Hi all,

I am using the AD8237 with variable output gain and an offset of Vcc/2 (circuit below)
The circuit was implemented on a proto-board and during testing it was noticed that the output gain is not varying when the switch SW1 is rotated.
Could anyone please highlight what is the issue which is causing this to happen.
Any kind of support would be highly appreciated.

 

So the amp is working properly but the gain will not change?
Does the input signal have a reference connection to ground?
What gain does it have?

I see no obvious reason that the gain setting should not work.
Have you continuity checked out the circuit through the switch to make sure it is connected per the schematic?

 

hi D,
I agree, as drawn it should work.
Carry out the suggested checks.
E

 

Hi all,

Thanks for the prompt feedbacks.
Problem solved. For some reason I decided to include two 47nF capacitors in the implementation of the circuit, one from FB to ref and the other from FB to the output of the IA.
Removed these capacitors and problem solved.

 

Now I encountered another problem.

The two waveforms shown below were fed to the inputs of the IA. With a gain of 20, the output should be (10mV*20) + (Vcc/2)
Why is the output being clamped and with no offset of Vcc/2 ?

 

hi D,
What is the value of the Vcc voltage.?
E
EDIT:
Also check the actual Vcc/2 voltage.

 

Hi Eric,

Vcc is 3.3V. Measured Vcc/2 and it is 1.65V

 

I'll ask again.
Do the input signal sources have a path to the circuit ground?
The circuit won't work if the inputs are floating with respect to ground.
There must be a path for the input bias currents.

 

Sorry for not responding to your question. Yes, the inputs have a ground reference.

 

Then it would appear you have a wiring error.

 

Now I encountered another problem.

The two waveforms shown below were fed to the inputs of the IA. With a gain of 20, the output should be (10mV*20) + (Vcc/2)
Why is the output being clamped and with no offset of Vcc/2 ?

View attachment 154080

View attachment 154081

That looks like exactly what would result if one input got the signal you described, while the other was steady 1.65V. It's perfectly amplifying (and clipping) a roughly 0.1V signal about its 1.65V midpoint instead of amplifying the difference between the two signals.

 

hi D,
The input signal in post #5 is not 10mVppk its 100mVppk according to the scope images.
The Vout is limiting at +Vcc and 0V as expected.
E

 

hi D,
The input signal in post #5 is not 10mVppk its 100mVppk according to the scope images.
The Vout is limiting at +Vcc and 0V as expected.
E

I thought that at first too. One signal is 100mVpk and the other is 110mVpk. I think the intent was to amplify the difference between the two, but instead it's amplifying just one or the other.

 

You do need a bias return path. See: http://www.analog.com/media/en/technical-documentation/data-sheets/AD8237.pdf#page=24

Generally, you cannot drive the inputs from a low Z source without a series resistor.

Not sure if your artificial Vcc/2 is low Z for the REST of the circuit to work.

Not sure I folllow the input offset of 1.65 V for the input signal. That's relative to what? So, you not using a sine wave referenced to the artificial ground of Vcc/2 Your artificial 1.65 isn't the same as Vcc/2.

Is there a "ground loop" I don't see.

When I started to play with OP amps, a few things got me.
The bias return path. Even if you have 1 pA, 1 pA across nearly zero ohms is not good. Having an OP amp drive another OP amp input w/o any feedback creates the same mess. A few pA of current with no where to go, so, sometimes series resistors are required.

Understanding common mode range is something else to worry about.

 

Hi all,

Thanks for the replies.

@ericgibbs and @ebeowulf17 , yes the inputs were set to 100mV and 110mV, i.e. a difference of 10mV. The output should be (10mV*20) + the 1.65V offset (is my reasoning right?)

@KeepItSimpleStupid both inputs have the ground connected to the same ground of the instrumentation amplifier. The inputs are driver from a function generator (Analog Discovery 2). Should I include series resistors to both inputs? Won’t that increase the noise?

The REF is connected to a voltage divider followed by a voltage follower, i.e. low impedance, although the datasheet states that the impedance at the REF pin does not matter for this particular instrumentation amplifier.

I will try rewiring the circuit with a different IA and with a fixed gain to simplify the circuit.

 

Hi all,

Thanks for the replies.

@ericgibbs and @ebeowulf17 , yes the inputs were set to 100mV and 110mV, i.e. a difference of 10mV. The output should be (10mV*20) + the 1.65V offset (is my reasoning right?)

@KeepItSimpleStupid both inputs have the ground connected to the same ground of the instrumentation amplifier. The inputs are driver from a function generator (Analog Discovery 2). Should I include series resistors to both inputs? Won’t that increase the noise?

The REF is connected to a voltage divider followed by a voltage follower, i.e. low impedance, although the datasheet states that the impedance at the REF pin does not matter for this particular instrumentation amplifier.

I will try rewiring the circuit with a different IA and with a fixed gain to simplify the circuit.

As I said before, it looks to me like it's amplifying one signal and ignoring the other. Have you tried any variations on what signal you feed it? Here's what I would try if I were in your shoes:

1. Don't change any wiring yet.
2. Only change function generator output signals.
3. Send top signal as is while sending steady 1.65VDC as bottom signal. Observe results.
4. Send 1.65VDC as top signal while sending the original bottom signal. Observe results.

I have a sneaking suspicion that one of the tests above will give the exact results you're getting now, and the other will give something totally different (possibly a flat line.) If my guess is right, you'll be able to tell which signal is being ignored, which should help in tracking down the root cause.

Everyone has their own diagnostic style, but personally I hate the approach where you just take everything apart and rebuild it, hoping it works next time. I much prefer gathering evidence and identifying what went wrong. I feel like I lean much more from my mistakes that way. If I were you, I'd experiment with a variety of input signals to see what effect they have on the output, then go from there.

Unless it's the series resistance issue someone else mentioned, in which case my proposed experiments would probably lead nowhere...

 

Hi all,

Finally found out what was causing this problem. There was a phase shift between the two input singles which was causing the output to be different from what I was expected.
Another quick question. I need to re-amplify the signal following some signal conditioning. Is the circuit below a good prective for obtaining adjustable gain (from 1 to 10), where R1 is a 50k pot and R2 is a 5.5k resistor.

 

Is the circuit below a good prective for obtaining adjustable gain (from 1 to 10), where R1 is a 50k pot and R2 is a 5.5k resistor.

That should work fine.