ad620 / ad8221

When I connect my oscilloscope probe GND pin to AGND problem %95 gone. I am powering this cuircuit with a DC/DC converter. I do not make any connection to earth. sometimes in industrial environements earth connection can be noisy. thats why I do not make any earth connection on amplifier circuits.

Am I doing something wrong?

Hello dcIm,

dclm said:

Am I doing something wrong?

that's at least possible. Let me ask some questions which may help to clarify.

1) You say, that the output voltage shifts when you touch the IC. How far does it shift? Is it just some mV or is it several V? You have a very high gain. Touching the IC will e..g. change its temperature, and even a very small temperature drift of the input offset will result in mV at the amplifiers output.

2) Is the load cell connected when you observe the input drift, or are the inputs open? Open inputs will give you an undefined voltage at the output, it would be normal to observe drifts under that conditions.

3) In case the load cell is connected: does it share the same GND as your amplifier? If this is not the case, then the input current return path is missing. This would again lead to undefined behaviour of the output and drifts would be normal.

4) In case the load cell is connected and shares the GND of your amplifier: do all amplifier work equally well? Do they react on signals of the load cell and do they deliver correct output? How big is the drift in comparison with this output signal?

5) In case the load cell is connected and shares the GND of your amplifierl and the amplifiers deliver correct output (besides the shif): are all signals inside the allowed voltage range of your amplifier? (See e.g. fig. 8 in the datasheet of AD8221)

I hope the answers to these questions may help to find the root cause of your problem.

best regards

Achim

Hello Achim,

It shift in mV range. It gradually shifts. The output shifts even if I do not touch the IC. Also when I move around resistors at the input output again shifts. Not only ad8221. time to time I test amplifier by applying some hot air to amplifier. Compared to boards that works good. Temperture drift is huge.

Loadcell is connected and they share same ground.

Amplifiers work equally well. It changes according to datasheet. If I remove tare and put 500GR weight it shows 500.

I am applying 10V to a 2mV/V loadcell. AD8221 gain is 660. I further apply analog filtering with some gain. This signal is fed into a PLC which has +/-10V input. So normally my signal is around -8V. And when I put load my output goes upto 9.5 ( I use OP177 for filtering ) My weighing range is around 900GR. So additional 12V/V gain at filter section.

Each GR in display approx. corresponds to 18mV. When I touch IC weight changes approx. 3-4GR and stays like that. when I remove my hand output shows -1GR. But sometimes output shifts non-stop until I see no weight. that means near 0.5V output change on ad8221. But change in weight gradually increases.

My supply voltage is +10V and -10.5V and loadcell common mode output voltage is 5V. I use reference pin to change output of ad8221.

About earth connection... Why do you think connecting AGND and earth on board has this kind of effect. 

Thanks for the description you gave. Let's ensure, that I understand it correctly:

regarding the correct voltage range: Your output voltage range is -8V to +9.5V. But that is probably after the additional filter/gain stage. So at the output of AD8221 the voltage range would be -667mV to 792mV, correct? That should be well in the allowed range. As you shift the output voltage by driving the reference pin: you are driving it with a low impedance source which is also reference to the same ground, correct?

After your first description I assumed, that you really have a major issue on your pcb (like bad contact, illegale voltage conditions, ...). But as I understand it now, all your pcbs show correct funcionality. And the pcbs, that react to external influences, shift by a very small fraction of the output voltage range. (4mV from 17.5V).

Considering the high total gain of 660*12, these 4mV corresponds to an input voltage shift of just 0.5µV. The temperature coefficient of the input offset voltage is 0.4µV/°C - so a shift of 0.5µV in response to external influences does not seem unlikely to me. 

Therefor I do not believe in a real "defect" of a part of the boards: they all work in principle, and some are more susceptible to external interference than others. For the hand soldered pcbs I could imagine, that this different susceptibility stems from slightly different mounting conditions (small differences in the cleanliness of the pcb, small differences in the mechanical stress on the IC housing, ...) Keep in mind that we only talk of an input shift of some hundred nV.

dclm said:

About earth connection... Why do you think connecting AGND and earth on board has this kind of effect. 

As I understand the schematic in your second post, the cables from your load cell are shielded, and the shield is connected to earth. Maybe the connection from AGND to Earth reduces the remaining coupling on the input signals. Again keep in mind, that we only talk of an input shift of some hundred nV.

If something on the bad pcbs had been really been "broken" or "wrongly designed", you might have had a good chance of finding the fault by remote diagnosis in the forum. But when it comes to working circuits that only differ in their sensitivity to interference in the range of a few 100 nV, I'm afraid I can't really help much by remote diagnosis.

best regards

Achim

Hello,

I have bought these ad8221 IC's from LCSC. Do pictures give any clue that these are counterfeit products. 

Unfortunately, we are unable to provide information about parts in the EngineerZone community based on their markings.  We suggest going back to your point of purchase to get this information.  Note that ADI only provides warranty coverage or any quality, reliability, failure analysis, applications, or other support on products purchased through ADI authorized distributors and sales offices.  To ensure you are receiving unused, legitimate ADI product, please ensure all purchases are made through the ADI authorized distributors and sales offices listed on our website at: www.analog.com/.../continent.asp.