Please note that E2E has undergone a major upgrade. Before logging in, please clear ALL your browser's cache and cookies.

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.

THP210: THP210

Part Number: THP210
Other Parts Discussed in Thread: THS4561, THS4541, , OPA1637

happened across this newer part, the higher supply range super beta input FDA version, sure there is plenty of other stuff, but the output swing specs should probably say headroom, 

Scanning on down to the typical MFB example, I had found a little better phase margin to connect the feedback caps on the output side of the 2 inside the loop output resistors - that was on higher speed parts like the THS4561 - might not matter here, but the Zol looks pretty reactive, I can also see this is a slightly higher NG peaking solution than necessary, but no big deal. If I get time, will run a comparison from this to a slightly lower peaked solution. Lower NG peaking is of course higher LG and slightly lower integrated noise. 

  • Going on further in the document, I was surprised to only see this reference for the matched input impedance single to differential design flow (probably not that useful for this sub 10MHz part) - 

    https://www.ti.com/lit/an/slyt310/slyt310.pdf?ts=1599349343966

    This 2008 work by Jim Karki is great and actually where I got the starting point for the closed form quadratic solutions. I have found that you normally end up with a quadratic solution if the circuit involves an active impedance. But his doesn't really arrive at any more than the iterative approach ADI and NSM had achieved.

    The closed form solutions are available in TI, starting with the THS4541 datasheet, and so on.Kind of a disservice to the designers to hark back to Karki's material. 

  • Going on, I was going to add the necessary parameters for the THP210 to some design tools, one key term is the true GBP, This simulation gives that, looks like 8.5MHz. Kind of unusual phase characteristic - looks like a zero pole pair to buy phase margin around xover, 

    A wider sweep of this shows the Aol riding along just under 0dB for quite aways up in freq. 

    A little bit of numerical chatter up there, indicative of a pretty complicated model . 

  • Hello Michael,

    Yes, the THP210 TINA macro approximates/follows the AOL and output impedance over frequency plots provided by design.

    As you have mentioned, the device has a zero pole-zero doublet around the zero crossover; providing the phase response seen in the simulations.  As it was explained to me by design, there were some trade-offs optimizing for current consumption vs bandwidth.  The amplifier offers low offset, low drift and very low 1/f noise, making it attractive for some precision DC and relative low bandwidth ADC applications.

    I had involvement in some of the circuits on the OPA1637 and THP210 datasheet; but not directly the author of the datasheet.  Some of the filters were generated using standard Filter software tools such as Filter Pro and others; and tweaked using TINA simulation.  The frequency response and stability analysis were confirmed using TINA.   Some of the circuits were tested on the bench with a SAR ADC, providing optimal SNR and THD results.  Nevertheless, I did not spent time optimizing for noise gain.  If you happen to have time optimizing the filters for lower integrated noise, I would be interested in learning about your analysis.

    Many Thanks and Kind Regards,

    Luis  

  • Morning Luis, 

    yes, I had 1st seen this part way back in 2017 and it looks very nice with that superbeta input for offset and noise issues, not sure you need that high a supply range heading down towards an ADC as the FDA level shifts the summing junctions not far from the Vocm voltage - normally. 

    Optimizing is a strong word for dynamic range issues in active filters - I use the economics term satificing, which means trading off complex issues to move in a better, but not necessarily optimum, decision in a finite time. 

    I did run some alternate designs in this file, in this case not much improvement on SNR (something like 0.3dB) but I did get 2dB improvement in the min LG towards Fo. Essentially trying to hit the target poles while increasing the Q of the noise gain zeroes (always <0.5, two real zeroes) to reduce the added NG peaking that comes from those. 

    Testing the MFB ckt in the THP210 data sheet.docx

    This actually comes from a lot of extended work based on this original application note,  I had heard through Tim that ADI has picked this up as well, 

    www.ti.com/.../sboa114.pdf

    Note the cubic coefficients in this app note were in error, I fixed that while building the Intersil online filter tool and published here, what is in here still does not account for the capacitance on the inverting summing junction to ground, but that can be done without raising the order beyond 3. That becomes very useful to apply decomp VFA to MFB filters (nice noise and GBP benefits heading that way)

  • One more little error in the spec table, that FPBW should not have a minus sign on the 1Vpp, 

    And incidentally, the calculation from FPBW to SR shows it is a lot lower than the 15V/usec reported here. 

    FPBW is -3dB down, so the actual Vpp at that point for 1Vpp test is 0.5Vp*.707 = 0.35Vp, then if the BW is actually 2.4MHz, then the implied slew rate is 0.35V*2pi*2.4MHz = 5.3V/usec. Oh well, who knows?

  • Hi Michael,

    Thank you for the time putting the word document with the analysis where you show the optimization of Noise gain and Loop gain.  I will go to the examples of your MFB application note and then follow through the word analysis.  Although the improvement is subtle in this particular case, I can see there will be other cases where the difference is significant; and optimizing for noise gain will result in lower integrated noise, increased loop gain at Fo will provide a lower distortion close to Fo.  

    Agree, the Full-power bandwidth spec is inconsistent with the measured slew-rate. The Vpp to Vp conversion and -3dB attenuation factors were omitted in the calculation.  I will provide this feedback to the systems engineer so they may correct this.

    Thank you and Regards,

    Luis

  • Again, I never use the word optimize in this multi-dimensional solution space - improve is all I can claim, And yes, higher Q cases can be dramatically improved. 

    On their slew rate, there is also a divide by root2 term as the FPBW measurement point is -3dB down is it not, 

    Far more than you ever cared to know in these two relatively recent efforts, 

    https://www.edn.com/what-is-op-amp-slew-rate-in-a-slew-enhanced-world-part-1/

    https://www.edn.com/what-is-op-amp-slew-rate-in-a-slew-enhanced-world-part-2/

  • Thank you,

    I have asked the team to update the FPBW spec on the next datasheet revision 

    Regards,

    Luis

  • And also Luis, all this shows up in the OPA1637 FDA as well, 

  • HI Michael,

    Correct, both datasheets need to be updated.

    Thank you again for the application note and publication links,

    Kind Regards,

    Luis