THS4551: FDA input driver for ADC

Change the feedback caps to 80pF to set a 2MHz pole and then check your phase margin initially leaving the output RC the same

Hi Michael / Kai,
My input signal is 250KHz, 0-4.5V sine signal.
I am attaching my circuit herewith. Please check and update me if there are any corrections.
My Vocm is set to 2.25V

My ADC is ADS8900B.

Hello Shibin,

  I have attached transient/AC/stability sims below. It has around 56 degrees of phase margin shown in stability file, and the AC response will pass your input signal of 250kHz.

THS4551_Shibin.TSC

THS4551_Stability.TSC

Thank you,

Sima

Hi Sima,
My signal is a unipolar signal ranging from 0-4.5V, 250KHz

Hi Sima,

I get these results:

I'm a bit unclear how to simulate the THS4551 internal 0.6pF feedback caps. Maybe Michael can shed some light on this 

shibin_ths4551.TSC

Kai

Couple of things here working from Sima's LG sim file, 

1. I don't think you need the Vcm into the other signal side Kai, just ground that side

2. What I have done in the past is take the internal 0.6pF out using external -0.6pF elements then add that  0.6pF to the feedback cap

3. The seperate 1.2pF on the inputs is incorrect, just use the differential input Z across the summing junctions

4. I scaled the LC up and moved the L just across the feedback, what Simi had will work too, This just looks cleaner

5. I changed the sense polarity to report phase margin directly - again about 55deg so looks fine, these changes shifted things just a little bit, not much, 

And this modified LG file 

2605.THS4551_Stability.TSC

Hai Kai,
Here is my simulation result.
Hope this will be ok.

2656.THS4551.TSC

Incidentally, here is what happens if you zero out those two 10ohm inside the loop at the outputs. <20deg phase margin, as time went on, it just seemed more and more like those should have been inside the part - this is the highly resonant open loop Zout interacting with the feedback C - the little series R helps a lot and we did put those inside the THS4561 output lines at the cost of the little output swing under load. 

And I see Sima is using the original model which has that more reactive Zout

The newer one I think doesn't show this as well, here that is, yes, moved up to a simulated 24deg phase margin, the older model is a little more accurate in this reqard. 

And this updated file with the 2020 model, which I had trouble finding now onthe web folder, but I had pulled it in some months earlier to test. 

THS4551_Stability 2020 model.TSC

Oh and one other note I think I referred you to another thread, the THS4551 is apparently have some supply issues, Here is that thread again - it may be resolved by the time you needs parts, but shouldn't wait too long to ask (THS4561 would probably enough for this app as well)

https://e2e.ti.com/support/amplifiers-group/amplifiers/f/amplifiers-forum/1011426/ths4551-ths4551idgkt---looking-for-alternate?tisearch=e2e-sitesearch&keymatch=ths4551#

Thanks for clarification, Michael

I just noted that I have made a mistake. R1=0 instead of R1=10. With this correction I get this result:

With your method of correcting the 600fF capacitances I get these curves:

shibin_ths4551_1.TSC

Comparing both, with and without correcting the 600fF, yields:

Kai

And there's actually no common mode input capacitance, Michael?

Kai

For the diff input pins that is correct, at one time I spent a few days trying to find an app that would fail if I didn't put some input Ccm on the FDA - couldn't find one, so the diff input Z is adequate. you might want to have a cap at the input Vocm control pin depending on where that is coming from. 

Any updates???????????

what does that mean? You are not seeing the foregoing material?

I think what you mean is there any circuit change needed, looks ok as is for stability, there are all kinds of other things to consider, like performance with the ADC sampling, those SAR folks might be able to help there

And then there is the availability question if that matters to you. 

Thanks, Michael.

in any case, avoiding stray capacitance at the inputs of THS4551 would be a good idea:

Kai

That's a good idea for all high speed devices