THS3491: Noise modelling of THS3491 doesn't seem accurate

I had extracted the model noise terms back in 2020

Here is that, 

The data sheet shows this, presumeably from bench measurements, 

Seems pretty close, maybe the model was changed later? I actually cannot find a model now on the web folder. 

Hi Steve,

can you post your simulation file?

Kai

Hi Michael, yes the .lib version you've posted is the one I'm using. As I'm currently using the part in a non-inverting configuration I've found that adding a simple 735 Ohm resistor between non-inverting and ground works pretty well to make modeled noise represent reality. But I do see the need to use this in a differential amplifier configuration to in which case all bets are off so if you can possibly clue me in how to edit the current noise parameter in the model this I can make my own models that match measured results and tailored to the feedback resistors that I need to use. Thanks Steve

Wrong button, didn't solve my issue because rest of the model is in LTSpice so easier hack model file to make simulation results correct

Hi Michael,

The model is simple 2k feedback resistor and 1.050k input resistor. In 150MHz. Measured value (AC coupled, 150MHz after offset calibration in each case) is 526uV but simulation shows only 438uV. By contrast when using 3491 as a symmetrical  differential amplifier with Rf=768 Ohms, Rin- = 384 Ohms input,and  Rin+=786 to ground and 384 Ohms input where both inputs are then tied to ground measured vale (150MHz) = 408uV and simulated value is 425.8uV. The gain of each stage is almost identical main difference is resistor values are higher. As you are  familiar with the inner workings of the THS3491 and the model do you have any thoughts about the discrepancy ? I can't send the whole schematic because it is proprietary however I've given you enough to model each stage in Tina and hopefully figure it out.

To Clarify, 1.05K/2k is an inverting stage with non inverting input grounded. Diff amp is Rfb/Rin on inverting input, and Rin/Rfb to ground on non-inverting input. In both cases inputs are grounded at side not connected to THS3491.

Hello Steve,

Thank you for sharing the information you were able to share with clarity.  I will also take a look at what you have sent over early next week, as I am on holiday soon.

Best,

Alec

Hi Alec do you have any updates on this issue ? Best Steve

Hi Michael,

Maybe the biggest issue then is that the TI THS3001 evaluation board is noisier than the simulations and datasheet predicts? With the TI THS3001 evaluation board for this part , non inverting input = zero Ohms, inverting input =50 Ohms. Rf = Rin = 1k Ohm I see 207uV RMS noise on the scope in 150MHZ bandwidth. The scope alone is 111uV RMS so backing this out gives 174uV amplifier noise. When I simulate using the THS3001 model it gives 130uV. So 30% higher than simulation model predicts. But where the model really goes off the rails is with 500 ohms in and 2.5k Rfb. Here the simulation predicts 172.3uV but the calculated value is much higher as is the measured noise is 334uV so backing scope noise out yields 321uV. RMS. So it appears model gets less accurate with higher levels. If you have a THS3001 feel free to verify my measurements but something is off.

Thanks Michael, so to recap you agree the THS3001 is lower noise than the 3491 but the 3001 current noise model is incorrect. Am I to infer from this that the current noise is the only issue ? For example does the model correctly cover voltage noise and frequency response for different values of Rfb ? Also in the interests of saving time and being sure that the issue really is solved would you be so kind as to post a link to the thread showing a correct application of external noise to the THS3001 so that I can correctly model the part? Thanks, Steve

The Th3001 is broken on all 3 noise terms as I noted above.

I cannot right now find that e2e where Art Kay posted his noise generators. Those will be TINA files which you might want anyway. Someone in TI needs to provide those I would think. 

Thanks Michael, unfortunately the second link appears to be broken. Also is this link to what Art Kay posted or do I still need to hear from one of those mythical TI apps support guys as well ? Best Steve

Hello Steve,

I do apologize for appearing less often than other mythical creatures.  A mix of end-of-year work requirements and outside-of-work events have rendered me near-nonexistent on the forum. 

As it stands, I am attaching the TINA file I found when I visited the link (which did populate a result).  This TINA schematic contains noise generators, which I believe Art originated & Michael reposted on E2E.

I will take care to be more timely in my interactions; I do prefer the team's mythos to be a positive one.  Happy Holidays as well!

Best,

Alec

Dear Alec, thanks for your reply, for a while I was beginning to think Micheal was the del facto Ti support team :) Unfortunately I couldn't see any attachment not sure that it me or if you forgot to attach ? Thanks, Steve

That 2nd link opened for me, but in any case - this is a TINA V9 File including the noise macro's that Art did quite some time ok, top level looks like this, 

If you don't like TINA, here is the listing for each inside the macro, 

And here is this file including the macros (when I use them I just cut and paste into whatever file I am working)

3755.TINA Noise Sources revC (1).TSC

The TI guys have many (ad nauseum) obligations internally, so there is sometimes a gap in seeing these things. Kai is the other very good resource on this site plugging the dikes as best we can. 

Hello Steve,

Wow, I cannot believe I did the classic "I will attach X" and forgot to do so.  I believe Michael has reiterated it in his response, but I will append the TSC file here also:

7178.TINA Noise Sources revC (1).TSC

Best,

Alec