1
\$\begingroup\$

AD9655 ADC

The signal is AC coupled so I don't "need" the DC coupling availability of FDAs. I don't need another gain stage either, so the FDA would be in a 0dB gain configuration.

My main concern is one of my requirements is to have below 7-9 nV/rt-Hz output noise, all stages included. A single differential amplifier stage is going to have 4-6 nV/rt-Hz output noise from resistors alone (if I choose low values,) plus the ~1nV/rt-Hz injected from the differential amplifier itself.

My understanding is that transformer baluns offer superior amplitude and phase balance, lower noise injection (if any,) so in my application where SNR is everything I can see why it would be the choice. Look at the AD9655 app notes, they say to use a balun rather than an FDA.

Meanwhile an RF FDA such like the LMH6554 would offer extra common mode noise immunity (the only benefit???) and inject unwanted noise into the signal, degrading SNR.

Another point of confusion for me is the author of this article claims that FDAs will have "better" balance error and distortion specifications. In terms of balance error, the transformer I'm going to use (https://cdn.macom.com/datasheets/ETC1-1-13.pdf) specifies a 1dB amplitude balance and +/- 20 degree phase balance, while the LMH6554 specifies a balance error of -64dB. How do I compare these specifications to decide whether the balun or FDA will create a more "balanced" differential signal.

https://e2e.ti.com/blogs_/b/analogwire/posts/driving-adcs-amplifier-or-balun

Questions:

  • Why are baluns preffered to FDAs at RF frequencies for driving a differential ADC?
  • Are there any benefits to FDAs besides CMRR?
  • Is it logical to compare the amplitude and phase balance of my transformer to the balance error of the differential amplifier? Can you convert between those specifications?
  • Is there any other way to increase CMRR of a differential ADC front end without adding a differential amplifier?
\$\endgroup\$

1 Answer 1

1
\$\begingroup\$

Why are balun's preffered to FDA's at RF frequencies for driving a differential ADC?

Mainly noise, and the ability to change impedance level. Being passive is a benefit. They can distort if you try to use them at too low a frequency, the datasheets are not always clear about this, measure the performance at the lowest frequencies you need.

Are there any benefits to FDA's besides CMRR?

The option to have some gain, and the ability to go down to DC

Is it logical to compare the amplitude and phase balance of my transformer to the balance error of the diff amp? Can you convert between those specifications?

Yes, once you know how they are each specified, so is the difference the total difference, or the deviation of each from the mean level. Making some assumptions, 1dB balance means the two signals are within 10% of each other, so the difference is -20dB balance error.

However, don't get hung up on balance being the most important specification, it only needs to be good enough. The main benefits of a balanced signal are (1) it increases the signal swing at the ADC, good for SNR, any balance gets you a benefit and (2) it cancels even order distortion at the ADC inputs, which may or may not be needed in your application.

Going from no balance to -20dB balance error gets you almost all the SNR benefit you can have, and all of the even order distortion cancellation you have a right to expect. Going to -40 or even -60dB balance error may sound like a big improvement, but remember that those big dB numbers are comparing tiny error numbers of 0.01 and 0.001.

It would be a very unusual RF application where better balance than one balun could provide would be needed.

Is there any other way to increase CMRR of a differential ADC front end without adding a differential amplifier?

If you really need to increase CMRR over that of one balun, you can cascade two. I've seen a design where the signal goes into a transformer with a grounded centre-tapped secondary, then the two outputs going through a further transmission line balun.

\$\endgroup\$
2
  • \$\begingroup\$ You say that "If you really need to increase CMRR over that of one balun, you can cascade two." But my understanding is that a Balun only has CMRR from the balanced/differential side to the unbalanced/single ended side. ( markimicrowave.com/assets/appnotes/balun_basics_primer.pdf ) I probably should have mentioned it in the post but I'm converting a single ended signal to differential then feeding it into an ADC. Or are you just talking about the "inherent" CMRR that comes from differential signalling (i.e Vsig_pos + Vcommon - Vsig_neg +Vcom= Vsig_pos - Vsig_neg)? \$\endgroup\$
    – Max W
    Jun 21, 2021 at 20:29
  • \$\begingroup\$ @MaxW Re-read the last paragraph of my answer, you add a transmission line balun between the transformer balun and the ADC. If you go to the wikipedia article, it's classical transformer followed by transmission line transformer. The latter can be obtained in bifilar rather than coax, so they're more symmetrical. \$\endgroup\$
    – Neil_UK
    Jun 21, 2021 at 20:36

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.