# AD623 Instrumentation Amplifier gain stuck at 2

Joined Jun 25, 2017
17
Hi,
I'm trying to amplify very small signals in the range of 200uV max. This is something normal using dual-voltage circuit, but I'm trying to build the device with a single uni-polar battery.
I'm using a biasing voltage by dividing the battery voltage into two. After this, an OPA2350 is used as an input buffer. The buffered input is then connected to inputs of AD623. The full schematics is shown below.

Here, you can see that R5 and R6 provide the voltage division, R11 feeds the voltage to the input, and R10 represents the device input resistance. These resistors are all calculated to provide the highest input amplitude. Following this stage are two OPA2350s to buffer positive and negative inputs. The buffered inputs are then connected to AD623 (in the picture it is AD620). Also, the Gain resistor is 1.02K 1% resistor as suggested by the datasheet. R4 and R7 provide an alternative current circuit. Finally, U2 provides the virtual ground, and U6 is integrator feedback to compensate input offsets.

I can't perform tests with actual subjects, but in my lab I have a signal generator that provides sinusoidal wave with a minimum amplitude of 7mV peak. This worked for me before by replacing it with VG1 as the input source.

Now, as I expect, I should get a fixed gain of 100, or a gain of around 50 because of the voltage dividers. However, I can't get any gains more than 2. It doesn't matter if I set the gain resistor for 1000, 100, or even 10. I don't get the amplification I want. What do you think the problem is?

#### ericgibbs

Joined Jan 29, 2010
12,597
hi Hamed,
Welcome to AAC.
Have you tried testing the circuit with a +5V supply, rather that the 3.7V, which is below the rated specification of the AD623.?
With the VG1 source pins shorted together, ie: 0V input, what voltage do you measure between the OPA2350 outputs Pins #1.?
E

Joined Jun 25, 2017
17
hi Hamed,
Welcome to AAC.
Have you tried testing the circuit with a +5V supply, rather that the 3.7V, which is below the rated specification of the AD623.?
With the VG1 source pins shorted together, ie: 0V input, what voltage do you measure between the OPA2350 outputs Pins #1.?
E
You're right. 3.7V is not rated supply in datasheet charts, but it is clearly mentioned there (in input voltage range) that AD623 works with supply range from 3V to 12V. So it should work with 3.7V.

anyway, I did test running the system with 5V, and the result is still the same. Here is what I get on the output:

The peak is around 14mV, which shows amplification of 2. I should mention that grounding is poor is my lab, so the signal does not look like pure sinus. If the grounding is good, I get a very good sinus output.

Also, I tested short circuiting the input pins together. The noise level is around 1.7mV, which is more than OPA2350 and AD623 added together. But I guess that's mostly because of poor grounding and cables that I use for testing. This is output noise level:

I can't get a clear screenshot from OPA2350 pin1, but it's almost the same as the output.

#### Attachments

• 27.9 KB Views: 1

#### ericgibbs

Joined Jan 29, 2010
12,597
hi,
I use the AD623 in a number of projects, I could never get reliable results when the Vs was less than 4.5V.
I see that you have a 2.4V voltage splitter for +/-1.2V for the AD623 Vref [pin 5].

So with no VG1 input [shorted] what is the DC level of the AD623 output pin #6. It should be approx +1.2V
Also the DC levels of theOPA2350's ?
E

#### ericgibbs

Joined Jan 29, 2010
12,597
hi,
A LTSpice simulation shows the circuit is working OK.
How are you connecting the SG to the input.? and is it ground free.?
E

#### Attachments

• 41.4 KB Views: 7
Last edited:

Joined Mar 10, 2018
4,057
The AD623 shows a min single supply rating of 2.7V, page 7 of datasheet.

Curious, why the input buffers ? Versus directly into the AD623 ?

Also whats the time base on scope shots / div ? Scope model number ?

Sig gen, what frequency ? Model number ?

Regards, Dana.

Last edited:

#### ericgibbs

Joined Jan 29, 2010
12,597
Hi Dana,
Thanks for the heads up, just goes to show how easy it is to miss the details.

I also cannot see any reason for the OPA2350's

Eric

Joined Jun 25, 2017
17
hi,
I use the AD623 in a number of projects, I could never get reliable results when the Vs was less than 4.5V.
I see that you have a 2.4V voltage splitter for +/-1.2V for the AD623 Vref [pin 5].

So with no VG1 input [shorted] what is the DC level of the AD623 output pin #6. It should be approx +1.2V
Also the DC levels of theOPA2350's ?
E
About that 2.4V, it is not 2.4V. It's actually V2 4, meaning that the name is V2 and it's 4V. But you're right. I made a mistake in simulation. It is actually the same supply as the whole circuit, 3.7V. Sorry for that.

About the DC levels, for AD623 they are about 1.8V, which is correct. But on input of OPA2350 it's around 0.5 volts, which is again correct based on calculations I did. To be precise, The DC level on pin 1 of OPA2350 is 0.143Vsupply + 0.571VG1 with the resistor choice in the picture.

Joined Jun 25, 2017
17
The AD623 shows a min single supply rating of 2.7V, page 7 of datasheet.

Curious, why the input buffers ? Versus directly into the AD623 ?

Also whats the time base on scope shots / div ? Scope model number ?

Sig gen, what frequency ? Model number ?

Regards, Dana.
You're right about the supply range, but all performance charts are made with either +-5 supply or +5 for single supply. So I'm guessing eric is not wrong, and 3.7 is not nominal voltage supply. On this issue, what do you guys think about AD627? It does not have performance charts with lesser supplies, but the datasheet stated supply range from 2.7V more clearly. Since the pinouts and footprints are the same I can buy and try that too.

About input buffers, The real signal is acquired from animal brain and are really weak. Using an input buffer before InAmp increases the signal integrity and makes it less prone to noise and disturbances. Just something I figured along the way.

I don't use a scope here. I have a USB DAQ (USB-4716) from Advantech, and I watch the signals in a program developed in LabView.

The signal on the picture is 900Hz and It's also a custom made signal generator with 600 ohms output resistance.

#### ericgibbs

Joined Jan 29, 2010
12,597
hi H,
If it wired correctly as your circuit diagram, you should measure the voltages I have marked on this image, in BLUE [ the SG input short circuit]

When you used the SG input signal what was the frequency set too.?
E

NOTE: if the Signal generator is not ground free, it will effect the balance of the input resistor network.

EDIT:

If I ground the SG source, I get approx 2mV output from the AD623.
I suspect the signal generator, which has an Zout of 600R is not 0V/Ground free.
Eric

#### Attachments

• 17.6 KB Views: 8
• 40.8 KB Views: 6
Last edited:

Joined Jun 25, 2017
17
hi H,
If it wired correctly as your circuit diagram, you should measure the voltages I have marked on this image, in BLUE [ the SG input short circuit]

When you used the SG input signal what was the frequency set too.?
E

NOTE: if the Signal generator is not ground free, it will effect the balance of the input resistor network.

EDIT:

If I ground the SG source, I get approx 2mV output from the AD623.
I suspect the signal generator, which has an Zout of 600R is not 0V/Ground free.
Eric

You're right! The problem IS actually because signal generator was grounded. The simulation confirmed that grounding the input results in signal loss.
I ungrounded the signal generator, but I can't get anything if it's not grounded. It's amplified noise and no signal. So I should find a better way to test my circuit. Do you have any suggestions on that?

Also, if I don't use InAmp (a normal OpAmp) the system works fine, regardless of grounding the generator. Of course in that way I don't get differential inputs, and I really want differential inputs.

One more thing, using dual-supply also has no problem using InAmps.

#### ericgibbs

Joined Jan 29, 2010
12,597
hi Hamed,
I would suggest that you use a Audio transformer between the SG output and the Amp input resistor network.
This will isolate the SG ground from the network.

E

EDIT:

This Sim shows a basic audio 1:1 transformer, used for ground isolation.

#### Attachments

• 39.2 KB Views: 5
Last edited:

Joined Mar 10, 2018
4,057
About input buffers, The real signal is acquired from animal brain and are really weak. Using an input buffer before InAmp increases the signal integrity and makes it less prone to noise and disturbances. Just something I figured along the way.
I would think the input buffers in this case -

2) Add noise due to finite PSRR of buffers

Regards, Dana.

Joined Jun 25, 2017
17
hi Hamed,
I would suggest that you use a Audio transformer between the SG output and the Amp input resistor network.
This will isolate the SG ground from the network.

E
Audio transformer? any suggestions on the IC?

Also, I used to ground the animal body when getting test subjects. Do you think that the same applies to that situation or not? The real situation is acquiring brain signals using an electrode, and I grounded the animal body to lower the noise level. If I have to cut the ground from there, I'm pretty sure the noise level would huge.

#### ericgibbs

Joined Jan 29, 2010
12,597
hi H,
This PDF may give you some idea's for the ECG amplifiers.

As the project is battery powered, wouldn't that mean the animals body should be considered the ground/common reference and not the actual Earth ground.?

E

#### Attachments

• 357.6 KB Views: 8

Joined Jun 25, 2017
17
I would think the input buffers in this case -

2) Add noise due to finite PSRR of buffers

Regards, Dana.
You are right about the offset. However, I don't think it's that important since the feedback on REF pin on InAmp suppresses the offset to some extent, and I filter the output signals.

I don't know about PSRR, but my system is battery powered, and I don't think it varies much. What do you think about that?

Joined Jun 25, 2017
17
hi H,
This PDF may give you some idea's for the ECG amplifiers.

As the project is battery powered, wouldn't that mean the animals body should be considered the ground/common reference and not the actual Earth ground.?

E
My application is not ECG, it's actually ECoG, which is far more difficult to deal with. But thanks for the PDF. I'll read it.

And yes, the animal body is actually used as common reference. But it's also a common practice to ground the body to achieve higher SNR.

#### ericgibbs

Joined Jan 29, 2010
12,597
hi H,
Looking thru some information I understand that the ECoG refers to tests on the auditory system.
May I ask where you would place the two electrodes, that are represented by the two contacts of the SG, on the actual body.?
E

Joined Jun 25, 2017
17
hi H,
Looking thru some information I understand that the ECoG refers to tests on the auditory system.
May I ask where you would place the two electrodes, that are represented by the two contacts of the SG, on the actual body.?
E
Not just auditory system. ECoG acquires extracellular signals. They are basically electric pulses travelling between brain neurons. These pulses that are called spikes are in the form of electric current, and to get them you need to actually place a really sharp electrode inside the brain. There are spikes on certain places of brain dedicated to certain tasks. So for example, if you need to monitor auditory response of an animal, you need to monitor spikes on the part of the brain responsible for hearing tasks. The same applies to vision, sense, smell,...

The electrodes placed inside the brain get voltage potential variations compared to another location, which can be another electrode placed close, or simply skull or animal skin.

I hope I explained the problem simple enough. I'm not an expert in biology

#### ericgibbs

Joined Jan 29, 2010
12,597
hi,
Thank you for that brief explanation.
Thinking about how your electrodes [ SG pins ] will be electrically biassed, with reference to the bodies common/ground electrode, due to the input resistor network potential divider, surely they will 'inject' a potential difference between to electrode points and ground?
Will this be a problem.?
E