THS3061: CMRR at 10 MHz

Hi Colin and welcome to E2E,

Unfortunately, I was unable to determine why these two curves are so different. One suggestion is that different supply bypassing is used for a dual chip. As for why its CMRR increases above 120MHz, that is outside the bandwidth of the amplifier and should be ignored. For your application, is it possible to use a differential output? If so, a fully differential amplifier such as the THS3491 will give you good CMRR.

Best regards,

Sean

Hi Sean,

Thanks for looking into this. It does seem unlikely that power supply bypassing would explain the vastly different CMRR plots. Surely we could assume that the professional TI engineers that produced these plots know how to provide correct and effective bypassing on the test fixture.

The question remains unresolved, in that we cannot be sure that there is not an error in the published plots, and we can find no credible explanation for why the plots look so different for the single and dual versions of the same op-amp.

I have ordered some THS3061, and may measure the CMRR at 10 MHz myself as this does not appear to be a difficult measurement. The CMRR vs frequency plot for the THS3001 shows the test measurement circuit used for that plot. Was the same test circuit used for the THD3061 plot of CMRR vs frequency? I presume it was, but TI have been naughty and not told us. :)

Cheers, Colin

Hi Sean, I forgot to thank you for pointing me to the THS3491, which is not a fully differential amplifier as far as I can see, but I don't need a differential output anyway. The CMRR is not given, but broadly speaking the performance of this op-amp is truly state-of-the-art. I may not use it for this application, but will file it in my head for future reference. Colin

Hi Sean,

The more I look at this, the more I am inclined to think that one of these plots is wrong. As an example of what we should expect, take a look at the specs for the OPA694 and OPA2694, two op-amps that are also identical except that one is single and one is dual. And we find, as expected, that the CMRR plots for these two op-amps are identical, being 44 dB at 10 MHz.

Given that all evidence points to one of the THS3061 / THS3062 CMRR plots being wrong, I think that TI needs to look into this. As a professional designer, I need to know the correct CMRR for these op-amps, and for the THS3061 in particular. Would it thus be possible for you to look into this further? 

Thanks,

Colin

Hi Colin,

I have ordered some THS3061s and THS3062s and will work with characterization to revisit this specification, and I will get the datasheet revised if necessary. I apologize about the THS3491, it is single ended.

Best regards,

Sean

Hi Sean,

Thanks so much for that, really appreciated. I look forward to see what you come up with, and will share anything I learn as well.

Cheers, Colin

Hi Colin,

you might be interested in the CMRR TINA-TI simulates?

TINA-TI shows the same result for the THS3061 and THS3062. Both models are identical, very probably.

colin.TSC

Kai

Thanks for that Kai. The simulated result lies somewhere between the published THS3061 and THS3062 results. While the simulation provides a guide, simulation is never perfect and measurement is required to get an accurate CMRR plot, especially at higher frequencies.

At this stage I have 2 issues. I think we would all like to resolve the apparent error in the published CMRR specs, and appreciate Sean's effort here. But with that said, I have come to realise that I have issues with my wideband unity gain differential amplifier regardless. My THS3061 arrived today, and I found it performed no better than my original lowly LM6171. My circuit is identical to yours above except that the resistors are 820R. It is well known that the resistors will need to be matched to a high degree of precision to obtain good CMRR from this circuit, but within limits that is easily achieved. But at higher frequencies, unmatched stray circuit capacitance will also unbalance the circuit and degrade CMRR. For example, at 10 MHz a stray capacitance of only 2 pF represents an impedance of about 8 kohms, which is only 10 times greater then the 820 ohm resistors that I am using, and significantly degrade the CMRR at 10 MHz. Right now I am sure this is the major factor degrading my CMRR performance. In retrospect, I would have laid out the PCB with particular care to minimise stray capacity on all connections associated with the + and - inputs. I can balance the stray capacities by adding a few pF here and there, and then see if I am eventually limited by the CMRR of the op amp itself. Due to unmatched stray capacitances, it is not as easy as you might think to build a differential amplifier with good CMRR up to 10 MHz and beyond. 

Cheers, Colin