Potentiometer configuration in the Wheatstone Bridge

Thread Starter

beedees

Joined Mar 29, 2015
37
Hi to all! I have some doubts about the use of a digital potentiometer in a wheatstone bridge. I have to realize a conditioning circuit to amplify the signal of a strain gauge so I want to set to zero the output of wheatstone bridge using a digital potentiometer by SPI (Serial Peripheral Interface) with a microcontroller. I know that the POT can be put in a branch of the the bridge or in parallel as you can see in the attached schematics.

Schermata del 2015-03-29 20:41:37.png

Schermata del 2015-03-29 20:40:40.png

But I don't know what are the advantages and disadvantages by using one of these configurations. Consider that one of resistors of the right branch (R3 or R4) or both are strain gauges.

Thank you and sorry for my bad English!
 

OBW0549

Joined Mar 2, 2015
3,566
Your second diagram (the parallel approach) is the better arrangement, as it will minimize temperature shifts: digital pots usually have rather large temperature coefficients of resistance, and your second diagram comes closer to operating the pot in a ratiometric manner that minimizes the effect of the pot's resistance tempco. However, I would recommend putting a resistor between the wiper of the digital pot and the junction of R1 and R2, with the value of that resistor chosen so that you get all of the adjustment range you need, but no more; this will allow your adjustment steps to be as fine as possible. The way you have shown, your adjustment steps will be coarser and a large portion of the pot's adjustment range will be un-usable.

Your English is fine, by the way.
 

crutschow

Joined Mar 14, 2008
34,285
You could also put Rv in series with (and between) R1 and R2 and use the wiper output for that half of the bridge.
As OBW0549 noted, you should select the values of R1 and R2 relative to the value of Rv so that you can use the full range of the pot to give the adjustment range you need but no more.
 

MikeML

Joined Oct 2, 2009
5,444
Here is how I would do it. This way the digital pot is operated near 0V, and any mismatch in thermal tracking between R1-4 and the digital pot will be minimized with this connection. Make the pot resistance a small fraction of R3-4.

br.gif
 
Last edited:

crutschow

Joined Mar 14, 2008
34,285
I realize my suggestion won't work with most digital pots since they require one terminal be at ground potential.
Mike's connection would appear to be the best if both end terminals of the pot can float some above ground.
 

OBW0549

Joined Mar 2, 2015
3,566
I realize my suggestion won't work with most digital pots since they require one terminal be at ground potential.
Mike's connection would appear to be the best if both end terminals of the pot can float some above ground.
Which digipots require one end to be at ground? I don't know of any, and the ones I've used such as the ones made by Microchip, do not.
 

crutschow

Joined Mar 14, 2008
34,285
Which digipots require one end to be at ground? I don't know of any, and the ones I've used such as the ones made by Microchip, do not.
Perhaps it's that the terminals voltages must be between ground and the supply voltage, which is certainly less restrictive.
 

Thread Starter

beedees

Joined Mar 29, 2015
37
Thanks to all. I didn't consider that the digipot and a completion resistor have different tempco and this fact could unbalance the bridge. But I don't understand why this way to balance the bridge will be coarse. Maybe because the potentiometer is digital and some values will not usable? Need I a digipot with more bits?
 

OBW0549

Joined Mar 2, 2015
3,566
But I don't understand why this way to balance the bridge will be coarse. Maybe because the potentiometer is digital and some values will not usable? Need I a digipot with more bits?
To see what I meant, consider the case in which the pot wiper approaches either end of the pot: as the wiper moves away from the middle, its effect on the bridge becomes greater and greater, in a nonlinear fashion. At the extreme ends the steps become huge, and the R1/R2 side of the bridge is ultimately forced to either ground or Vcc regardless of R1 & R2. Inserting a resistor between the wiper and the R1/R2 junction reduces the step size, makes the step sizes even (or more even) across the adjustment range, and also further reduces the effects of digital pot resistance tempco.
 

OBW0549

Joined Mar 2, 2015
3,566
It just occurred to me that since you're thinking about using a digital pot in this application, you've doubtless got a microcontroller operating it. Am I right?

If so, that microcontroller probably has one or more PWM (pulse width modulation) outputs, and you could spare yourself the expense of a digital pot by using PWM to perform the adjustment. Simply take the uC's PWM output, pass it through one or more stages of RC low-pass filtering, and connect the filtered output to the R1/R2 junction through a resistor, just as you would have connected the digital pot wiper-- and voila, you're done.

An added advantage of this approach is that whereas most digital pots have only 64, 128 or 256 positions, microcontrollers can often generate PWM outputs with much more precision, sometimes up to 16 bits. That will allow you to make much finer bridge balance adjustments.

Don't know why that didn't dawn on me before... old age, I guess.
 

Thread Starter

beedees

Joined Mar 29, 2015
37
Wow:) It's a good idea! In practice I control the balance of the bridge selecting a voltage value with the PWM. Am I right? And I have to realize a RC low pass filter to dampen the generated pulses, I think... For this reason do I have to use LPF?
So the schematic is this

Schermata del 2015-03-30 18:39:22.png

Once again thank you very much!
 

OBW0549

Joined Mar 2, 2015
3,566
In practice I control the balance of the bridge selecting a voltage value with the PWM. Am I right?
Yes, that's it exactly.

And I have to realize a RC low pass filter to dampen the generated pulses, I think... For this reason do I have to use LPF?
Yes. You want the output of the filter to be the average DC value of the PWM waveform, with pulsations/ripple removed.

So the schematic is this
That's correct. However, I would add a second RC stage like R6/C1, after the first stage to get more ripple attenuation. Or you could use one filter stage as you've shown and simply make C1 very large to get more attenuation, since (I assume) you probably don't need very rapid response to changes in PWM setting.
 

Thread Starter

beedees

Joined Mar 29, 2015
37
Ok! Thank you so much! I've learned a lot. I have only to understand if use a bridge completion module for strain gauge or create a wheatstone bridge. In fact strain gauges produce a low level signal so I have to pay attention to select resistances with small tolerance and temperature coefficient, but other problems could be wire resistance or noise pickup. After that I apply your advices! Thank you a lot!
 

crutschow

Joined Mar 14, 2008
34,285
You could use a feedback approach to adjust the bridge balance.
When commanded (when there's no load on the strain gauge) the micro could adjust the PWM duty-cycle until the bridge output is (near) zero.
 

cmartinez

Joined Jan 17, 2007
8,220
Yes, that's it exactly.


Yes. You want the output of the filter to be the average DC value of the PWM waveform, with pulsations/ripple removed.


That's correct. However, I would add a second RC stage like R6/C1, after the first stage to get more ripple attenuation. Or you could use one filter stage as you've shown and simply make C1 very large to get more attenuation, since (I assume) you probably don't need very rapid response to changes in PWM setting.
This is very interesting... I've been working with load cells for a while now and this is the first time I've seen this configuration. Normally I'd just compensate what's being read from the Wheatstone bridge by adding or subtracting an offset.
In this case, wouldn't it be better to use PWM on both sides of the bridge to do this compensation? That is, add a second RC stage with PWM being fed at the node where R2 and R4 meet?
Also, how does one avoid output signal drift with this circuit?
 

MrChips

Joined Oct 2, 2009
30,714
Now there is an interesting thought. Use the PWM scheme to null the bridge and use the PWM value as your measured value. Mission accomplished.
 

OBW0549

Joined Mar 2, 2015
3,566
This is very interesting... I've been working with load cells for a while now and this is the first time I've seen this configuration. Normally I'd just compensate what's being read from the Wheatstone bridge by adding or subtracting an offset.
Usually you can do that. But when working with raw strain gages, you can get some pretty wicked bridge offsets if the gage bonding is not done right. Years ago, I developed this product, here:

http://crlsensors.com/signal-conditioning-instrumentation.cfm?product=5804

in response to a customer who, for various reasons, was experiencing initial bridge offsets of as much as +/- 3000 microstrains. I used the PWM technique that I described above in #14 to zero out that strain offset, along with the feedback approach that crutschow described in #16.

In this case, wouldn't it be better to use PWM on both sides of the bridge to do this compensation? That is, add a second RC stage with PWM being fed at the node where R2 and R4 meet?
I don't think that would make any difference, one way or the other.

Also, how does one avoid output signal drift with this circuit?
The PWM circuit doesn't have much effect on the temperature drift of the bridge circuit. But you can use the PWM to compensate for any temperature shifts in the bridge itself, by making the PWM value dependent on measured temperature. I did that on another design, to essentially remove the temperature drift from a Wheatstone bridge pressure sensor. It's a very versatile technique.[/QUOTE]
 

cmartinez

Joined Jan 17, 2007
8,220
Now there is an interesting thought. Use the PWM scheme to null the bridge and use the PWM value as your measured value. Mission accomplished.
Ahhhh... let me add my two cents then...
If you have enough resolution on the PWM signal, then simply use a comparator to detect when the signal has been completely nulled... you could then infer the load value and report it... no ADC needed!
 
Top