FEDEVEL
Platform forum

simeon , 03-04-2025, 11:24 AM

Hey. I just realized by accident that I made a huge error in my schematic.. For the RFID module, I used 5V instead of a 3V3. Now, the 5V makes a polygon connection towards from the RFID module towards the RFID display. I am thinking of three options right now to fix this mistake. I am in doubt because I am not sure which one's best to minimize the EMI whilst also not making the current path too large.

**option 1:** connect LCD 5V through right side polygon
The 5V polygons circles the PCB. Here, crossing the RFID tracks is avoided but *doesn't this make the current path from the power source connections which are on the left to the 5V LCD too long (i.e. larger voltage drop?*

**option 2: **
I continue the 5V polygon form the point where the two vias are. This is the shortest but crosses right through the spi signals.

**Option 3:**
connect from the outside board. This is also possible, but then the 3V3 that's replacing the 5V of the RFID module will be interfering/crossing each other basically.

**How the 3V3 will be routed for the RFID module:**
as seen in the picture.

Robert Feranec , 03-04-2025, 12:14 PM

I would just do it the simplest and fastest way and made a note do not forget to fix it in the next version. I would probably just made two cuts from both sides to isolate the pin, connected it to 3V3 and connected the cut pieces together.

QDrives , 03-04-2025, 08:00 PM

Is the board already assembled?
If not, you can drill a big hole in the 5V pad so that the pin does not connect the PCB (eg 2mm).
Then solder wire directly to the pin.

simeon , 03-06-2025, 01:41 AM

Thanks for the help guys!

simeon , 03-06-2025, 01:43 AM

Nope, not yet. I had to make the ESP32 PCB today. So, glad I caught it by accident 🙂

simeon , 03-06-2025, 01:54 AM

Edit: now that I read it again... I realize that it might've sounded that I already have the PCBs. If I'm not mistaken, the suggestions made are for PCBs that I have physically? So, with this one.. I haven't ordered them yet. I was about to order them, and caught this error by mere luck.

My question was more in terms of what the best of THE 3 option would be to fix this error in the PCB layout with minimizng EMI. Ofc this only cuts through SPI which not high speed, but it's always better to not cut through such tracks, right?

simeon , 03-06-2025, 01:57 AM

I implemented the easiest fix here. 3V connection at the outside of the NRF and 5V reaches the upper pin by following 270 degrees ccw. I don't know if this is best though.. good for EMI, but the track to reach the 5V pin of the LCD display is quite long?

simeon , 03-06-2025, 03:00 AM

In the mean time: Improved the desgin in the 3V area a bit, if it's of any use.. I had the nrf module placed into the board. Decided to place the female header more inwards so that the antenna itself sits at the outside of the board. I predict that this helps with EMI as well.

Robert Feranec , 03-06-2025, 06:22 AM

ah, if it's only in a file, you can do it any way. that is a very simple board, even if you need to reroute something it is not going to take much time. Also, that is only a 2 layer PCB it means the layout will never be perfect. PS: I noticed you have the 5V in a loop - usually for power I try do not mix high current paths with low current paths so I would route the connection from the source to destination separatelly for different kind of currents.

Robert Feranec , 03-06-2025, 06:22 AM

but if these are not high curents, I would not think too much about it

simeon , 03-06-2025, 10:57 AM

Yep, I know it's 2-layered, but I always worry about EMI too much and don't really know when to say enough is enough in terms of perfecting the layout.😅 There has to be a point where I stop, as you said it, 2-layered will never be perfect...

And thanks for the advice, that's interesting! I'll keep that in mind for the nxt revision as I just sent it in. Right now, I have a servo motor taking quite a lot of current under load which is also immediately connected through the same route with a keypad, buzzer, and LCD display
If I think about it now, it totaly makes sense. I can imagine the noise from all of these different modules causing constructive inference on the 5V line which a 100nF cap doens't fix.

QDrives , 03-06-2025, 04:33 PM

If you do like option 2, add some Gnd return jumpers parallel to the signal traces.
Like the green lines in this altered picture of yours.