1
\$\begingroup\$

Can someone please review my PCB? Simply, I'm running 24 IR Leds at medium power, from a 5v arduino MCU pin, all pulsing together (38khz). Big fat traces, lots of space, SMD components. The LEDS are in 8 3-serial runs, and I have power/ground traces around. The "leds" on the board are just 2.54 headers, as I'll be running them off-board. But I don't know if I've got my head wrapped around the mosfet circuit correctly. Please be kind (but feel free to talk to me like an idiot, because I am) :) I'll fab at seeedstudio if it's relevant.

enter image description here

Edit: I believe this is the schematic for the board as is currently laid out (with the exception of lowering the gate resistor as a commentor suggested). Also, I do believe I'm an idiot with the leds, as I'm bridging their legs, rather than just the leds, so can remove those trace portions (ie on the top left led, 1 and 2 are the +- legs of an led so ought not be 'traced', but 2 and 3 need the trace to connect led1 to led2.):enter image description here

schematic

simulate this circuit – Schematic created using CircuitLab

\$\endgroup\$
  • 3
    \$\begingroup\$ A schematic diagram would be helpful... \$\endgroup\$ – Peter Bennett Dec 23 '15 at 3:08
  • \$\begingroup\$ Judging by the trace colours, you've got a 7 or 8 layer board (I lost count :-) ) I don't see any holes in what I assume are your header positions for the LED connectors, and have no idea what the red zig-zag patterns represent. \$\endgroup\$ – Peter Bennett Dec 23 '15 at 3:17
  • 3
    \$\begingroup\$ My eyes are bleeding looking at that. \$\endgroup\$ – Ecnerwal Dec 23 '15 at 5:02
  • 1
    \$\begingroup\$ I added what I believe is the schematic for the board as drafted, to the OP \$\endgroup\$ – dirtyharry2 Dec 23 '15 at 18:50
  • 1
    \$\begingroup\$ If your schematic is correct, and looking at your PCB, they seem to match. You have 2 of the 3 LEDs shorted out. \$\endgroup\$ – Tyler Dec 23 '15 at 19:00
5
\$\begingroup\$

Whew, what a color-riffic board! That said, please try to use one color for each layer, such as red and blue. That's the "standard" way of doing it and most others will better understand it represented this way.

The AOD510 boasts "Latest Trench Power AlphaMOS (αMOS LV) technology, Very Low RDS(on) at 4.5VGS, Low Gate Charge". Now to extrapolate on that, it is an N-channel power MOSFET. With 4.5V on the gate, it's Source-Drain resistance will be ~4mΩ. Which is really good: when "on", the LED's will be fully on, with minimal loss at the FET.

However, not explained, is that these power FETs typically have very large gate capacitances. From the datasheet, \$C_{ISS} = 2719pf\$. What this means is that your CTRL signal, fed through the 10k resistor to the gate, combined with the Ciss of the gate, forms a series RC filter. Ultimately this behaves as a low-pass filter... i.e., as the CTRL switching frequency increases, the resistance and capacitance (RC filtering) will diminish the effective signal.

If you do the math at that link, you'll run across

\$f_c = \frac{1}{2\pi R C}\$ ... solving for frequency "corner",

\$f_c = \frac{1}{2\pi \times 10k \times 2719pF}\$

\$f_c = 5853Hz\$

So at 5.8kHz, using a 10k resistor and device with Ciss of 2719pF, you will have already lost half of the signal due to the R/C combo.

Now you can reduce this low-pass filter effect by choosing a smaller value for R. If you re-solve the formula for R = 1kΩ, the cut-off frequency becomes 58kHz. The drawback to this however is, CTRL has to source and sink more current - each time the gate is toggled.

This is the drawback to power mosfets - their large gate capacitance makes switching them quickly problematic. You should experiment with this, and don't be surprised if you have to add a "driver" element to make that AOD510 switch as quickly as you'd like.

\$\endgroup\$
  • \$\begingroup\$ Thanks; I'll limit colors when posting. I found it helps me visualize the nets, but I get if others see it as nonstandard. As for the rest, it sounds like the MOSFET circuit is wired correctly to the MCU and LEDs then? You're just worried about kHz, and I can tinker with the pulldown resistor for that? I've seen so many schematics, all different, that I get confused. Some have some form of voltage divider and 1M res! \$\endgroup\$ – dirtyharry2 Dec 23 '15 at 5:05
  • \$\begingroup\$ It should work, if I can understand the wiring correctly, but the performance may not be exactly what you want. Just be prepared to tinker with it and make a few revisions. \$\endgroup\$ – rdtsc Dec 23 '15 at 17:05
  • \$\begingroup\$ I added what I believe is my current circuit in schematic form to the OP. \$\endgroup\$ – dirtyharry2 Dec 23 '15 at 18:49
  • \$\begingroup\$ Next you may want to consider what current to place through your LEDs. Current = brightness, but each LED has a maximum limit. Check the datasheet for your LEDs, then look at this post: electronics.stackexchange.com/questions/17179/correct-formula-for-led-current-limiting-resistor. \$\endgroup\$ – rdtsc Dec 23 '15 at 23:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.