INA333 used in an EMG circuit

Thread Starter

Dritech

Joined Sep 21, 2011
847
Hi all,

I am trying to implement a simple EMG circuit using the INA333 instrumentation amplifier (schematic shown below)
When implementing this circuit, I am getting a output with an offset of approx 4V. Could anyone please explain why I am getting this offset instead of Vs/2 (which should be approx 2.5V) ?

Below are some observations done from the actual circuit implementation and some changes from the schematic below:

1) Resistors R1 and R2 are actually 200 Kohms in the actual circuit, and resistor R3 is 150 ohms
2) VG1 and 2 are connected to the electrodes in the actual circuit and an additional reference electrode is connected directly to Vs/2 supply.
3) Although the reference electrode is connected to Vs/2 supply, I am still getting approx 1.1V as the offset voltage (baseline) at the output of the two detection electrodes. Although I am getting aprox 1.1V at both detection electrodes, if I simulate this on TINA simulator, the output baseline is still Vs/2 (as shown below)

1581783154049.png
1581782559261.png
1581783314650.png
 

Thread Starter

Dritech

Joined Sep 21, 2011
847
@crutschow , no not when they are floating, but when the detection electrodes are connected to the arm. When probing the - and + pins of the IA (marked in red below), I am getting a voltage of approx 1.1V as baseline.

1581785376227.png
 

Thread Starter

Dritech

Joined Sep 21, 2011
847
Is the body connected to a circuit ground electrode?
The electrodes are connected as shown below. The two detection electrodes placed on the muscle (approx 20mm apart) and the reference electrode to the elbow. The reference electrode is connected to Vs/2 as the circuit is powered on a single supply rail (GND and 5V), hence why the reference electrode and the REF pin of the INA333 are both connected to Vs/2.

1581786537858.png
 

Thread Starter

Dritech

Joined Sep 21, 2011
847
What output voltage do you get when you directly connect the electrodes to Vs/2?]

How is Vs/2 generaged?
The reference electrode (Vs/2) is connected directly to a bench power supply set to 2.5V. The same 2.5V is connected to REF pin of the IA. Then there is a 5V supply connected to V+ of the IA.
 
I never DC-couple EMG preamplifiers because the electrodes get salts and sweat (edit: and electrode gel) on them, generating a galvanic half-cell potential. It depends what metals you are using for the electrodes. Try measure the DC potentials between the electrodes.
 
Last edited:

Audioguru again

Joined Oct 21, 2019
851
Your inputs are connected to your negative power supply voltage. But the datasheet says that the inputs do not work properly if their voltage is within 0.1V from a supply voltage.
So add a negative supply so that half the supply and signal ground is 0V or bias up the inputs a little.
 
Last edited:

TeeKay6

Joined Apr 20, 2019
572
Hi all,

I am trying to implement a simple EMG circuit using the INA333 instrumentation amplifier (schematic shown below)
When implementing this circuit, I am getting a output with an offset of approx 4V. Could anyone please explain why I am getting this offset instead of Vs/2 (which should be approx 2.5V) ?

Below are some observations done from the actual circuit implementation and some changes from the schematic below:

1) Resistors R1 and R2 are actually 200 Kohms in the actual circuit, and resistor R3 is 150 ohms
2) VG1 and 2 are connected to the electrodes in the actual circuit and an additional reference electrode is connected directly to Vs/2 supply.
3) Although the reference electrode is connected to Vs/2 supply, I am still getting approx 1.1V as the offset voltage (baseline) at the output of the two detection electrodes. Although I am getting aprox 1.1V at both detection electrodes, if I simulate this on TINA simulator, the output baseline is still Vs/2 (as shown below)

View attachment 199144
View attachment 199143
View attachment 199145
The "Ref" terminal of the INA333 should connect to the victim at whichever point on the victim's body you select as reference. In your schematic, the negative terminals of VG1 & VG2 should connect, via the body reference probe, to the Ref terminal of the INA333--not to the ground you have shown. Ideally, when the INA333 "+", "-", and "Ref" terminals are connected together, the output would be 2.5V--relative to the ground shown. That is an acceptable operating point; the output will go higher or lower than 2.5V depending on the probe signals from the victim. If you measure the output relative to the Ref (i.e.+2.5V point) terminal, the output will swing plus or minus according to the victim's signals.
 

crutschow

Joined Mar 14, 2008
24,306
What output voltage do you get when you directly connect the electrodes to Vs/2?

You didn't answer the above question.

I think your amp connections are all correct but you may be experiencing the galvanic problem that prairiemystic mentioned, so you may have to AC couple the signal.
If you do, be sure to add a high value resistor from each of the inputs to Vs/2 to provide a DC bias point for the inputs..
 

Thread Starter

Dritech

Joined Sep 21, 2011
847
Thanks a lot for your replies.

@prairiemystic , as electrodes, I am using normal electrodes with gel and adhesive.
@Audioguru again , no, the REF pin is actually connected to Vs/2, which is 2.5V
@crutschow ,

"What output voltage do you get when you directly connect the electrodes to Vs/2? "

Oops, misunderstood this question. I will try this and monitor the output voltage.

By AC coupling both inputs, do you mean i should implement the circuit below at both - and + terminals of the IA ?
1581839318165.png
 

crutschow

Joined Mar 14, 2008
24,306
The 1meg resistors will reduce the signal level by a factor of 220/(1220) due to the voltage divider action.
Since that's probably not desired, you may want to increase the value of R4-R5 or reduce the value of R1-R2.
The amp input offset current is 200pA maximum so, the limit on R4 is how much offset voltage you can tolerate.

The value of C1-C2 determines the low frequency response of the circuit which is f = 1/(2pi(R1+R5)C1) where f is the -3dB low-frequency point of the high-pass filter.
Your circuit values give a value of 0.395Hz.
What is the lowest frequency from the muscle?

The addition of C1-C2 should have little effect on the noise.
 
Last edited:

Thread Starter

Dritech

Joined Sep 21, 2011
847
The 1meg resistors will reduce the signal level by a factor of 220/(1220) due to the voltage divider action.
Since that's probably not desired, you may want to increase the value of R4-R5 or reduce the value of R1-R2.
The amp input offset current is 200pA maximum so, the limit on R4 is how much offset voltage you can tolerate.

The value of C1-C2 determines the low frequency response of the circuit which is f = 1/(2pi(R1+R5)C1) where f is the -3dB low-frequency point of the high-pass filter.
Your circuit values give a value of 0.395Hz.
What is the lowest frequency from the muscle?

The addition of C1-C2 should have little effect on the noise.
Thanks or the reply. Your are correct about the voltage divider. If I increase the 1Mohm resistors to lets say something in the range of 6Mohms, will that have negative effect on the circuit (noise or other issues) ? or since it is not connected in series with the input, this will not make much difference? In such case, I will also change the value of the capacitors to approx 80nF.
I used 220K resistors in series as I read that these are necessary to protect the circuit against ESD and to protect the subject (the person) from any electrical shock. Is this correct? will a 220K resistor prevent damage from ESD or prevent an electrical shock? (although ideally this circuit should always be powered by low power batteries to prevent this from happening)


"The amp input offset current is 200pA maximum so, the limit on R4 is how much offset voltage you can tolerate."

Did not quite understand this. How will the resistance of R4 and R5 affect the input offset current and voltage?
 

crutschow

Joined Mar 14, 2008
24,306
If I increase the 1Mohm resistors to lets say something in the range of 6Mohms, will that have negative effect on the circuit (noise or other issues) ?
The thermal noise is determined by the parallel resistance of R1-R5 and R2-R4 so increasing R4-R5 will have only a small effect on that noise.
will a 220K resistor prevent damage from ESD or prevent an electrical shock?
It should.
Even with 240V the fault current would be only 1mA which is well below the lethal value.
How will the resistance of R4 and R5 affect the input offset current and voltage?
It has no effect on offset current, but the offset voltage generated by the offset current equals 200pA * R4 when R4=R5.
Thus for R4=R5=6Mohms, the offset voltage from that would be 200pA*6M = 1.2mV (could be plus or minus polarity).
The output DC voltage from this is then 1.2mV times the amp gain.
 

Thread Starter

Dritech

Joined Sep 21, 2011
847
Hi Again,

So i built the circuit below and it looks like it works fine, except when touching anything which is earthed (example the housing of the bench power supply). This is because the output is saturating to Vs, then goes to 0V when removing my hand, and then back to Vs/2 after a while (please see snapshot of the output waveform below, where A is Vs/2 before touching anything which is earthed, B is the saturation which happens when touching earth, C is the saturation to 0V when removing my hand, D is when it gets back to Vs/2)

NOTE: The 5V and 2.5V rails are both directly powered from a bench-top power supply.

1583007051835.png
1583007505345.png

Following this, I modified the circuit as shown below. The changes are that the 2.5V (Vs/2) is now being generated by a voltage divider, and the same principle is being used for the REFERENCE electrode (where before it was getting 2.5V directly from the bench power supply).

As can be seen from the waveform below, now I am getting the inverse of before (i.e. output saturates to 0V when touching earthed equipment and then goes to 5V when releasing hand)
During this test, I also noticed that when touching the reference electrode (which is being attached to my elbow), I am getting the same symptoms as I was getting with the previous circuit design - please see 2nd waveform below.

Can anyone help me in finding a way to prevent this from happening.

1583007882761.png
1583007849417.png
1583008157817.png
 

eetech00

Joined Jun 8, 2013
1,835
Hi Again,

So i built the circuit below and it looks like it works fine, except when touching anything which is earthed (example the housing of the bench power supply). This is because the output is saturating to Vs, then goes to 0V when removing my hand, and then back to Vs/2 after a while (please see snapshot of the output waveform below, where A is Vs/2 before touching anything which is earthed, B is the saturation which happens when touching earth, C is the saturation to 0V when removing my hand, D is when it gets back to Vs/2)

NOTE: The 5V and 2.5V rails are both directly powered from a bench-top power supply.

View attachment 200279
View attachment 200281

Following this, I modified the circuit as shown below. The changes are that the 2.5V (Vs/2) is now being generated by a voltage divider, and the same principle is being used for the REFERENCE electrode (where before it was getting 2.5V directly from the bench power supply).

As can be seen from the waveform below, now I am getting the inverse of before (i.e. output saturates to 0V when touching earthed equipment and then goes to 5V when releasing hand)
During this test, I also noticed that when touching the reference electrode (which is being attached to my elbow), I am getting the same symptoms as I was getting with the previous circuit design - please see 2nd waveform below.

Can anyone help me in finding a way to prevent this from happening.

View attachment 200283
View attachment 200282
View attachment 200286
hello

There should be a 0.1uf bypass cap connected at each +supply pin to ground. Also from the each resistor divider junction to ground to prevent spurious noise from affecting the circuit. Also...the input bias resistor values seem high...might reduce them to 47k-100k (I'm not sure but seem very high, figure 34 shows 47k each).

eT.
 
Top