1
\$\begingroup\$

I am trying to build an audio amplifier PCB with S/PDIF and single ended line level inputs and want to use the CS42528 CODEC (datasheet).

My questions are regarding the input and output stages of this IC.

  1. The Analog input requires buffering and biasing for the following ADC-Input (AIN+/ AIN-, 4.2.1 of datasheet) and there are multiple ways of doing this. Since my analog input is single ended, I could use either a single-to-differential-ended amplifier or simple single ended configuration (see this appnote by Cirrus). How much of an improvement will the differential output configuration bring about?

  2. Is there any benefit in using a fully differential amplifier IC (like the OPA1632) over a configuration with two op-amps like in the following image?cirrus_recom

  3. Similar to 1., the output stage requires a low-pass-filter. It will connect to the amplifier TPA3004D2 and can either be differential-to-single-ended (like below), single-ended (?) or fully-differential. What are the performance differences between these options? I know that the differential signals will give the best common mode rejection. But what advantages/disadvantages has the differential-to-single-ended setup over the other two?

cirrus_output_filter

All components will be on the same PCB. So in summary, will the usage of differential signaling for this single ended source and differential output bring any noticeable quality improvement? Is there any reason to use a fully differential amplifier IC?

\$\endgroup\$

2 Answers 2

0
\$\begingroup\$

Personally, I don't see the point in fully-differential amplifier ICs for low frequencies such as audio, and I usually use a dual op-amp package for it, as you showed in the first image. It has the advantage that I have a greater variety of op-amp parts to choose from and that I can easily substitute if out-of-stock. That circuit can be modified in many places to implements gain/attenuation/filters and so on.

Potentially, an integrated IC could do better at very high frequencies (>10 MHz), because the internal routing will be tighter than going out of and back into the package. But I cannot really share experiences here.

All components will be on the same PCB. So in summary, will the usage of differential signaling for this single ended source and differential output bring any noticeable quality improvement? Is there any reason to use a fully differential amplifier IC?

That depends a lot on what else is on your board and how you realize the single-ended audio input. If you receive a single-ended input (i.e. signal + ground cable) and attach that ground wire to your local ground, then basically all is already lost, and going differentially from there on, will not remove the introduced interference. If however you manage e.g. to differentially receive your audio input, it sure makes sense to preserve the excellent signal integrity by going on differentially. Maybe this post also helps for some background.

\$\endgroup\$
2
  • \$\begingroup\$ Thank you! So do you think I could use something like this as a receiving element? And is it in general a bad practice to connect the grounds of the single ended input and the receiving PCB? \$\endgroup\$ Jun 30 at 7:38
  • \$\begingroup\$ @celeryperson The transformer coupling is not really feasible to do directly for audio frequencies. You would need Henries (not micro) of inductance to have sufficient impedance down to ~20 Hz. The op-amp based receiver is more like it for such low frequencies. About GND connection: If you have single-ended source and single-ended receiver, you will get inteference if a ground loop is created (e.g. between two mains powered PCs). In contrast, if you can guarantee to connect to a floating piece of equipment, you will not have a ground loop and single-ended receivers are ok. \$\endgroup\$
    – tobalt
    Jun 30 at 8:23
0
\$\begingroup\$

Fully differential amps are a useful niche but I don't think they would be so useful in audio range.

The idea is that in your your two amp circuit resistors and everything should be perfectly matched to reach an high CMRR while a fully diff amp is already manufactured that way.

On the other hand filters need substantially double components and these would need to be matched to achieve both CMRR and filter performance (usually you check the results with Monte Carlo)

The other useful properties:

  • (typically) at the end even PSRR is a little better, which is useful if you don't have a perfectly clean dedicated analog rail;

  • The effective signal is doubled so you gain in SNR;

  • TI says that harmonic distortion is better but I never got around their explanation (here);

  • If you need to drive an ADC they are simply perfect (but if you look around you could find dedicated ADC drivers)

Personally, for consumer audio, they are simply excessive unless you wade in the esoteric pool; for pro audio they could have a use when you need some level of performance (also the pro audio electrical environment is horribly noisy)

\$\endgroup\$
2
  • 1
    \$\begingroup\$ In the two-op-amp kind of discrete FDA, a part of the signal goes through the amps sequentially, which introduces different amount of "lag" in the positive and negative output. I guess that will lead to THD at higher frequencies. If an integrated FDA uses a more symmetric structure (Fig. 2 in your link, they use the same buffers on each output after a differential output OTA) - that would lead to improved THD rating. \$\endgroup\$
    – tobalt
    Jun 29 at 7:16
  • \$\begingroup\$ That's true but TI says it's more deeper than that. Anyway probably even using a TL072 (notoriously slow but a staple of audio processing) I doubt that would be significant for consumer audio \$\endgroup\$ Jun 29 at 7:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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