Jetson Nano Gpio interrupt lagging

[mohamed-keshk]

I am using the Jetson nano GPIO to read the output of an AB bi-phase rotational encoder encode, I wrote a simple python script to trigger a rising edge interrupt, the callback function prints the current state of the pin that triggered.

import time 
import Jetson.GPIO as GPIO 

# Define the GPIO pins for the rotary encoders 
PIN = 13

# Initialize GPIO GPIO.setmode(GPIO.BOARD) 
GPIO.setup(PIN, GPIO.IN) 
def callback(PIN): 
   print(GPIO.input(PIN))

# Add event detection for rising edges GPIO.add_event_detect(PIN, GPIO.RISING, callback=callback, bouncetime=1)

try: 
   while True: 
      pass   
except KeyboardInterrupt: 
   pass 

# Clean up GPIO 
GPIO.cleanup() 

After a lot of debugging and many many tries, I stopped using the actual encoder, I’m using a function generator set to output a 3.3v 500hz square wave on pin 13 on the Jetson nano. The output I expect should a stream of ones, but the actual output is random mix of ones and zeros.

The actual output from the terminal: 1 0 1 1 1 1 1 1 1 1 0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 0 1 0 1 1 1 1

What am I missing or doing wrong. This my system: L4T 32.6.1 JetPack 4.6 Ubuntu 18.04.6 LTS Kernel Version: 4.9.253-tegra

[23100048]

Having the same problem. Should be getting a constant stream of ones but get a mix of ones and zeros when the input signal is high. @mohamed-keshk, Can you please share any information on how to fix the issue?

[anhmiuhv]

It is possible that the sampling rate is not high enough. In that case, you should write a c program to read sysfs or the gpio registers directly.

[mohamed-keshk]

@23100048 I'm sorry to disappoint you but I couldn't find a satisfying answer to why that happened, nor have found a solution. I abandoned the project, I mean to get back to it some time later, but till now, I simply can't read the encoder reliably. @anhmiuhv the sampling rate of the jetson nano gpio is a hell lot higher than the 500 hz signal I used to test it. If it can't handle such a slow signal, I guess we'd be better off using an uno.

[anhmiuhv]

Yeah, it is likely that you will need to use real time kernel to reliably sample. Uno is probably easier

[anhmiuhv]

If you still want to use Jetson Nano, it is worth exploring options to implement in C using libgpiod https://www.ics.com/blog/gpio-programming-exploring-libgpiod-library or sysfs https://www.ics.com/blog/how-control-gpio-hardware-c-or-c . Another option to explore is to read directly from Jetson register (for examples, https://github.com/mdhardenburgh/jetsonNanoRegisterAccess).

I think this is a good blog (albeit for Jetson TX2). https://www.lancesimms.com/JetsonTX2/JetsonTX2MappingGPIOs.html

The Python library contains some overhead.

[23100048]

Thankyou so much for your responses. Yes, I have shifted the process of reading the input signal to an arduino (mega 2560) and am communicating that via rosserial to the jetson nano. That has solved the problem.