Custom board with nRF5340 hardware connection issue

[Maxoravn]

Background:

Currently working on a project involving the use of a servo motor, and encountered a problem with the length of the wire used to control the servo. The servo operates correctly when connected to a shorter wire, but when using a longer wire, it does not function as expected. My colleagues and I could not figure out the problem. The long wire works with an arduino but not with the custom board programmed by the nRF5340DK and so we thought it has something to do with the DK and not the physical connections.

Issue Description:

The servo motor is being controlled by toggling the serv2/serv3 GPIO pin (custom board) on and off with the appropriate delay. When connected with a short wire, the servo operates smoothly and responds to given signals. However, when we switch to a longer wire, the servo fails to respond at all. We are programming the custom board using the nRF5340 DK.

We have tried measuring the signals from the pin using an oscilloscope (images attached). As the wire gets longer, the signal becomes worse. The first image shows the expected signal (short wire) whereas the second image shows the signal we get from using a much longer wire.

/cfs-file/__key/communityserver-discussions-components-files/4/20240124_5F00_124131.mp4

I have also attached our code set in a zip file. It contains all the necessary files to run the sample program (toggling the servo motor) to work on our custom board. Basically, we just plug the servo motor (MG92B) to our custom board serv2/serv3 pin. When we flash servotest_main_v1.c as main, it should hold the servo at a certain angle. Again, the servo will only turn when using the short wire and not the long one.

project.zip.zip

 

Additional Information:

• Long wire works fine when using an Arduino
• Tested resistance with multimeter and signal with oscilloscope
• Already tried making a new wire and checked physical connections.
• Using serv2 and serv3 (configured as &gpio0 2 and &gpio0 3 in .overlay file)
• Using VS code and nRF extension to build and flash the application.
• OS: Windows. nRF toolchain version: v2.5.1
  

I am a beginner and would really appreciate your help with this issue. Thanks! 

[Naeem Maroof]

Hi Francis,

As you have mentioned that using a short wire there is no issue but using a long wire there are issues.

In fact, the resistance, capacitance and inductance of long wire could imploy signal loss that you are having.

I do not know how long the wire is, but maybe you could use a good quality wire and not some ordinary one. For example like the wires used in radio signal testing. Or probably you could also think to reduce the wire length for better purposes.

As you have mentioned the arduino, the voltage level that NRF SoC supports vs what arduino board supports could be different.

You could also look to use the high drive strength for the GPIO

(from the product specification: PIN_CNF register on page 233 nRF5340 (nordicsemi.com) )

Regards, Naeem

[DubiousDooby]

Thanks Naeem. Francis currently has a cold and I'm the colleague. I'll be looking into this shortly.

[Maxoravn]

Thanks for your response Naeem.

The wire is only about 7-8 feet long and is directly connected from the custom board to the servo.

The voltage reading for the custom board is 5V and the arduino has 4.9V/~5V. The Arduino code that we are using to test the signal is working fine for the long wire and gives us a proper signal output. However, when trying this test to the custom board, it only works for the short wire and the signal becomes worse as the wire gets longer. I am not sure if this is a software issue because we are just toggling the gpio pin on and off to give a PWM signal to the servo motor.

As for the high drive strength for the GPIO, I tried enabling it for pin 2 and 3 (used for servo2 and servo3) but did not noticed a difference. I tested the signal again using the oscilloscope I am not sure if these pins support this drive configuration.

Currently, my colleague is working on a transistor circuit to possibly amplify the signal. We are also looking into getting a better quality wire.

Would there be other solutions that might help with this issue? Thank you so much again for the help.

2024/1/30 UPDATE:

Hello. we have also performed an experiment where we stripped the wire completely and measured the signal at different points. As we go down the wire, the signal goes from 5V logic to 3V/1V logic digitally (not a continuous drop). This results in a bad signal as we use a long wire. We believe that this is a microcontroller problem rather than a hardware/electrical issue based on the tests that we have done. Thank you again.

[Maxoravn]

Hello,

Here are additional information on the experiment:

In our experiment, we noticed something interesting with the signal along the stripped wire on our custom board using nRF5340dk. So, the stripped wire connects to the servo pin for the signal output. As we move down the wire, the voltage shifts from 5V to 3V in a digital way – not smooth.

We checked the signal using an oscilloscope, we found that the amplitude of the square wave changes suddenly, as we progress down the wire. It's kind of a quick switch rather than a smooth transition. Attached some pics for a visual.

Experiment setup:
The nRF5340 is connected with the custom board. The stripped wire is connected to the servo pin that we are using. The oscilloscope shown is what we are using to measure the signal output. The program in the custom board is a simple program that holds the servo at an angle (sends PWM signal; which is attached in our project zip file main.c)

Results:

Start of the wire:
  Setup
 Signal output

Around the midpoint:

Past the midpoint (before the voltage changes in logic):

Around the end of the stripped wire (where the change in voltage logic happens):

Note that this is also the signal that we are getting when using a proper wire that is of similar length.

I have also attached a zip file of the results we get at different points of the wire. The files for the custom board are in the project zip file under "tricca_cboard".

stripped wire test.zip

Your insights on this issue would be super helpful! Thanks again for looking into it.

[hmolesworth]

The MG92B is a 5 volt servo; the nRF5340 port pins cannot drive more than 3.6 volts absolute maximum regardless of port pin drive strength whereas the Arduino is using 5 volts. Some form of level-translator buffer is really required;  this could be a simple FET of some other driver.

Worth considering changing the oscilloscope probe settings to x10 probe settings instead of x1, assuming there is such a switch on the probes. Sometimes a x1 probe can prove misleading since it loads the circuit with 1MOhm; x10 is 10MOhm.

7 to 8ft is a huge distance in terms of port pins, by the way.