Sampling frequancy for EMG signal

Other Hardware Sensors
41

you still can not get valid 10 digit results.

Further, while the RESOLUTION of your readings is better,
what is the ACCURACY of the ADC's voltage reference?

Here are the best I could find:

Note the initial accuracy is only 0.2% (max) ; and they aren't cheap!

And what are the characteristics of the ADC? Does it GIVE 10 figure accuracy (even 1LSB accuracy) throughout its range? The ESP32 does NOT.

For 10 figure accuracy you need a 32 bit ADC.

I feel you are misunderstanding something; a 10 BIT adc will not give you 10 DIGIT results.

Do the calculations as INTEGERS. The result from the ADC is NOT A FLOAT.
int millivolt = (sensorValue * 2608)/1024; //5000-2392 = 2608

Print the results in a sensible form - like this:
1012

(thats millivolts.)

42 
void setup() 
{
  Serial.begin(115200);
}
 
void loop() 
{
  float offset = analogRead(25);
  float offsetv=(offset/4096)*3.3;
  float sensorValue = analogRead(26);
  float millivolt = (sensorValue/4096.0)*3.3-offsetv;
  Serial.println(millivolt/201*1000,3);
//Serial.println(offsetv);

}

Here is the output when im using the ESP32

image
image1655×862 21.2 KB

I DONT KNOW if that graph of the emg should be looks like that when im flexing my muscle for 5s
there are spikes in the graph i dont know if that fine or not

43

Do you have a Arduino Uno ?
Connect the muscle sensor to a Arduino Uno, print out the raw ADC signal (from the analogRead function), and use the Serial Plotter of the Arduino IDE. Then you can use the muscles and see what it does. Then you can also try to position the probes/pads better to get a better signal. Do you have different probes/pads to try ?

Can you give a link to your EMG module and can you show a photo of your project ?
It should output a rectified signal.
The Myoware module should work, although it failed when I tried to demonstrate it for my physiotherapist. There are other modules on Aliexpress/Ebay/Amazon that don't work.
Moving an arm or a leg is not enough, you have to use muscle strength to measure something.

Can you tell more about your project ? Please give us a broader view.
Why do you want 1000Hz sampling rate ? A higher sampling rate does not make it better.

44

Im using myoware sensor im collecting the RAW data not the rectified one so i need to sample the signal at 1khz even at 2khz would be better im doing that for machine learning to control the prostatic hand.Here is the link of one of the research which im trying to follow the similar approach

45

Please give a link to your Myoware sensor. I assume that you have the triangle shaped version 2 ? Do you have extra probes ?
Can you try the rectified output, to see if it is at least working a little ?

I did not read the whole article. My experience is that when the location of a probe is moved a little, it might stop working. I used a laptop (without the power supply connected) or a battery, to avoid noise from the mains. As I wrote, even then it failed when I tried to demonstrate it, and that was with the rectified output.

Sparkfun has a Myoware guide: https://learn.sparkfun.com/tutorials/getting-started-with-the-myoware-20-muscle-sensor-ecosystem

46

im using this version 1
Actually, everything was fine when i was using the rectified output which is coming form the SIG pin and the reading is fine .The challenge comes when i use the RAW pin to read the Signal and sample it .At the beginning, i was using Arduino uno to sample data but uno is only 10-bit ADC so i had to use ESP32 for much more resolution and now when i sample data it looks different so i don't know is that right because the sampling change or there is something wrong

47

You have it better working than me because it was never fine when I used the Myoware :face_with_raised_eyebrow: I also have version 1.

Can you do the project with the Arduino Uno ?
The EMG signal is noisy and not accurate anyway, I think that the 10-bit ADC is good enough.

The ESP32 might start Bluetooth or Wifi if that was used in the past on that board, even if the current sketch does not have it. Maybe that causes noise for the sensitive Myoware module.

48

Now im able to get 1 KHZ when i printed 3 figures .In case if im using 2 sensor thus i have to print a total of 6 figures therefore, the frequency will drop again.What to do in this situation

49

Your problem has two sources:
First, you are measuring the raw signal, which has a bandwidth of around 200Hz; so to record its SHAPE you need a LOT more than 2 samples per period.

Second, you are using the serial interface to output results; and its SLOW.
115200 baud is ABOUT 10 characters per millisecond.
So its inappropriate for this application.

What do you hope to do with the data?
If you just want to see what the signal is doing you would do better to use an oscilloscope.
If you want to process and act on the data in some way you will need a more powerful platform than an ESP32.

50

What other platform could be more powerful than ESP32? Im not only observing the behavior of the signal, the purpose of doing that i need to record this signal for signal process and machine learning so i need accurate data where i can capture more character

51
52

I dont have the Raspberry pi 4 i only have 3

53

That depends a lot on sensor and processing. With analog sensing sent via long wires to a 10-bit ADC it really would be terrible - but if the sensor has MCU on board, has decent ADC and is wireless - then it can become quite decent. We developed uMyo to match those criteria (while keeping it at a similar price as Myoware)

54

This topic was automatically closed 180 days after the last reply. New replies are no longer allowed.