Differential amp voltage ref question

[Annakin4]

Hi All,

I am working on a simple scale using some strain gauges, which is my first time working with differential analog signals. There is a initial amplifier with 1000 gain that takes in the differential signal from the two sides of the wheatstone bridge, and outputs an amplified differential signal into a ADC (ADS8865). Now, both the ADC and the amplifier take in a reference voltage, I didn't think long and hard when making my prototype, and I have a two volt reference (Rest of the system is 3.3V just if curious) being fed into both. Inside the amp datasheet, it states "The REF pin is also available on this device and can be used to provide a DC shift of the output signal" so my first question is, do I even need this pin and a offset? My prototype works while feeding in 2V to REF, and I see a 2V offset on the signal, but I want to do this correctly. As far as the ADC is concerned, it supports "input signals with a differential input swing of –VREF to +VREF. " and I'm assuming the REF is setting the maximum difference it can read between the two differential lines, to support the fact even though there is an offset of +VREF which is the positive limit the ADC can still measure weight I put on the gauge.

Am I correct? Can I ground the amp's ref pin or should I keep a DC offset from the amp?

Thanks,
Samuel

[David Hess]

The amplifier's REF pin can be set to any value which keeps the differential output within the common mode range of the ADC.  Drift of the common mode voltage should have very little effect.  This would not be the case if an ADC with single ended input was used.

Note that if the bridge excitation is tied to the reference voltage for ratiometric operation, then reference voltage drift is removed as an error term.

[Annakin4]

Thanks David,

I apologize I am very tired and struggling to understand, so the common mode voltage range is 0-3.3V (I understand it isn't really rail to rail but for the sake of conversation) and so I want the common mode voltage outputting from the amp to keep the differential signal between 0-3.3V? In this case, would I be correct in saying I should keep the 2V REF going into the ADC, than I should use a voltage divider to put the common mode voltage at say 1.5V?

Also, I plan on exciting the bridge using the reference through a buffer (Currently using the 3.3V supply but too much noise from the switching regulator), and there is a single ended output version of my amp, if I excite with my REF, is there any advantage to using a differential ADC?

I appreciate your time

[David Hess]

I apologize I am very tired and struggling to understand, so the common mode voltage range is 0-3.3V (I understand it isn't really rail to rail but for the sake of conversation) and so I want the common mode voltage outputting from the amp to keep the differential signal between 0-3.3V? In this case, would I be correct in saying I should keep the 2V REF going into the ADC, than I should use a voltage divider to put the common mode voltage at say 1.5V?

With the way these modern delta-sigma instrumentation converters work, the only requirement is that both inputs be within the input range of the converter which will be ground (really common) to the positive supply or something.  Usually that will mean applying the reference voltage divided by 2 to the instrumentation amplifier's REF input and if it drifts around, it should not be a problem because the ADC does not care.

Also, I plan on exciting the bridge using the reference through a buffer (Currently using the 3.3V supply but too much noise from the switching regulator), and there is a single ended output version of my amp, if I excite with my REF, is there any advantage to using a differential ADC?

I am not sure what you are saying here.  Is that two separate issues?

If the ADC's reference is taken from the output of the buffer, then the drift of the buffer is removed as a factor.  (1) When I worked with load cells, we got by with just the output from the 7815 for the analog positive supply which of course also supplied the reference for the ADC so noise and drift from the 7815 was rejected through ratiometric operation.

An ADC with a differential input has a lot of advantages whether the output from the amplifier is single ended or differential.  But a single ended ADC works just fine by tying its "common" directly to the instrumentation amplifier's reference input.

(1) The temperature coefficient of the excitation leads affects the gain and they make six wire load cells to prevent this with Kelvin sensing but usually it is a minor effect.  Or current excitation can be used.  Thermocouple errors can be removed by AC chopping the excitation.

[Annakin4]

Okay, thank you.

I am not sure what you are saying here.  Is that two separate issues?

I apologize for going off on a tangent, to restate, I was considering exciting my gauge with a reference, so if the single ended output was cheaper, and also had an extra circuit for an active shield, would there be any advantage to sticking with a differential amp? (Idk how useful the active shield would be, not really important. I'm having fun here why not play with it.)

Sounds like no, and sticking with the differential amplifier makes sense. EDIT: I meant ADC

Or current excitation can be used.  Thermocouple errors can be removed by AC chopping the excitation.

Ah, that is very interesting, thank you.

[David Hess]

I apologize for going off on a tangent, to restate, I was considering exciting my gauge with a reference, so if the single ended output was cheaper, and also had an extra circuit for an active shield, would there be any advantage to sticking with a differential amp? (Idk how useful the active shield would be, not really important. I'm having fun here why not play with it.)

Sounds like no, and sticking with the differential amplifier makes sense. EDIT: I meant ADC

The differential ADC connection should be more accurate and makes layout easier because the input signal is not shared with the ADC's ground return.  I would use it by preference and it also removes the need for a differential to single ended conversion.

A singled ended signal chain is theoretically lower noise but in practice I doubt there is any advantage.  Other error and noise sources not to mention the basic accuracy of the transducer itself have a greater effect.