This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

THS4531A: Difference between THS4531 and THS4531A

Part Number: THS4531A
Other Parts Discussed in Thread: THS4531


I am contemplating on using THS4531A to build a Fully Differential Transimpedance Amplifier.

I realised that there is another component THS4531 which is quite similar to THS4531A.

I figured out the two components are different in following respect:

1. Harmonic Distortion@1kHz

2.Input referred offset voltage and drift

3. Input referred offset current and drift

Apart from these, are there any other differences in the two components ?

Is there any restriction on using THS4531 instead of THS4531A ?

Is there any document telling the difference between THS4531 and THS4531A ?



  • Hi Ashish,

    There are a few differences between those two parts. They have their own datasheets. You can refer to this to compare which parameters are of importance to you. We don't have a specific document listing the difference between these two parts.

    There are no restrictions on using either part. You want to just go through the values you think will be important and make sure they are suitable for your application.

    Here is the compare tool that compares both these parts and highlights some of the more common parameters.

  • Hi Ashish,

    Ignore my last comment about that compare tool. It looks like those values are wrong.
    My assumption is that the THS4531A uses a different process which causes it to have the slight differences. Their GBW and AOL curves look the same. Their output impedance plot also looks the same. So I went through the macro model line by line for the older versions of the netlist and they were the same as well with a difference in offset.

    However, they released a new version of the model for the THS4531A in November last year, which looks a lot different than the THS4531 model. My assumption is that they made updates to the THS4531A model to make it better and more accurate to the part. Now I would tell you to use the newer model with the THS4531A since it should be more accurate.

    Have you tried ABA swapping the parts in the same circuit in PSPICE and measure their Aol and phase response? If there is a drastic difference between them then one of the models is not working properly. Then I think your best option is to calculate the gain and phase by hand calculation and comparing it to the simulation models. However their simulation results should be very similar.

    Also looking at your circuit in the last thread, I would recommend reducing the capacitance on your feedbacks and/or increasing the capacitance between the differential inputs. This will make the circuit more stable.



  • Hello Karan,

    Indeed there is a significant difference in the performance if I change the THS4531 model with THS4531A model. My circuit was perfectly stable with THS4531 model and it became Unstable/Oscillatory with THS4531A Pspice model (see the step response image)

    Also looking at the Aol plot, we see there is drastic difference between the Aol plots obtained with THS4531 (which is lot cleaner at Gain CrossOver Frequency (Unity Gain Frequency) and also has larger DC gain) compared to that with THS4531A (which shows peaking at the gain crossover frequency/unity gain frequency).

    Now, I am in huge dilemma. Which model is correct ?



  • The THS4531A is the THS4531 with die coat to reduce package stress induced offset issues. The models go through various revs and it looks like the THS4531A is a 2018 update - would apply to the THS4531 as well in the AC specs.
    Assuming the THS4531A model is more correct, would need you current schematic to improve phase margin. I was working on that model update some years back, It does look like the updated A model has a 2nd order pole pair at high f in Aol, and that is often commonly there -
  • yes, I was testing the 2018 THS4531A model, and it has the things I was updating back in June 2016. It will be much more accurate than the original release model - the THS4531 model should be updated to this with perhaps higher nominal Vos and Ios.
  • THS4531A LG simulation Zt.TSCI found your earlier circuit - yes, with the new model it simulates unstable, I recomped to nominal butterworth with single pole Aol model - that takes Cf to 1.9pF, but not enough with higher poles and reactive Zol - adding 100ohm inside the loop fixes it nicely, files attached.

    THS4531A closed loop Zt.TSC

  • Hi Michael,

    yes, the performance of THS4531 model looks out of date and somewhat unrealistic. The THS4531A model appears to be much more realistic.

  • Targeted Aol response for model update - here was the unloaded shape I sent out to match what designer files were showing, loading will change this a lot due to reactive Zol. 

    Lumped element Aol model results for the THS4531 model updates circa Jun 2016.docx

  • Hello Michael,
    I am not able to open "THS4531A-closed-loop-Zt.TSC" . Can you resend it ?

  • Ok, I am learning I cannot put more than one TINA file into a post, or I need to close and re-open TINA to to a 2nd file, hope not - I will try again here, 

    THS4531 closed loop Zt.TSC

  • Hello all, some more comments on the THS4531 and THS4531A models,

    The updated THS4531A model is the more current macromodel approach where the input offset voltage is simply set to the typ, 1sigma, 100uV value. However –

    1. The Aol and Zol I think I provided back in June 2016 (which may have been updated later) is giving a peaking gain of 1 response, not matching the datasheet. Normally, the model is tuned to match the over gain curves and Cload curves – maybe more tuning is needed.
    2. The much older THS4531 model is transistor based – usually more accurate. It has the typ Vos of 200uV in its model.
    3. If the THS4531A model is more accurate, it is easy to change the Vos entry to 200uV and rename a THS4531 model and update to 2018.
    4. All of these models have a typ Vos and Ios in some kind of random polarity. However the full material span is bipolar at the ATE test limits.

    I tried to discuss some of these issues these two recent articles.



    Where I discovered a couple of other interesting things.

    1. CMRR has a polarity, and it appears to be random across models (except for CFA).
    2. I think I made a mistake in the FDA output DC error discussion related to input common mode current – fixed in the articles, but I think now the input CM current shifts from the output CM voltage controlled by the CM loop to an input CM offset – that shift and drift then gets to the output Vos through DC ratio mismatches on the two sides – probably not a dominant error.


  • Hi Michael,

    your inputs are highly appreciated!

    Thank you very much. :-)

  • Thanks Michael.