AD8207 experiencing noise on the OUT pin


I want to use AD8207 in combination with an ATMEGA328PU microcontroller. ATMEGA takes care of 10 bit AD conversion. The AD8207 is setup in a Vcc = 3,3V configuration with Vref1 connected to Vcc and Vref2 connected to GND. So I can use the AD8207 for bidirectional current sensing. If I short circuit the inputs of AD8207 pin 1 & 8 I notice the voltage on the out pin is straight in the middle at 1,65V (measured using a voltmeter). However the ATMEGA328PU ADC shows the values jumping up and down between 499 to 520 (where Vcc equals a quantized value of 1023), where I expected a steady value of 511 or 512. It seems there is noise present on the OUT pin. When I connect a dc battery of 1.2 V directly to the ADC input of ATMEGA (so disconnect AD8207 out pin first), the quantized value is steady and does not jump up and down (using the same VCC). To me that proves that it's not the ATMEGA showing problems. I probed the out pin with an oscilloscope, and sure enough there seems to be a broadband noise superimposed on the DC of about 20mV p-p.

I also scrutinized the Vcc chip (LM2931A33) and replaced it by a 3,7V battery powering AD8207 and ATMEGA: no change in behavior.

I replaced the AD8207 with another sample, but with no effect.

Now all this circuitry is build as a prototype on experimental board with the AD8207 soldered on top of a SOIC-8 break-out board. Still I did my best to decouple Vcc using a 100nF capacitor near to the chip. I cannot seem to get it stable. I checked the datasheet but I don't see any instructions an issue like I am experiencing now.

Any tips anyone?



added voltmeter info
[edited by: PeetD at 11:34 AM (GMT -5) on 31 Dec 2020]
  • 0
    •  Analog Employees 
    •  Super User 
    on Jan 2, 2021 6:32 AM

    Hi peetd,

    I'm not an expert on AD8207 behavior, but I'll give you my thoughts anyway until someone more knowledgeable can get back to you.

    The input referred broadband noise of AD8207 is 0.6uV/RtHz and the device has a gain of 20V/V. With approximately 300kHz -3dB small signal BW (from datasheet plot on page 6), the RMS noise at the output will be around 6.6mV_RMS:

    0.6uV/RtHz * 20V/V * sqrt(300kHz) = 6.6mV_RMS

    So, it's not unreasonable to see 20mVpp at the output as you've observed (Noise_pp = ~ 6.6* V_RMS).

    So, it seems like some kind of low pass filtering would be needed to knock the noise level down.

    Is that allowed in your application?

    Let me know what you think and whether my noise estimate makes any sense?


  • Hello Hooman,

    Thank you for looking into this. I get your point, I needed to catch up on my electronics skill as they have been stale for some decades (more into sw engineering and testing nowadays) and I conveniently overlooked this of course. But the spec states it quite clearly and I agree that my rather crude observation of noise can be explained that way. I somewhat mistakenly expected that a short circuit between inputs would lead to zero noise at outputs, but I now see that assumption was incorrect.

    Using a low pass filter is acceptable in my situation and I will start working on it right away..

    I was worried that the chip was somehow oscillating (causing problems), but I am glad that I got that right and do not have to worry about that.

    Thanks again.