EMG Amplifier Circuit - Decoupling Caps

[ANTALIFE]

Hello

I am currently working on a simple EMG Amplifier and am in the process of laying out the PCB. Currently I am using x3 100nF cap's for decoupling, one between power & ground, one between power and ref, and one between ground and ref. I want to keep the board size small (50mm x 30mm) but am finding that the ref to gnd/pwr decoupling cap's are a bit awkward to place.
So the question is are the last two decoupling cap's worth having, keep in mind that the peak frequency of the amplifier is a whopping ~300Hz.

Cheers, Anton

EDIT: Updated schematic with proper input resistors (had none before), and changed the OPAMP to TL064 (from LT324)

[Audioguru]

The inputs of your instrumentation amplifier are floating with no reference or bias voltage. Most ECG circuits have the reference as an inverted common mode signal that is fed to the patient (right leg) to provide a reference voltage near ground and to cancel interference.

[ANTALIFE]

Oh crap totally right, had the referencing resistors in my ltspice simulation forgot to add them to the schematic...

[Audioguru]

The LM324 has crossover distortion that will probably produce spikes on your outputs. The filters do not affect crossover distortion. No other opamp except the sister LM358 dual opamp has crossover distortion.

[ANTALIFE]

The LM324 has crossover distortion that will probably produce spikes on your outputs. The filters do not affect crossover distortion. No other opamp except the sister LM358 dual opamp has crossover distortion.

Thanks for the help, have updated the schematic like you suggested.
Couple more questions though, is the TL064 a better alternative for the OPAMP? Also are the 470k resistors OK for working with EMG signals with the instrumentation amplifier?

[danadak]

Crossover distortion occurs not only in OpAmp output stages but also
in the input stage of RRIO CMOS OpAmps. Many manufacturers do not
discuss this, a few of the good ones show in datasheet this problem.

It occurs fundamentally because of the input stage N and P MOS transistors
not having a symmetrical threshold curves. Many designs do not care about this
because the magnitude of the problem is small. However increasingly in DelSig
A/D design, ~ 16 bits or better, it can be an issue.

As an aside, just a thought, but Cypress has low cost microcontrollers with
DSP, 20 bit A/D onboard, DMA. So doing all the filtering, running in background
with DMA, makes a great way to eliminate passive component issues associated
with OpAmp derived filters.

See attached and https://shreeyashsalunke.wordpress.com/2015/10/31/myoelectric-prosthetic-arm/

http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=6948910&url=http%3A%2F%2Fieeexplore.ieee.org%2Fiel7%2F6937053%2F6948870%2F06948910.pdf%3Farnumber%3D6948910


http://www.thaibme.org/thaibme/2007/papers/23.pdf


Regards, Dana.

[ANTALIFE]

Crossover distortion occurs not only in OpAmp output stages but also
in the input stage of RRIO CMOS OpAmps. Many manufacturers do not
discuss this, a few of the good ones show in datasheet this problem.

It occurs fundamentally because of the input stage N and P MOS transistors
not having a symmetrical threshold curves. Many designs do not care about this
because the magnitude of the problem is small. However increasingly in DelSig
A/D design, ~ 16 bits or better, it can be an issue.

As an aside, just a thought, but Cypress has low cost microcontrollers with
DSP, 20 bit A/D onboard, DMA. So doing all the filtering, running in background
with DMA, makes a great way to eliminate passive component issues associated
with OpAmp derived filters.


Regards, Dana.

Righto, come to think of it's I will just make sure that my ref voltage is such that I will stay above the crossover point (0V) at all times 

[StillTrying]

Righto, come to think of it's I will just make sure that my ref voltage is such that I will stay above the crossover point (0V) at all times 

LOL. Opamps don't know about 0V / Gnd they don't have a Gnd pin. The inputs' crossover point is usually towards the V+ end of their CM.
The TL064 will be fine as long as you keep it's inputs and outputs at least 1V away from the V+ and V- rails.

[Audioguru]

Your schematic shows no power supply voltages. The minimum recommended for the TL064 is +-5V. The TL064 is one of the noisiest (hiss and rumble) quad opamps made. Maybe you should use a low noise quad opamp made for audio (TL074).

[ANTALIFE]

Righto, come to think of it's I will just make sure that my ref voltage is such that I will stay above the crossover point (0V) at all times 

LOL. Opamps don't know about 0V / Gnd they don't have a Gnd pin. The inputs' crossover point is usually towards the V+ end of their CM.
The TL064 will be fine as long as you keep it's inputs and outputs at least 1V away from the V+ and V- rails.

Wait aren't you talking about saturation? I'm pretty sure crossover occurs when you switch between the top and bottom BJT/FET (see video), so the solution is to add an offset to your input signal such that you only operate with either the top or bottom switch (but not both).

[ANTALIFE]

Your schematic shows no power supply voltages. The minimum recommended for the TL064 is +-5V. The TL064 is one of the noisiest (hiss and rumble) quad opamps made. Maybe you should use a low noise quad opamp made for audio (TL074).

Cheers for the suggestion, will have a look at TL074 after work.

[StillTrying]

Wait aren't you talking about saturation? I'm pretty sure crossover occurs when you switch between the top and bottom BJT/FET (see video), so the solution is to add an offset to your input signal such that you only operate with either the top or bottom switch (but not both).

The inputs of a rail to rail op amp can have it as well.
https://www.maximintegrated.com/en/app-notes/index.mvp/id/656

But with +- rails and not using LM324s you won't have that problem on inputs or outputs now.

[ANTALIFE]

Wait aren't you talking about saturation? I'm pretty sure crossover occurs when you switch between the top and bottom BJT/FET (see video), so the solution is to add an offset to your input signal such that you only operate with either the top or bottom switch (but not both).

The inputs of a rail to rail op amp can have it as well.
https://www.maximintegrated.com/en/app-notes/index.mvp/id/656

But with +- rails and not using LM324s you won't have that problem on inputs or outputs now.

Ah, well the more you know. Thanks for the link will have a good read.