Using Capacitors on a PCB

Question

So I have built a custom PCB with a NRF24L01+ 2.4Ghz transmitter breakout board on it. I have heard the modules are incredibly sensitive to input voltage, so I have it connected to its own dedicated 3.3V regulator and a 100uF tantalum capacitor. When I turn this board on, I still have a significant amount of drops per minute. When I take the tantalum capacitor off the SMD pads and solder a 100uF Aluminium Electrolytic capacitor on the pins of the breakout board, the problem seems to go away.

I have the schematic and board shown below.

After doing a significant amount of testing, I have found the following results and plotted them on the following graphs. I let each test run for 10 minutes and then averaged the drops per minute. I kept one of the NRF modules on another PCB connected to a Aluminium Electrolytic capacitor on its pins, and changed the others capacitor configuration.

Can anyone tell me why this is? I can't figure out why just moving the capacitor from the SMD pads to the pins of the NRF increases stability. I also don't understand why an Aluminium Electrolytic capacitor would be better than a tantalum capacitor, but that seems to be what the data is showing.

Another issue that I found is that when no capacitor is connected to NRF module at all, this line of code stalls and stops. Anyone have a clue why that might be?

  if (!radio.write( &myData, sizeof(myData) )) {    // Send data, checking for error ("!" means NOT) 
Serial.print("Transmit failed ");                   //When Capcitor Not Conencted, Code Freezes in this if Statement;
count=count+1;
RF_Flag=false; }

If anyone can help me with any of these issue that would be greatly appreciated! I cannot seem to figure out why this is.

Answer 1

Without more details on the capacitor specifications we can just speculate on the exact reasons, however there is one problem that clearly jumps out.

Technically speaking, your power traces are crappy. Their resistance and inductance would be too high for a power supply trace. This introduces power droops and ringing into the system.

If you have too good of a capacitor (low ESR) ringing will become a problem (this could be why tantalum is worse) and, regardless of the capacitor type, the thin segments of trace between it and the connector will introduce inductive and resistive droops in the power supply.

Add some wire between the connector and the capacitor to reduce trace impedance and start from there

---

Oh, and regarding that stalling line of code. If the power distribution to your microcontroller is as bad as in that small section you posted, it’s very likely that power to it is dropping too much when the RF transmitter turns on.

Answer 2

It could be due to:

• Capacitor ESR

The cap with lower ESR (within reasonable limits) will provide the best decoupling. The tantalum is specced around 1-1.5 ohms ESR, and a junk alu cap from the bottom of the drawer would be about the same or higher, unless it is a low-ESR model. So, this isn't conclusive. Also the RF chip's current draw is pretty low, couple tens mA, so this shouldn't matter.

• Layout

Your power and ground traces are really thin, which increases resistance. This layout looks like it was done by autorouter, which is not a good sign...

• Connector malfunction

When the capacitor is "on the pins" it works better than when soldered on the board. This isn't normal. If the connector is defective and doesn't make a good contact, a capacitor soldered on the module would work better than on the carrier board (on the other side of the defective contact). Of course, fixing the contact problem is the correct solution here.

• Regulator instability

This depends on regulator, caps (including ones on module) and layout. Having a scope would definitely help to check VCC voltage drop when the module transmits.

Answer 3

1. Should have used a ground plane on top & bottom layers vs running 6 or 8 mil wide ground trace around the board.
2. Power trace should also have been a lot wider.