Fluctuation with Poti for Offset trimming (Wheatstone bridge)

[DesertEagle]

Hy,

if searched this forum for quite a while but I can't get a clue what is wrong with my circuit. I have following problem:

I have two strain gauges mounted on a beam connected to a soldering pad with two 120Ohm precision resistors. The strain gauges and the resistors are aranged as a Wheathstone bridge and connected to 5V supply, GND and to an Amplifier for the output signal.

At the beginning (beam unloaded) I have quite huge offset voltages of around 3 to 4mV (without the gain). I tried now to put a poti in paralell to the bridge as shown in many application guides to zero the bridge.

Without the Poti, the arrangement deliveres stable signals. As soon as the poti is installed, the signal shown huge fluctuation (after the amplifier).

The poti is installed at the end of the cable connection (apprx. 4m cable inbetween) since the arrangement is not accessible at the point where the bridge is installed (see attachment for details).

Does anybody have an idea where this fluctuation may arise from? Is the Poti creating a loop which causes the fluctuations?

Any help is highly appreciated :thumbsup:

Regards,
Dan

 

[:Dark_Force]

I don't see why there should be an offset trimming function to the circuit shown! Did you forget to draw the connection from the poti wiper to the center of the bridge?

 

[DesertEagle]

I don't see why there should be an offset trimming function to the circuit shown! Did you forget to draw the connection from the poti wiper to the center of the bridge?

You are absolutely right! The one connection of the Poti should be routed to the amp inlet (Paralell to the 120Ohm Resistor)

Missed on that one!

Greetz,
Dan

 

[:Dark_Force]

How many turns does your potentiometer have? I observed a similar behaviour when I used a single-turn ceramic trimmer in a capacitive bridge and solved it by exchanging it with a multi-turn part.
And you should also reduce the range of the potentiometer by choosing a lower value potentiometer and putting resistors in series with the potentiometer. I mean you certainly don't need the whole range of a 1MOhm potentiometer in this curcuit since you have a gain of 1300 following the bridge.

- - - Updated - - -

With such an arrangement, you can adjust for almost +/- 2.5V at the amplifier output with minimum sensitivity to wiper position fluctuations!

 

[DesertEagle]

Hy DarkForce,

thx for your reply. I use a 10T Trimpoti for this application.

One like this with different resistance value:

https://www.ebay.de/itm/Spindeltrim...-Trimmpoti-Poti-64Y-200R-083606-/310494645962

I have no idea from a theoretical point of view why this causes the fluctuations. The only thing I can imagine is that it creates a loop to ground which causes the fluctuations.

Do you think it might be caused due to the Poti is installed at the end of the wires?

Greetz,
Dan

 

[BradtheRad]

Your graph without the pot is perfectly smooth. I wonder if your signal is amplified until it is clipped at the supply rail.

With the pot...
It looks like a normal signal, greatly amplified. The jittery line comes from noise from all sources: two strain gauges, long wire runs, possibly your power supply, etc.

 

[KlausST]

Hi,

if you have an offset of 4mV at input and a gain of 1250 then the output should be 4.9V,
but the AD620 is supplied with 5V and according datasheet it will output max. 3.8V.
So your output signal is clipped. Therfore you don´t see any noise or other fluctuations.

As soon as you compensate the offset then the output voltage is below 3.8V and the AD620 is working as it should.
But now you see all noise and fluctuations.

There is noise caused by every resistor, by every junction, induced by cable, noise in strain gauge supply and others....

If you want precise output then first use a voltage reference, add good low pass filtering to supply your strain gauge. (here you could use a filter with less than 1 Hz, if possible)
Next step is to low pass filter your input to AD620. (select cut off frequency to meet your needs= maximum useful strain gauge signal frequency.)

If you use an ADC, then please use it´s reference voltage for generating strain gauge supply. This further reduces thermal drifts.

Good luck

Klaus

 

[DesertEagle]

Maybe there is some misinterpretation of my graph. (This one was for reference)

I'd like to explain the two situations and parameters in more detail:

Circuit:

• +/- 12V Supply Voltage from a car battery (stable)
• INA125 Instrument-Amp
• Transistors at 5V Precision Reference for current boost -> 5V are absolutely stable
• 120Ohm Precision resistors with very low drift
• 0,5mm2 shielded cables

Situation 1) No Poti

The reading is stable at 2,677V. No fluctuations.

Situation 2) Poti in parallel to the 120Ohm Resistor at the end of the cables

Trimming to zero works but the reading fluctuates between -100mV to +100mV. The fluctuation seems to be "randomly". No special pattern can be identified. (measured with Oszi).

A collegue suggested I might have created a loop to GND with the poti in prallel but I can't see a loop in this configuration.

Greetz,
Dan

 

[:Dark_Force]

Regarding the Datasheet of the INA125 it seems not very realistic to me that this is an oscillation issue. The GBWP is just too low to sustain an oscillation. Can you do an FFT of the output signal when the poti is connected? If it's an oscillation issue you should see a distinct peak.
Maybe you also want to shorten the cable length for a test to see if this changes the output signal (which should be the case if it's an oscillation issue)

 

[DesertEagle]

Hy DarkForce,

at the moment I have no access to the setup but I will try your recommendations as soon as the application is available again.

Unfortunately I cannot use the Ref Input of the INA for zero trimming due to the high offset voltage of 4mV at the initial setting.

Is there a method of trimming the signal before the amp without using a TrimPot?

Greetz,
Dan

 

[:Dark_Force]

Maybe you find some answers in this application note from the brilliant Jim Williams!

http://cds.linear.com/docs/en/application-note/an43f.pdf

 

[DesertEagle]

DarkForce,

thx for the document. Really very informative.

Based on this I prepared a schematic with offset correction. (see Attached)

Any comments are highly welcome!

 

[KlausST]

Hi,

your schematic is hard to read, especially because the INA125 has only text and no symbol.

* C5 has wrong direction!!!
* there is a missing wire from C4 to "CAP+"
* for C4, C5, C6 read datasheet "capacitor selection" a standard elko won´t work good enough
* add ceramic capacitors at the INA125 V- and V+, and at the OPAMPs
* add capacitors after D1 and D2.

* do you really want to compensate for -1V ... 0V ?
* there is no low pass filter in signal lines
* there is no EMV filter in sensor lines
* there is no ESD protection in sensor lines

It seems the circuit with REF200 is from an application note, so i´m sure it will work.
**************

My personal point of view:
This seems to be a very expensive solution.
My hint:
* The INAs have built in reference. Use one 1.24V output and feed it to an unity gain opamp. The resualt is a stable +1.24V output.
* use an inverting OPAMP circuit to generate stable -1.24V
* connect all three offset pots to these +/- 1.25V (pins 1 and 3)
* feed all three pot´s wiper outputs with 10k/100nF low pass filters to individual unity gain OPAMPs
* feed the three OPAMP outputs to the INA125 IAREF inputs.
* i´d use cheaper OPAMPs with: unity gain stability, low voltage noise (not extremely low), low current noise (not extremely low), low offset voltage drift (not extremely low)
*******
* i personally don´t like those very low ohmic 10R pots to adjust signal gain. Try to use fixed resistors. The fine gain adjust can be made with software.

Hope this helps

Klaus

edit: missing wire...

 

[DesertEagle]

Hi,

thanks for your reply and helpful hints.

I have implemented most of them in my schematic. Nevertheless I wonder if I really have to use Tantalium C for C3 and C4 (Symbol name was updated, former C4 and C5). They are really expensive and rather difficult to source. I don't have one of them actually on hand.

With respect to the REF200 solution you are right and for further layouts I will give it a try. I have the Ref200 in my box and I intend to give them a chance in this config.

With respect to Filter, EMV and ESD design:

I am not shure how to design the EMV and ESD filters for this layout. Should I use a simple inductor coil in series?

I designed the low pass filter for 10kHz since I inted to switch the circuit on and off in an 2ms intervall via Microcontroller. Should be possible with 10kHz fc.

Attached the updated circuit. Again, all inputs highly appreciated!View attachment Strain_Gauge_Amp_INA125_with Offset comp_Rev01.pdf

Greetz,
Dan

 

[KlausST]

Hi,

It's ok to work with the REF200.
Tantalum Cs are widely used, they should be available and the price should be low compared to the REF200. The problem with elko is that the standard ones have high ESR at high frequency. Voltage drop, low power efficiency, voltage ripple and heating may occur. If you use them, then put at least 10uF ceramic capacitors in parallel to them.

The low pass filter between pot and opamps are still missing.
You have placed one between Ina output to adc input. This is ok.

Emv...small capacitors, maybe 10nF at the sensor outputs.

Esd .. two diodes on each sensor line to gnd and supply.

Just to clarify:the capacitors at v+ and V- are drawn as elkos. Use ceramic here.

Klaus

 

[DesertEagle]

Hy,

I have now manged to finish the circuit. Due to the additional components it is pretty messed up...

I am now creating the circuit design for my PCB but routing this thing is really challanging...

Thank you for your helpful inputs. I will give some feedback whether this works now.

View attachment Strain_Gauge_Amp_INA125_with Offset comp02.pdf

Greetz,
Dan