Excitation of Wheat-Stone Bridge Measurement With Multi-channel ADC

Thread Starter

trodav

Joined Jan 25, 2021
5
Hello,

I'm currently working on a project, where I want to measure force using strain gauges. The strain gauges are configured in four point Wheatstone bridges. I'm using a four channel ADC. On each channel I feed in the amplified differential output voltage of a Wheatstone bridge. My question is how to properly excite the 4 Wheatstone bridges with very little error?

Here is more detailed version of my problem and my approaches regarding excitation the so far:

Typically one would use some linear regulator to power the bridge (see example). Additionally you would connect sense wires directly to the strain gauges and connect them to REF+, REF- of your ADC. Firstly this removes offsets due to voltage drop caused by the wire resistance. This can be really important, e.g. 5 V excitation voltage leads to 41 mA for 120 Ohm stain gauge. Secondly any error or ripple would be common to exication voltage and ADC reference.

Using 6-wire measurement like this is not possible for my case because I have 4 Wheatstone bridges and I could just connect Vref to a single bridge. Apart from that my ADC only has a single (Vref+) terminal.

At first I was thinking about current excitation, since one can easily build a high quality current source using a voltage reference, an op-amp, a series pass transistor and a shunt resistor. With current excitation, I would need to put all my Wheatstone bridges in series to ensure they all have the same current. I could then feed the voltage over the shunt resistor to the opamp, since it would be proportional to the exicitation current ripple. If it helps I can also post a picure of this.

One of my requirements is that my design should work for 120 Ohm, 350 Ohm, 1k Ohm strain gauges. Ofcourse sensitivity and also the total error of measurement is better with a higher exiciation current/voltage. But using 120 Ohm gauge would require 41 mA current source for good total error. On the other hand a fixed current source of 41 mA would mean a voltage drop of 166 V over the 4 Wheatstone bridges with R_nom=1k Ohm, which is obviously far out of the range of the rails I'm planning to use.

Furthermore I was thinking about using current exication but with all 4 Wheatstone bridges in parallel, in combination with current mirrors. I heard that current mirrors often have high error (even with matched pair transistors), so I'm not sure if that would be a good approach. A second option would be to use a high quality voltage reference for the ADC, use voltage exciation with the Wheatstone bridges in parallel and just use low resistance wire maybe even offset calibrated. But I'm not really happy with option 2 since it does not compensate for aging, temperature drift and you would need to calibrate every time you switch the wiring.

Finally I should also mention that an ac exciation is not really an option, since my application also requires higher kHz speed signals.

Are there any suggestions, how I could find a neat sollution to this problem? Thanks alot for reading this long question! I'm happy for any help :)

Greetings
Dave
 

Thread Starter

trodav

Joined Jan 25, 2021
5
Hello,

I came up with an easier solution, I think. I'm planning to use all Wheatstone bridges in parallel powered from a linear regulator. One can measure the actual bridge voltage directly at the strain gauges for offset correction. Using some sort of multiplexer I can measure the actual bridge exicitation voltage on every channel with my ADC. Because of its high input impedance the on-resistance of the additional multiplexer should not be a problem for the measurement with the ADC.
The actual bridge exiciation voltage is still proportional to the output voltage of the linear regulator and the differential bridge output voltage is still proportional to the bridge exiciation voltage. Therefore I can use the output of the linear regulator as the ADCs reference voltage. So error (ripple) on the output of the linear regulator will again be common to the ADC and the Wheatstone bridge. To use the output of the linear regulator as voltage reference, I will probably need to reduce it using a resistive voltage divider.

Any opinions or comments on this are very welcome.

Greetings

Dave
 

ericgibbs

Joined Jan 29, 2010
18,766
hi Dave,
What is the actual voltage of the bridge excitation voltage.?
E

BTW: Are you planning to use the AD7191 ADC module.?
 

Thread Starter

trodav

Joined Jan 25, 2021
5
hi Dave,
What is the actual voltage of the bridge excitation voltage.?
E

BTW: Are you planning to use the AD7191 ADC module.?
Well, I should have expressed myself more clearly. Using voltage excitation there will be a voltage drop on the wire resistance that connects the output of the linear regulator to the Wheatstone bridge, same for the return path. So if I use a 5V linear regulator the voltage seen at bridge terminals might be reduced depending on the current draw. This is meant by "actual voltage of the bridge". So in my example the actual voltage of the bridge might just be 4.90 V or something. Which would result in an error on the differential bridge output voltage.

The AD7191 ADC module looks interesting, but it's not suitable for my application, since I require faster sampling rate. But I already picked an adc and tested it.
 

ericgibbs

Joined Jan 29, 2010
18,766
hi trodav,
I appreciate what you mean by voltage supply not being exactly at 5.0v at all times. ;)

If you use a ADC that is ratiometric and the bridge excitation voltage, via remote bridge sensing wires, as the ADC reference voltage, the ADC output should track voltage changes.

BTW: I guess you know that most MCU internal ADC's are ratiometric, if the Vsupply to thr MCU is used a Vref for the ADC.

E

Posting a rough sketch of the project would help our discussion.
 

Thread Starter

trodav

Joined Jan 25, 2021
5
hi trodav,
I appreciate what you mean by voltage supply not being exactly at 5.0v at all times. ;)

If you use a ADC that is ratiometric and the bridge excitation voltage, via remote bridge sensing wires, as the ADC reference voltage, the ADC output should track voltage changes.

BTW: I guess you know that most MCU internal ADC's are ratiometric, if the Vsupply to thr MCU is used a Vref for the ADC.

E

Posting a rough sketch of the project would help our discussion.
Hi ericgibbs,

what do you mean by the ADC being ratiometric? As far as I know ratiometric measurement is a type of configuration of the ADC where the reference voltage is derived from the exicitation voltage of the sensors (see). So its rather a design technique for a circuit than a property of the used ADC. As I explained in the post at the top I can not use a ratiometric measurement. That would require to connect the sensing wires of the Wheatstone bridge to the Ref terminal of the ADC, but I only have one Ref terminal yet multiple Wheatstone bridges. Correct me if I'm wrong.
 

Thread Starter

trodav

Joined Jan 25, 2021
5
hi Dave,
Do you have a draft sketch you could post.?
E
I find that the INA125 is useful device in L/C projects.
Sure. Attached is the schematic that shows how I'm going to excite the Wheatstone bridges and connect them and the reference voltage to the ADC.
 

Attachments

Top