabouts).

The main problems I see are:

1) You will have to mount the 555 on the copper side of the board (or your text will be in mirror writing)
2) There is no track leading from the diodes to the pads used to connect your variable resistor.
3) The pads for components are a bit small. They may be difficult to drill and/or to solder to. Make them larger (in the case of the IC pads, make them longer).
4) Make the "hole" in the centre of the pads about 0.5mm -- that way it will act as a guide for your drill and make drilling far more accurate (your drill will be larger than 0.5mm).
5) If you have the track connected to pin 3 of the 555 going between the 2 rows of pins (i.e. exiting through the inside rather than the outside) you will save yourself a long track snaking around the board.

That's all at the moment. Give me another look when you've fixed them. Note that I have not made sure the circuit is correct!

 

[anearey]

Thanks for the assistance once again. My knowledge of these circuits is quite limited and I'm having a hard time figuring out whether or not to leave pin 5 detached from the rest of the circuit, or (I've heard this elsewhere) that I should have it sent back to ground through a 0.1uF cap. Very confused.

As far as the revisions you mentioned;

1.) Is there an issue with soldering the IC to the copper side? I mean, the first version of this circuit I made, I used a SMD NE555 chip (because I ordered the wrong one ).. could that have caused an problem?
2.) oops.
3.) I've adjusted pad sizes the best I could, and I've had to entirely remove the silkscreen layer (not really necessary anyways, since I'll have to take it off before printing)
4.) Holes are as small as possible (a little larger than 0.5mm)
5.) Thats how I had it on my old design, I just wasn't sure if that was a big PCB "no-no" but if not, I'd rather do it that way, myself.

Thanks again!.. heres the revised board

 

[(*steve*)]

1) Not really, other than it looks odd. You may find it hard to solder both sides of some pins, so you have to be careful if there are conductors joining under the component that may have been severed by drilling (make the pads larger is the answer)

3) Leave the text that you have on there now. Documentation in copper is always a good thing if you have the space (and it doesn't short stuff out or cause other problems). Make the IC pads as large as possible consistent with leaving good clearance between them. Typically gaps will widen when you etch at home.

5) I've made that change. It's fine. In your case, if you're mounting the IC on the copper side, you can bend the leads flat at the end (where they get narrow) and simply don't drill the holes. I do "surface mount" of DILs like this fairly often. I also gets around the problem of your drilling cutting the pad in half.

Here is a revision showing what I mean:

Oh, and retain the rectangular pad for pin 1 and rounded pins for the others -- it's good practice.

 

[anearey]

I must have uploaded the wrong file somehow, because I had actually fixed those traces. But I now see what you mean by extending the pads in order to make drilling easier. I went ahead and flipped the board so that I could mount the DIP to the reverse side of the board. I made sure directions and connections were consistent with the original, as to not wire something backwards.

As far as the circuit goes, do you think this should function properly??

 

[(*steve*)]

As far as the circuit goes, do you think this should function properly??

As I've said before... Show me the circuit diagram and I'll try to check that for you.

 

[anearey]

This is what my dimmer was modeled after:

http://pcbheaven.com/circuitpages/LED_PWM_Dimmer/

 

[(*steve*)]

I am not particularly enamoured with that circuit. The issue is the way the load is in the emitter rather than in the collector. It means that it is much harder to turn the transistor on.

If you're using a P channel mosfet as a high side switch, or an N channel as a low side switch, then you'd be right.

OK, you have an N channel mosfet and... it's set up as a high side switch. That's the wrong way around. And the way you have the mosfet, it will always be conducting through the body diode. I'd recommend you get a P channel mosfet and place it in this circuit (that seems to be an IRF9530)

Here is a page discussing high side switches. Note that the 555 has an open collector output on the discharge pin, so, with the resistor you have to +12V, you have the equivalent to what they recommend for switching the MOSFET.

The fact that I'm seeing *everything* backward makes me wondering if I've made some fundamental error looking at your circuit. Get a second opinion

OK, I've checked the circuit and it looks like it's right apart from my comments above. I finally understand what W2 and W8 are! I have made a few changes to eliminate them.

As you no doubt imagine, the board can now be made substantially smaller. I'd also consider making the high current tracks fatter, just because you can There is also probably no need to have all the power connections taken to one point as you have here. The parallel traces from the +12V could be a single trace.

 

[anearey]

Ah, I think I understand now, P channels are for altering positive voltage, while N channels are for negative. (High/Low) Yes? And since I have my circuit turning on and off my +12VDC supply, I need a P channel MOSFET which would correspond to the 5A load a IRF530 would have taken on if I were using it on the low side of the circuit. I would actually use a IRF9540, due to the fact that it is more readily available to me than the IRF9530. The only difference would be the higher current rating, correct?

....long story short, putting an IRF9540 in place of where the IRF530 is now should solve the problem?

I redrew the circuit by hand (rather than mirroring the image and fixing the text in a graphic editor, like I did previously) so that I could be more sure of the connections, and I changed the N channel out with the P channel

 

[(*steve*)]

Well, an enhancement mode P-Channel device needs the gate pulled more negative than the source to turn it on. And an enhancement mode N channel device needs the gate pulled more positive than the source.

It is just easier if the source is close to the opposite supply rail, Otherwise you need to pull the gate above (or below) the supply rail, which is possible, but is more work.

 

[anearey]

i just saw your edit on your last post, which is good to know. But I should I really make the high current traces larger? They're currently at 0.100" width, and the low current ones at 0.060"

And I think I understand what you clarified about the N and P channel mosfets, so the way I have the P channel mosfet configured on the board should be functional, since it is pulling negative charge on the gate?

 

[anearey]

possibly a final?..

 

[Mitchekj]

I hate to throw in more complexity, but 700 LEDs is a lot! That says to me there's going to be tons of parasitic capacitance (and resistance/inductance) in all the wiring, traces, and the LED dice themselves.

I would expect some issue with the pwm acting as expected. Namely the output of the dimmer not going to zero each cycle. This may or may not be an issue, but I have seen this cause some serious flicker when you attempt to dim.

The best guess I have is that the level your output drops to is not consistent. It drops to only 2 volts one cycle, a few cycles later it goes to only 7 volts, etc. Due to all the parasitics and the time constants which are created due to their interaction. This may be creating some fundamental or harmonic frequency of the PWM frequency, which would be visible. That's just my theory though. I know the effect happens though, I've seen it myself.

Now 700 LEDs makes me nervous. That's a lot of parasitics. I think you may have some issues getting them to PWM as you'd like. I've been able to fix the issue in the past by using a resistor across the dimmer's output to damp things. Then all my cycles went completely to zero and the flicker went away. With so many LEDs you may need to insert more resistors across your bus down the line in strategic locations. But I don't know for sure what the extent of any issues you may have will be.

Just hoping to prepare you a little bit for some possible heartache.

 

[(*steve*)]

I'd say that's looking much better.

 

[anearey]

Mitchekj, thats a very interesting phenomena you are describing, but how do large manufacturing companies like LitePanels create such a large array with success? This is essentially what I am trying to emulate since I don't exactly have 2 thousand dollars to spend on a single light source. I've never heard of these panels having a natural resonance, and they are a tried-and-true reliable light source for video with no flicker or strobing??

http://www.bhphotovideo.com/c/product/708734-REG/Litepanels_LP1_MS_1_x_1_Mono.html

 

[Mitchekj]

They solve it (probably) similarly to how I did it. I was just giving you a heads up that you may get some unwanted happenings when you build it, now you'll know what it is if it does happen.

 

[anearey]

Haha, I'll definitely keep an eye out for it. I may tack a resistor on the end of that circuit too. Thanks!

 

[anearey]

thanks so much guys! works like a champ!!