How can I adjust offset or bias to an amplifier input

[Spike]

Hi all,

As part of a project I'm working on with my son, I lashed together this rectified difference amplifier on a breadboard.  I looked at more traditional circuits but tried this mostly because I found it interesting and it saved me some breadboard space.  The stability has not been an issue for my project as I'm dealing with frequencies from mHz to a few tens of Hz (very low hundreds at most).

The inputs are a pair of TI DRV5053 linear Hall effect sensors.  These sensors output between about 0 and 2V, with a 1V output with no magnetic field.  One problem I'm having is that the 1V output is about +- 20-30mV and so I'm starting out with a a couple tens of milivolts offset between the sensors.

My question is, what is a good/proper way to add a method to adjust/null this offset at the input?  I'm using a single supply so I think I may only be able to offset an input upwards, but that is fine for my application. I experimented with some things I found on p6 of this TI app note, and in this Maxim app note, but I wasn't able to get anything working.

If anyone can provide some gentle guidance it will be much appreciated.

[Ammar]

I would use a better IC. The AD8293 from Analog devices has an inbuilt chopper that would solve this problem.

If you want to stick with your current IC, you could use a potentiometer at the input to create a voltage offset.

[Spike]

I would use a better IC. The AD8293 from Analog devices has an inbuilt chopper that would solve this problem.

To be clear, I'm not talking about the input offset of the amplifier itself, but rather the fact that the two sensors output a slightly different quiescent voltage.

If you want to stick with your current IC, you could use a potentiometer at the input to create a voltage offset.

Yes, great, this is what I'm asking about, but I've tried several things in the application notes that I linked and could not get them to work with this circuit.  I'm obviously missing something in my understanding of how this should work.  So if you could provide some specific ideas to try, perhaps I could improve my understanding.

[RandallMcRee]

The app note that you referenced was talking about nulling the offset of an opamp. Typically this range of adjustment is small.

Thats not what you want.

Put an opamp configured as a summing amplifier in between one of your sensors and the absolute value circuit. The summing amp has two voltage inputs, have one come from one of the sensors and the other feed from a potentiometer set up with a varying dc voltage.

[Spike]

The app note that you referenced was talking about nulling the offset of an opamp. Typically this range of adjustment is small.

Thats not what you want.

Yes, I was a bit contradictory in my previous reply.  I realized at the time that it wasn't quite what I wanted, but I was hopeful that I could apply a similar technique such as raising one side of the input divider above ground.

Put an opamp configured as a summing amplifier in between one of your sensors and the absolute value circuit. The summing amp has two voltage inputs, have one come from one of the sensors and the other feed from a potentiometer set up with a varying dc voltage.

I guess I'll give this a go, but I was hoping there was a solution that wouldn't involve another amplifier (per sensor channel).  One of the reasons I thought I'd give this difference/rectifier a try is that it saved me a pair of op amps (though I guess it didn't  ;-) )

[David Hess]

What power supply voltages are you using?  They are are high enough, then low value resistors in series with the sensor outputs and a potentiometer with the wiper connected to the high supply voltage and the ends connected to the sensor+resistor outputs through larger resistors will allow adjusting the offset differentially.

[StillTrying]

You could pass a small adjustable current through Rc if you're more concerned with the quiescent voltage than linearity at large outputs, you might have to swap the sensors, you'd want the 20mV low one connected to Rc.

[Spike]

What power supply voltages are you using?  They are are high enough, then low value resistors in series with the sensor outputs and a potentiometer with the wiper connected to the high supply voltage and the ends connected to the sensor+resistor outputs through larger resistors will allow adjusting the offset differentially.

Excellent, thank you.  This seems like it might fit my needs.  I'm using a 12V supply for the amplifier, which I'll presume is high enough, but what does high enough mean in this context?

Not sure I completely understand what you mean by "through larger resistor," but I've tried what I think you mean on a breadboard (minus the amplifier) using a 12V supply.  It raised the quiescent voltage to around 1.5 volts per sensor, which would be fine so long as I'm still able to get the full swing from the sensors.  I'm assuming if I mucked with the supply voltage I could alter this?

One additional question: the output of the sensors was already through a 100 Ohm resistor that was part of a low-pass filter.  Would there be a poor interaction between the output filter and this adjustment circuit if I left the capacitors in place?

Thanks very much for this

[David Hess]

What power supply voltages are you using?  They are are high enough, then low value resistors in series with the sensor outputs and a potentiometer with the wiper connected to the high supply voltage and the ends connected to the sensor+resistor outputs through larger resistors will allow adjusting the offset differentially.

Excellent, thank you.  This seems like it might fit my needs.  I'm using a 12V supply for the amplifier, which I'll presume is high enough, but what does high enough mean in this context?

The higher voltage means the resistance is higher so it affects the gain of the sensor less as the adjustment is changed.  Ideally a pair of currents are used but that is more complicated.

The adjustment could also be made in the amplifier but there are some complications because fiddling with the resistances will spoil the common mode rejection.

Not sure I completely understand what you mean by "through larger resistor," but I've tried what I think you mean on a breadboard (minus the amplifier) using a 12V supply.  It raised the quiescent voltage to around 1.5 volts per sensor, which would be fine so long as I'm still able to get the full swing from the sensors.  I'm assuming if I mucked with the supply voltage I could alter this?

One additional question: the output of the sensors was already through a 100 Ohm resistor that was part of a low-pass filter.  Would there be a poor interaction between the output filter and this adjustment circuit if I left the capacitors in place?

If you added the RC circuits shown in the application note, then you are already setup to do this and there is basically no interaction.  With a 12 volt supply and a nominal 1 volt output, 20 millivolts is 200 microamps through the 100 ohms and 200 microamps from 12-1 volts is 55 kilohms.  So use say a 25k trimmer and a pair of 24.9k resistors with one going to each 100 ohm output.  The wiper of the trimmer goes to the positive supply.  None of the values are critical.

The offset adjustment stability depends on the supply stability but that is unlikely to be a problem with hall effect sensors; they have terrible stability anyway.

[Spike]

The higher voltage means the resistance is higher so it affects the gain of the sensor less as the adjustment is changed.  Ideally a pair of currents are used but that is more complicated.

Got it, thanks.

If you added the RC circuits shown in the application note, then you are already setup to do this and there is basically no interaction.  With a 12 volt supply and a nominal 1 volt output, 20 millivolts is 200 microamps through the 100 ohms and 200 microamps from 12-1 volts is 55 kilohms.  So use say a 25k trimmer and a pair of 24.9k resistors with one going to each 100 ohm output.  The wiper of the trimmer goes to the positive supply.  None of the values are critical.

Ah, excellent!  This helps to clear things up for me.  I've implemented this in my circuit and it works beautifully.  Thanks very much for the assistance, it's exactly what I needed.