Go Down

Topic: Buildng ECG with instrumental amplifier + low pass filter - help! (Read 1 time) previous topic - next topic

lovebes

Please see the image for schematics.



I am trying to build a ECG circuit using INA121 (instrumental amplifier).

It worked for like 5 minutes two weeks ago: https://www.youtube.com/watch?v=wZN98gna-l8&feature=youtu.be

The whole circuit is on a breadboard.

And then I don't think I changed anything but started getting a lot of noise, never getting the signals back.

I am at my wits end. Please help me, please guide me find out what is wrong. Can instrumental amplifiers go bad? If I test it with sin waves, I can see that the signal gets amplified cleanly. The 33Hz low pass filter also works very well, seen with the sin wave.

What I have done to no avail:
- After seeing about 19Hz repetitive noise, I bumped up the low pass filter (LPF) to filter at 10Hz.
- I noticed the noise has the same characteristic even if I don't connect the leads to my chest!
- Leads might be the problem but I even put some saliva on the area where the gel contacts are met with the skin
- I even tried a 33Hz pre-LPF before going into the INA, but that really didn't help. Only after removing it was I able to get the signals seen on Youtube clip.


My next plans are to go for Arduino based LPF based off (http://jeroendoggen.github.io/Arduino-signal-filtering-library/). If that doesn't work, I really have no other clue.

Also, I can try to see if I can eliminate the noise caused by the rails on the breadboard, but I doubt that is the reason. The fact that noise characteristic is same even if leads on/off seems fishy.

Yes, I am well aware that safety is a concern. I am aware of all the risks involved, and that even milliAmps can kill a person. The circuit comes only to at max a couple microAmps at the leads end.


Any advice is undoubtedly appreciated in a wholesome level. My project of piping this signal onto bluetooth into Android is stalled because I can't get the signal..


Thank you so much!

Edit:
Found the problem. I made 47.2nF by parallel-joining 6 capacitors. Removing that and replacing it with a single capacitor (10uF), adjusting the resistors to still make a 33Hz LPF, resolved the issue!!


DVDdoug

I don't know anything about EEGs...

How strong is the 19Hz noise?   19Hz is odd... Normally, you'll get 50/60Hz power line noise and white noise.  

If the 19Hz signal is full-voltage (near the capability of the amplifier) You've (somehow) built an oscillator.  An oscillator needs positive feedback (which can be phase-shifted feedback into the negative input).   But your only feedback is an internal feedback resistor, and there is no phase-shift in the feedback since your filter is passive following the preamp's output.   But, there could be some positive feedback through your breadboard layout (with wires being too close together, etc.).

Quote
Yes, I am well aware that safety is a concern. I am aware of all the risks involved, and that even milliAmps can kill a person. The circuit comes only to at max a couple microAmps at the leads end.
I think you are supposed to be measuring  electrical signals generated  by the brain, rather than pumping milliamps into the victim.    

And, it generally  requires a bit more than a few milliamps to kill someone.   And, because of the body's high resistance, it generally requires high voltage to get lethal current.   From what I've read, less than 50V is considered safe and the user doesn't need to be protected from 50V or less.   At more than 50V, you need to make sure a user can't touch the voltage supply.  

lovebes

Wow that is amazing!
Hm.... a feedback!? Yes, the 19Hz is full-voltage oscillation.
Here's the inner workings of the INA121 Instrumental amplifier:



If a feedback onto a negative input is needed for oscillator, can you guide me to how it would happen on the connections to this chip?
 
Also, sin waves work fine - noise filtered, shown crisp-and-clear amplification on output. When I do that, I put one lead to ground, and the other to tap in the sin waves generated via arduino. Then, I read in the signals generated by output of my schematics by AnalogInput().

Thank you so much for that idea!

Grumpy_Mike

Quote
From what I've read, less than 50V is considered safe and the user doesn't need to be protected from 50V or less.
No not true when you are talking about electrodes connected to a body. The 50V referrers to the "low voltage directive" and it is not considered safe at all.

For electrodes you need at least a 1M input impedance and isolation, normally done by driving the system off batteries.

Go Up