THS4551: Slow SPICE model

Hello Erik,

  Would you be able to share your schematic or recreate it in Tina-TI. In a simple circuit schematic, it should not run slowly. Are you getting an output in LTspice?

Thank you,

Sima

On second review, its not behaving properly.  I followed the instructions you gave me in an earlier forum post.  Here is the schematic, can you try and get the model to work on your end and then attach here?THS4551_simple.zip

Yes sure, I can look into this and get back to you by tomorrow.

I think I figured it out, looks like I had my feedback connected improperly, so it was positive feedback instead of negative feedback.  

However, I'm also getting used to TINA now, can you point me to documentation that shows how to measure phase margin in TINA?  In LTSpice I break the feedback loop and insert an AC source, but I'm getting very low open-loop gain.  It seems a bit tricky with this THS4551 diff amp.

I'm trying to make a LPF for the THS4551 at 20kHz, but from what I read there is gain peaking due to the feedback cap resonance, so that's why I'm trying to look at it in SPICE.

Morning Erik, 

One way to do the LG phase margin is shown in this article - some preparatory material is in #5 as well, 

https://www.planetanalog.com/stability-issues-and-resolutions-for-high-speed-fully-differential-amplifiers-fdas-insight-8/

If you need a particular MFB filter, let me know - have done quite a lot in that area using the THS4551, 

Hello Erik,

     To add onto Michael's response, this video series goes over a few examples on analyzing stability for fully differential amplifiers (FDA): FDA Precision lab Stability. To change component properties in Tina-Ti, right click on the component and select properties. 

Thank you,

Sima

Ok, thank you Sima and Michael, I will work through this next week.

Hi Sima and Michael,

I finally got around to this.  I am following ADS8900B datasheet, figure 111.  Here are my results and attached simulation.  Can you review to make sure I did everything correctly?  It looks like there is plenty of phase margin (105deg) probably due to the 2.2ohm/10nF output filter, correct?

Thank you!

ErikTINA_Stability_THS4551.TSC

Hey Erik, 

You had quite a few issues with your set up, it is more like 15deg phase margin, This article tried to show one way to get this done - #5 in this series is background on this, 

https://www.planetanalog.com/stability-issues-and-resolutions-for-high-speed-fully-differential-amplifiers-fdas-insight-8/

Here is the modified file, 

TINA_Stability_THS4551 corrected.TSC

I have to take off now, if you want more detail on what I did here, tomorrow AM. 

No problem, I'm reviewing now and we will connect back tomorrow

Ok, a few questions/comments:

1.  Your article on PlanetAnalog is great except the graphics resolution is too low and I can't see any detail in the images.  Do you have a better source for the document?

2.  It looks like you forgot to connect VCOM to GND, although after I connected it to ground I got the same result as you.

3.  I'm getting a bit turned around on phase margin.  Ideally a graph will show 0 and then shifts towards -180 (the first -90 is typically from the internal dominate pole).  But in all cases I've been simulating here I can't seem to get it to start at 0.  If I start the simulation at 10Hz I get a starting phase shift of -160deg.  I'm not sure what is going on here.  What does it mean to start at 100deg phase shift as per your screenshot?

4. It doesn't seem like the 1.2pF capacitor is doing anything, is that just a placeholder that I should put in the schematic?

5. What is the significance of 99.97MEG vs something like 100MEG?

Thanks!

Erik

Ok Erik, these are all good questions, 

1. Yes, when EDN planet Analog migrated platforms back in August 2019 the links to expand images were all broken - here are pdf versions of #5 and #8 (you should review #5 first). I prefer to break the loop at the input since that will keep the feedback network as part of the load to the internal open loop output impedance. Breaking it at the output as shown in other sources isolated that effect missing some phase margin issues in some cases (like putting a feedback cap across the feedback R). 

8078.The SSU part 5 Stability issues for high speed op amps_introductory issues article text Feb. 2_2019.pdf

Stability Issues and Resolutions for High Speed Fully Differential Amplifiers May1.pdf

2. Vcom input control will default to mid scale on the supplies if floated. If you connect to ground on single 5V supply that is out of range but the model might still work - the device will not. 

3. Over time we have mainly migrated to rotating the sense meter to report not phase shift around the loop but phase margin - rotation is adding 180deg to the report phase so it starts out at 180 and moves to 0. 

4. As you will read in part 5 the breaking at the input approach I use only requires you to manually place the input impedance of the FDA at the summing junction, that is the 1.2pF - often does not do anything, but sometimes does so I always include it, 

5. For a while there I was running into numerical chatter in TINA on some models, using those odd values seemed to help that, (especially with the THS4541 model)

Thanks for your reply.  A couple more questions:

1.  Intuitively it seems that the feedback inductors aren't needed.  Since the inputs are disconnected from feedback, the gain should be large (open-loop) anyways.  But obviously it is needed to get a valid simulation.  What am I missing?

2.  Any ideas how to increase the phase margin from 15 to something closer to 20 or 30?

Erik

I'm having trouble opening that file, giving me an error, can you check it?  I tried to rename it to something, but still I get an error "unable to open ..."

Nevermind, I fixed everything on my end per your screenshot.  Thanks for your help Michael!

Here you go, 

2703.TINA_Stability_THS4551 corrected.TSC