3
\$\begingroup\$

I'm trying to create completely DIY EMG. As many of you may know, I will be dealing with microvolt inputs. I know that op amps amplify voltage, but I am wondering if I can somehow get a 1,000,000x amplification. Do I need to create my own op amp by soldering a few components or are there already op amps available that can do this?

Note: I am amplifying voltage in order to have the signal be read by an Arduino Uno.

\$\endgroup\$
  • \$\begingroup\$ You mean like an in amp? \$\endgroup\$ – Ignacio Vazquez-Abrams Aug 10 '15 at 0:02
  • 2
    \$\begingroup\$ Modern op amps cannot be created by "soldering a few coomponents". There are many available IC's (especially instrumentation amps as pointed out by Ignacio), that could do this job. You will probably have to do it in several stages as a gain of 1 million is not realistic in one stage. Also, you will have to pay considerable attention to grounding, filtering and shielding if you expect any success. \$\endgroup\$ – Barry Aug 10 '15 at 0:14
  • 1
    \$\begingroup\$ Something like an INA118 feeding to an LMV710 would be a good start. \$\endgroup\$ – whatsisname Aug 10 '15 at 18:28
  • \$\begingroup\$ @whatsisname Thank you for giving me component names as it was hard for me to find some. \$\endgroup\$ – Application Developer Aug 11 '15 at 0:19
  • \$\begingroup\$ My guess is that you need a gain closer to 10^4 than to 10^6 \$\endgroup\$ – Scott Seidman Aug 13 '15 at 19:29
5
\$\begingroup\$

You can definitely do it with op amps. Some things worth investigating

  1. A 'T' feedback network in the op amp
  2. Be advised of the issues with that much gain in an op amp. You will need to account for things like
    • Dynamic range
    • Signal to noise ratio
    • Current/Noise voltage (i.e. you will probably want most of the gain in the first of several stages as the noise voltage will go up with the square of the gain instead of linearly in subsequent stages)
    • The bandwidth of the signal you are capturing

This is do-able, but it will be work. I'd advise looking up Ron Mancini's "Op-amps for everyone", Any book by Gerald Graeme, and a fresh copy of LTSpice to do your analysis first before even breadboarding.

Sounds like a fun project. Good luck!

\$\endgroup\$
  • \$\begingroup\$ Most of the gain can't go on the input stage unless you AC couple the inputs, as there will be a possible electrode offset of about 150mV to deal with (and that's with good electrodes and good contact). About 50 usually leaves enough headroom. AC coupling the inputs causes some other problems, like different impedances on the inputs because of low tolerance components, and even worse CMRR \$\endgroup\$ – Scott Seidman Aug 13 '15 at 17:30
1
\$\begingroup\$

My experience is that doing this in 2 or 3 stages of 1000x or 100x works better. Be sure to consider the voltage offset of the amp as well, since a small offset x 1E6 can easily exceed your output range.

\$\endgroup\$
1
\$\begingroup\$

I disagree with the previous answers. This should be approached by using an INSTRUMENTATION AMPLIFIER input stage with a gain so as not to saturate an approximate 200mV electrode offset that you should expect. After that, you should use an op amp active High-Pass filter with some gain, and low pass filter with some more gain. Keep in mind that your op-amp input offset bias will be multiplied by your gain at EACH STAGE, so watch out for that.

Can you do this with op-amps? Yeah, sure, but you will NEVER achieve a CMRR as high as you will with an instrumentation amplifier IC on the front end. You just can't practically duplicate that with op amps. It is $5 well spent.

\$\endgroup\$
  • \$\begingroup\$ -1 for baseless asertions that do not map to the requirements (CMRR). Hopefully Nick has the answer he needs in any case. \$\endgroup\$ – cowboydan Aug 13 '15 at 17:22
  • \$\begingroup\$ @cowboydan -- have you tried to amplify microvolt-level signals with a 10^6 gain without a high CMRR? The only assumption I'm making is that my the time he's done, he'd like something related to EMG on the output. Can you do this with op-amps alone? Of course, but you have to be much better at it and work much harder than if you're trying to do it with an instrumentation amplifier input stage. \$\endgroup\$ – Scott Seidman Aug 13 '15 at 17:27
  • \$\begingroup\$ yes, I have. You win. \$\endgroup\$ – cowboydan Aug 13 '15 at 17:34
  • \$\begingroup\$ Well, with a 40dB CMRR even 1 mV of common mode noise will saturate a +/0 10V supplied op amp circuit with a gain of 10^6. Why bother when you can get 70-90dB with one chip? The advice in comments to use a INA118 was right. \$\endgroup\$ – Scott Seidman Aug 13 '15 at 17:41

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.