AD623 unity gain, different results depending the length of Rg path?

Hi,

If you look at Figure 28 in the datasheet of the AD623, you’ll see the peaking as you’ve observed. But look at Figure 29 too: there is no peaking. The only difference is the package. Therefore I think the solution is the stray capacitance at the inverting inputs of the internal opamps (where you connect the gain setting resistor). This can depend on the package, but also on the external wiring. If you put the jumpers on, the stray capacitance is higher. This results in slightly higher gain at higher frequencies, since the feedback factor is lower, the noise gain is higher.

I hope it can help to understand the reason.

Hi, 

Thanks to your replay. 

I think I understand a bit more.

Do you have a internal schematic more detailed to see about the internal of the In Amp?

BR

You can find the internal topology in the datasheet (https://www.analog.com/media/en/technical-documentation/data-sheets/ad623.pdf), see Figure 1 (first page).

Hi seth1er,

I would agree with ZoltanDTI's answer. Just a few notes, could you verify that your operation of frequency is at 500kHz? For AD623, operating at unity gain 500kHz is not recommended as seen in figure 28 that there is some peeking at around that area.

Even in figure 29 that there is no seen peaking, the flat response is only at around 100kHz.

Cutoff frequency is usually set as 1decade roll-off to assure stability of the operation. I'm not quite sure that your operation is at 500kHz, if it is, I'll be suggesting that we look for another pin-compatible/footprint compatible InAmp.

Best regards,

Kristine

hi, 

Thanks for your answer.

my operating frequency is really at 500KHz ( and I already test until 1MHz) I noticed that the cut off is around 700KHz (-3dB). I don't understand why working at 500KHz is not recommanded for AD623.

if you have another inAmp better I'm interested. 

It's for a low power application (ad623 is under 1mA)  with human skin(noise environnement). 

Thanks for your help

Hi Seth1er,

As a rule of thumb to avoid instability, we want the flat response to be twice of the operating frequency. For example, AD8422 is the closest in specs to AD623. It's GBW plot is seen below:

G=1's flat response is about 1MHz, this is very suggested for your operating frequency 500KHz. I'm just quite sure if you have the same footprint/packaging.

Hi Seth1er,

As a rule of thumb to avoid instability, we want the flat response to be twice of the operating frequency. For example, AD8422 is the closest in specs to AD623. It's GBW plot is seen below:

G=1's flat response is about 1MHz, this is very suggested for your operating frequency 500KHz. I'm just quite sure if you have the same footprint/packaging.

hi, 

thanks for your proposition. I understand for GBW. but the one you propose need 4.6V in single supply and I have a batttery powered, 3.7V.

maybe AD8226 

but it's different from datasheet and the diamond plot tool below 

Hi,

You need to identify your Vcm and Vdiff for the tool to work properly. I could help you understand the tool, could you give details on what your Vcm and Vdiff is?

Thanks!

Kristine

If you need gain of 1 only, you don't have to use an inamp. If higher gains are also required, then you may not find an IC inamp that have sufficient bandwidth. So you may consider to build your own inamp (or, if gain of 1 is enough, a difference amplifier with input followers) using opamps. You can easily find many opamps that can meet all of the requirements.