Acquired data using MCC 172 with IEPE sensor have spike noise.

[tnag]

I have been using the MCC 172 HAT with PCB's IEPE sensor to acquire vibration signals. 
Using the HAT with my Raspberry Pi 3B, the Python sample code provided by MCC worked well without any noise.
If I switched the MCU to Raspberry Pi Zero 2W, the spike noise appeared at every measurement. 
Sometimes the spike seems periodic, but not always the case. 
Attached figures are the graphs of the raw voltage vs. time.

Measurement configuration is:
- Single channel input
- IEPE enabled
- 25.6kHz sampling rate
- 11 second measurement time duration
- Shorted circuit input signal
- Buffer size is 100,000
- Sensitivity is 1000.0 (no scaling)

Test I have tried on RPi Zero2W (but not worked):
- Rasbian 32bit Lite OS and 64bit Lite OS
- Lower/higher sampling rate
- Increase/decrease sample buffer size
- Overclock CPU to match with RPi3B clock status during acquisition
 arm_freq=1300
 over_voltage=4
 
 Any input would be greatly appreciated.
 
 Tatsuro

[Jeffrey]

Hello @tnag

Based on the information you provided above, You have proven the MCC 172 performs admirably when operated with the Raspberry Pi 3B.  The noise occurs when you use the Raspberry Pi zero 2W. 

Meaning the problem is not the MCC 172.  Suggestions:

Try a different RPi Zero 2W;

Try a different power supply than the one you used with the RPi Zero 2W or check it's output for noise;

Contact the vendor of the Raspberry Pi Zero 2W and provide them you test results.

Regards,

Jeffrey

[tnag]

Hi Jeffrey,

Thank you for the prompt response.
As you mentioned, my MCC 172 HAT does a great job in acquiring vibration signal for sure. 
It could actually measure even in the higher frequency domain ~10kHz without major amplitude attenuation.

Previously, I have tested with several RPi ZeroW (not Zero2W), and the results were the same.
I thought this was due to the lack of CPU power, and so this was why I purchased Zero2W.
I cannot perform test with a different Zero2W since I only have one Zero2W. It is hard to get RPi these days.

Regarding the power supply, I have used the same 5V/4A power supply for RPi3B and Zero2W.
To make sure that the noise does not come from the power source, 
I tested both RPi with the 5V/2.1A mobile battery today. Unfortunately, the results were still the same.

Do you think OS related loop time jitter could cause SPI communication unstable?
I have changed the priority or niceness of the python job, but there was no diffrence.

Best regards,

Tatsuro

 

Edited October 12, 2022 by tnag

[Jeffrey]

Hello Tatsuro,

The onboard micro on the 172 handles the SPI communications to the ADC and buffers the data, so it is not affected by the Pi processor load.  As long as you don’t get overruns the data should be considered valid.  Do you see the same spikes if you measure something without IEPE?

I posed your issue to one of our hardware engineers.  After reviewing the schematic of the Raspberry Pi Zero2W, here is his response

The MCC 172 is powered by 5V passed directly from the micro-USB connector, but due to its small size there is not much room for capacitors on 5V.  You may have better luck if you find a way to add additional capacitance to the 5V net right at the Raspberry Pi (between the GPIO header 5V and GND pins would be a good location.)  Pi CPU activity may be causing supply spikes.  I suspect more bulk capacitance would help (47uF – 100uF), but you would need to experiment with different values.

Regards,

Jeffrey

[tnag]

Hi Jeffrey,

First of all, I would like to thank you and your enginner for the support.
I have done testing the 2 approaches you mentioned.

1) Measure something without IEPE.
The same spikes exist. I have set the IEPE disabled in the python code, and shorted CH0+ and CH0- for the input signal.

2) Add additional capacitance to the 5V net right at the Raspberry Pi.
I still see the same spikes. I tried several configurations below:
C = 47uF, IEPE = disabled, Input = CH0+/- shorted, Power supply = microUSB port
C = 47uF, IEPE = enabled, Input = PCB, Power supply = Direct to RPi GPIO 5V and GND pins
C = 100uF, IEPE = enabled, Input = PCB, Power supply = microUSB port
C = 147uF, IEPE = enabled, Input = PCB, Power supply = microUSB port
C = 194uF, IEPE = enabled, Input = PCB, Power supply = miicroUSB port
C = 194uF, IEPE = enabled, Input = PCB, Power supply = Direct to RPi GPIO 5V and GND pins

I realized that some spikes show the periodic behavior, and found it is about 1 Hz or 1 second cycle.
I have tested with sampling rate 25.6kHz (first fig) and 12.8kHz (second fig), and both show 1 second cycle.
The buffer size is fixed to 100k samles. Does this help you to narrow down the possiblity? 

Regards,

Tatsuro

Edited October 13, 2022 by tnag

[Jeffrey]

The second from the last picture is 'ringing'.  it is most often caused by a bad ground somewhere.

There is something amiss with your Pi Zero2W. You stated previously, you have 2 different models of Raspberry Pi computers, one shows the problem the other does not.  If the same problem was seen on both Pi's I would say the issue is the MCC 172, but since it does not. I have to conclude it is the Pi Zero2W, and  recommend you contact the vendor of the Raspberry Pi Zero2W.

Regards,

Jeffrey

[tnag]

Hi Jeffrey,

Ok, I will contact the vendor.
Meanwhile, I could order another Zero2W last night, I will test to see if my first Zero2W is the bad indivisual.

Thanks for the support.

Regards,

Tatsuro

[tnag]

Hi Jeffrey,

A new Raspberry Pi Zero2W has arrived, and I have done testing.
Unfortunately, the result was the same. There were 1-second periodic spike noises.

Is there any possibility that MCC172 HAT does not work with the RPi ZeroW series?
Have you ever tried to test any ZeroW?

Regards,

Tatsuro

[Jeffrey]

6 hours ago, tnag said:

Is there any possibility that MCC172 HAT does not work with the RPi ZeroW series?

Based on your test results, it appears there are some performance issues with model RPi Zero2W.

As you stated in your original post, you don't see the noise issue when you use the Raspberry Pi 3B when used with this same MCC 172, only with the Pi Zero2W. 

Since the problem is not shown on Raspberry Pi model 3B but only a different Raspberry Pi model, I conclude the problem is not the MCC 172. 

I recommend you contact the makers of the Raspberry Pi Zero2W, and provide them with your test data for both the RPi's 3B and Zero2W

 

 

 

 

 

[tnag]

The results with ZeroW/Zero2W were not as I expected.
However, I can now use 3B with confidence, thanks to your help.

Tatsuro