Current sensing using AD8411a, with a step-down arrangement

Hello,

your circuit will probably significantly influence gain and accuracy of the AD8411. Both changes are caused by the high internal resistance of the voltage divider, which is 50kOhm || 200kOhm = 40kOhm. 

The change in gain should result from the fact that the divider resistor works in series with the internal resistors of the AD8411 (12kOhm and 600kOhm).

The worse consequence is the influence of your divider resistance on the offset voltage. The input offset current of the AD8411 can go up to 2,7µA. With your divider resistance of 40kOhm, this can lead to an offset voltage of up to 2,7µA*40kOhm=108mV at the input of the AD8411 (beeing amplified to a much higher offset at the ouput of the AD8411). The typical performance may be better, but that's what you have to expect as worst case. In the simulation model, the input offset current seems to be 0, so you cannot monitor this effect.  

The datasheet recommends to keep the input resistance at 100Ohm or below (chapter "input filter"). You effectively set it to 40kOhm by your dividers. 

best regards

Achim 

Hello CalvinJoe, 

CalvinJoe said:

Could a more accurate model be provided by Analog Devices of the AD8411a to analyse the input offset effect and various other non-idealities? 

That question goes to Analog Devices, not to my person. But as a general comment: simulation will never describe the exact behaviour of an individual chip. The model may describe the best case or the worst case or any typical case, and each of these approaches has its place. 

For many amplifiers the input offset current will be dominated by some mismatch of bias currents. For differential amplifiers like the AD8411 it will probably be dominated by mismatches of the internal resistors. 

In AD8411a.lib you can find the following line:

So some mismatch is modelled, but it's very small. The relative deviation of the resistors (12,000165kOhm versus 12kOhm, 569,9887kOhm versus 570kOhm) are much small than the ratio of offset_current/bias_current (2,7µA/484µA).

You may play around with these values to see effects in the simulation. But in the end, the worst case error analysis has to rely on the specs in the datasheet. And that is as stated and calculated above. 

CalvinJoe said:

what alternative product would you recommend that can withstand high common mode voltage of around 160V and can provide a bandwidth in excess of 2.5MHz?

I don't know an integrated solution that fullfills all these requirements. Maybe someone from Analog Devices does, but maybe not, cause your requirements are special. 

You tried to combine a current sense amplifier with external dividers. This is often unfavorable, as the relatively high input current of the amplifier loads the relatively high internal resistance of the divider.

My personal approach would be: use a current sense amplifier as long as you find one with sufficient CM-input range. If you don't find one and have to use precision dividers instead, then don't use a current sense amplifier but use an instrumentation amplifier with very low input current instead. 

best regards

Achim 

Hi CalvinJoe  and Achim ,

Achim said:

CalvinJoe said:

what alternative product would you recommend that can withstand high common mode voltage of around 160V and can provide a bandwidth in excess of 2.5MHz?

I don't know an integrated solution that fullfills all these requirements. Maybe someone from Analog Devices does, but maybe not, cause your requirements are special. 

Thanks Achim  for answering the simulation questions. CalvinJoe , if you want more info about products, you might want to start a thread in the amps forum. 

mike

Thank you Achim for your comprehensive explanation. I'll try the instrumentation amplifier approach. 

Regards, 
Calvin 

Okay, I'll do that.