Amplify 0.1-1mV from a bridge for Arduino use

[alec_t]

What is the minimum/maximum V the AD623 can handle on the inputs?

The datasheet tells you that.

I did see some designs that use a ref voltage IC (i.e.: https://www.ti.com/product/ref1004-2.5 ) - is that something I should try?

Unless you need to offset your output just use GND (as Eric suggested).

I am not too concerned by the 50Hz interference. Will be hard to avoid on a breadboard design I think.

If I understand the datasheet correctly any interference on the inputs can get rectified and appear as an output error. That may be why your output is stuck at -2.7V.

 

[Pcba]

I got a new AD623 yesterday and just did a test. I used a voltage divider made of 1M5 R and a 10K pot. When I put 1mV (rather noisy) on 2 + 3 I get 600mV output with a 120R as Rg. That is about what I expect.

So it was a defect AD623 in the end.

A 2nd test with my probes - and I do get the expected results!

The datasheet tells you that.

Honestly, quite a few of the figues I don't really understand, such as max high and low. Also, what is the max output? When I put 10mV to the input I obviously can't get 6.8V at the output with a 5V supply.

One important questions I got - can direct 5V or GND to pin 2 or 3 harm the IC? Since I work with probes this can happen. The datasheet said "save up to supply +0.3" - I can't get there.

 

[ericgibbs]

I got a new AD623 yesterday and just did a test. I used a voltage divider made of 1M5 R and a 10K pot. When I put 1mV (rather noisy) on 2 + 3 I get 600mV output with a 120R as Rg. That is about what I expect.

So it was a defect AD623 in the end.

A 2nd test with my probes - and I do get the expected results!



Honestly, quite a few of the figues I don't really understand, such as max high and low. Also, what is the max output? When I put 10mV to the input I obviously can't get 6.8V at the output with a 5V supply.

One important questions I got - can direct 5V or GND to pin 2 or 3 harm the IC? Since I work with probes this can happen. The datasheet said "save up to supply +0.3" - I can't get there.

Hi,
The AD623 is OK to operate upto +/-03V of the supply rails.

I would add a 4k7 resistor to each probe, it will not affect the readings , but it will protect the AD623 if an accidental over voltage is applied to the probes.

It also helps to protect the IA inputs against electrostatic damage.

The Vout can swing to within 0.5V of the supply voltage, I would normally allow 1V overhead.

E

 

[Pcba]

Hi,
The AD623 is OK to operate upto +/-03V of the supply rails.

I would add a 4k7 resistor to each probe, it will not affect the readings , but it will protect the AD623 if an accidental over voltage is applied to the probes.

It also helps to protect the IA inputs against electrostatic damage.

The Vout can swing to within 0.5V of the supply voltage, I would normally allow 1V overhead.

E

Thanks for all the help. To see it working gave me a big boost. Next step will be the finer details over the next few days. My testing today worked really great with normal thickness copper foils - but the output for 5oz was only just a little above noise. More tests are in the pipeline.....

 

[Pcba]

Everything is fine and my circuit works!

Just one little technical issue: when there is no input on 2 and 3 I get about 33mV on the output. Ref pin 5 is on GND. That is fine for most copper tests, but when the ristance gets very low then my OUT is near the noise (49mV)

Is there a way to get it to zero with a single power supply (I use 5V).

 

[alec_t]

Is there a way to get it to zero

Is the noise AC or DC? If AC, then minimising wire lengths and screening the whole system would help but will never quite get it to zero. If DC, then adding a DC offset to the input could get it to zero (but it would likely be subject to drift with temperature). Or could you use a 33-49mV reference at pin 5 instead of Gnd?

 

[Pcba]

Is the noise AC or DC? If AC, then minimising wire lengths and screening the whole system would help but will never quite get it to zero. If DC, then adding a DC offset to the input could get it to zero (but it would likely be subject to drift with temperature). Or could you use a 33-49mV reference at pin 5 instead of Gnd?

The noise I can handle when I move a bit away from my power supply. But even with no input (2 + 3 totally open) I get about 33mV.

I start to get it now - I could setup a voltage divider with 33mV and use that as REF, right? That will not be very stable but maybe ok for now. I tested the AD623 before wit a voltage divider is in that range already, I will reuse that. (go for a jog now and test it right when I am back)

 

[alec_t]

I could setup a voltage divider with 33mV and use that as REF, right? That will not be very stable but maybe ok for now.

Right. If derived from the 5V supply (which I assume is regulated ?) it will be as stable as the supply.

 

[ericgibbs]

The noise I can handle when I move a bit away from my power supply. But even with no input (2 + 3 totally open) I get about 33mV.

I start to get it now - I could setup a voltage divider with 33mV and use that as REF, right? That will not be very stable but maybe ok for now. I tested the AD623 before wit a voltage divider is in that range already, I will reuse that. (go for a jog now and test it right when I am back)

Hi,
If you have added the 4k7 resistors I suggested, you could add a 0.01uF or 0.1uF good quality low leakage capacitor across the IA input pins, this will reduce the noise.

You should not use a resistive divider to generate the Voffset, as it will degrade the CMRR .

Use a very low impedance source for the Vref, say an OPA out.

E

 

[alec_t]

Use a very low impedance source for the Vref, say an OPA out.

@Eric
How low is possible with an opamp at 33mV? Not all pull down well.

 

[ericgibbs]

@Eric
How low is possible with an opamp at 33mV? Not all pull down well.

hi alec,

It depends upon the OP's required accuracy.

IMO trying to remove a 33mV noise signal by applying a 33mV DC offset will only realise a 33mV DC offset with the 33mV noise signal superimposed on the top of the Voffset.

The REF pin 33mV voltage could be achieved by using a OPA powered by his existing +/-4.5V and using a suitable resistive divider on the input of the OPA.

He should attempt to reduce the noise being fed into the AD623 inputs by using R-C filters as I have suggested.
I often use the AD623 on weighing bridge projects and its the attention to the AD623 noise filtering that gives a good result.

The OP is only looking to measure input signals over a 100uV to 1mV range, so its important to minimise the input noise, Offsetting will not help when he tries to A2D the signal.

E

 

[alec_t]

IMO trying to remove a 33mV noise signal by applying a 33mV DC offset will only realise a 33mV DC offset with the 33mV noise signal superimposed on the top of the Voffset.

Very likely

 

[Pcba]

I don't worry about the noise too much. I get strong 50Hz spikes with wires and no signal attached to 2 + 3, but when I short 2 + 3, or remove any wires, I get quite stable 33mV DC. The noise in that signal is maybe 2mV. During actual measurement there are no 50Hz spikes and I get a stable and expected output.

Since the 33mV output is +33mV and pin 5 is to GND I can not get lower then 33mV. Tried caps too, no change.

I will try +/- 2.5V after breakfast.

My power supply is a cheap China one, Atten brand, dual 30V. I am thinking of changing that, or build one from a circuit design. Elektor had one a few years back, Unilab. Not sure if that's really better though.

Update:
When I split the rail into +2.5/-2.5V then I get <1mV output shorting 2 + 3 (through 1 pcs. each 4K7 ) when I put REF on GND. I can not wish for more (for now). I can measure now even very low resistance. Lowest sample I got on hand is 210µm/6oz copper foil (OUT 14mV - IN 17µV) - no 400µm/10oz sample on hand unfortunately.

 

[alec_t]

Good to hear you've solved the problem.