THS3491: Pspice for TI Simulation Issue

Hi PAB,

with RL1 and RL2 being in the same range, you have an inverting amplifier but not a TIA.

Kai

Hi Kai. The idea is to calculate the value of RL2 (which will be unknown) by measuring the current through it (translated into votlage at the output of the TIA). It would be a front-end stage of a LCR metre. Greetings.

So you kind of don't know what RL2 is? Sounds like a unity gain stable VFA requirement with a lot of output drive if you want to stay with that 100ohm feedback (looks like a load to the op amp)

The other way to stay with the THS3491 is to slip a 402 ohm inside the loop to the inverting node (from the normal inverting summing junction). For LG purposes, it thinks it sees that 402 for compensation. You might try that in sim, does of course increase the noise. 

Hi PAB,

why measuring at 10MHz?

Kai

That's right, it's unknown. There are two stages, one of them to measure the voltage, and this one, to measure the current through the DUT.

I guess I will have to move to a VFA architecture and sacrifice some voltage swing (I was expecting to be able to apply a DC offset of 10V to the DUT, hence a +/-15V split supply was the objective). I'm willing to apply 10MHz signals to the DUT too.

Because that would be the maximum signal frequency the LCR metre will be capable of.

That's great Michael, it indeed worked! How did you get to that value (402)? What would be the minimum value of the DUT I will be able to measure?

You should try some lower values in sim for DUT, 

I guessed, the LG for CFA is Rf + Ri*NG, I just increased the Ri but as the NG goes way up, the BW should be coming down but you are starting with quite a lot here, here is 10ohm input R, still flat through 10MHz, might have room to reduce that 402 say down to 348ohm

Here I went to 348ohm inside the loop and a 1kohm source, about 8dB peaking, stable but concerning, 

Michael, you've been of so much help to me. Thanks a lot!

You bet, obviously your response shape is changing a lot with that input Z. Might want to control the BW with some kind of 2nd stage filter

What about adding a 100Ohm resistor to the source, and measuring the DUT in series with that? Doing so would allow me to measure resistance values down to 10milliohms. I reduced Ri to 40.2Ohm and it's still stable.

Well yes, if you can extact the residual with that 100ohm fixed that will tighten things up of DUT range, you need to look out to a GHz on a part this fast, Might be ok with <10dB peaking - your call. 

The design is thought to apply 10MHz maximum frequency to the DUT, so I think it would be OK. I chose this part 'cause it's quite linear on this range of frequency (besides all of the other features like voltage swing and low noise).