Basic question about Class-D closed loop

Hi Suzuki-san,

Please see below for response, which is embedded in your original query.

Thanks,
Cody

Takumi Suzuki1 said:

Hi,

Let me ask you some questions about Class-D closed loop by comparing between TAS5710,TAS5754M and TAS5760M as below.

1) I thnik that TAS5710 and TAS5760M are "pure class-D amp" but TAS5754M is not pure.

    Because TAS5754M consists of "DAC+Class-D" and PWM signal is made from analog audio signal (from DAC).

    In other words, TAS5754M is analog input class-D amp.

    Is my understanding correct? All three amplifiers have the same job of converting a digital word into an analog word. How they do it is slightly, but not completely, different. Some of this is, of course, trade secret so it's difficult to provide you with all of the detail that I'm sure you'd be interested in. However, here is what we can say: The TAS5710 is a "PurePath" amplifier. It is a very old part that was created back when the "PurePath" marketing name was well-managed and only applied to the direct conversion from PCM to PWM. The modulator used in the TAS5710 is called many things, but is  generically referred to as a digital PWM modulated Class D amplifier. In both the TAS5760M and the TAS5754M, they do go through a DAC stage and then on to an analog modulator, which uses the traditional comparison of two analog signals to create the PWM.

The modulator of the TAS5710 can be thought of as black box which takes the digital word on the audio bus, cross-references it to a look-up table of given widths that are stored in the device, and then sends a low voltage pulse of that pre-determined width off to the error amplifier sitting in front of the PWM input power stage. One thing to note here is that the modulator is still "outside of the loop" and it depends heavily on the linearity of the clock signal being used to make the time comparison. In fact, one of the issues with the older closed loop devices was that the modulator is more susceptible to clock jitter because the conversion of the digital word to a pulse of a defined width depended on a very clean clock (hence the reason for the 2nd order PLL filter connected to the device).

Contrasting the modulator of the '10 with that of the '60M and the '54/'56M, this modulation is done more like a traditional DAC. There are, of course, special tricks that we do with the DAC to make it low cost (like the TAS5760M) but still have great performance. There are other tricks that we can do to squeeze even more performance out of the architecutre (like the TAS5754M). Irrespective of the techniques employed to increase performance, both of these devices have lower susceptibility to clock jitter than the older PWM modulated TAS5710.

Moving the discussion forward to the loop around the amplifiers, there is a little difference, but nothing that would be appropriate to discuss here. Suffice it to say that we close a feedback loop around the PWM signal in the TAS5710 and we close a feedback loop around the analog signal in the '54/'56M and the '60M. These feedback loops exist primarily to correct for errors in the output stage. We do that better in the '54/'56 than we do the '60M and '10. Then again, the '54/'56 devices are more expensive than the other two as well.

2) I think the performance of both TAS5710 and TAS5760M looks similar.

    So I suppose the both arechitecture of "Closed Loop Class D Amp" seems to be same or very similar.

    Is this correct? Actually no, they are very different- see above. It is true that they are similar, but the '60 has advanced architecture that does the same performance in a smaller space, with a smaller bill of materials, for a lower price.

3) About "closed loop", the feedback PWM signals are from "H-Bridge FET" out, right? For the most part, yes.

    And is the feedback PWM signal compared to input PWM signal? (Analog comparison) That's the part we can't tell you, but I think you should be able to tell from above explanation. :)

    Or the feedback PWM signal is "LP filtered + A/D converted" and compared to input digital audio signals??   ... another bit we can't tell you. :)

4) TAS5754M performance

    Though TAS5754M is using Analog signal to make PWM signal, the TAS5754M performance seems to be better than TAS5710/5760M.

    I do not understand well the reason why... Neither can our competitors... and that makes us very happy.

    The signal path of "Digital Audio data ---> DAC ---> Analog Audio data ---> Analog PWM" seems not to be better than all digital signal processing. As the saying goes, "The world is analog." so all digital roads must lead to an analog road eventually. You can do good and bad stuff in digital and you can do good and bad stuff in analog, its knowing how to do the good stuff for a lower price than the next guy that makes your product interesting.

    The reason of the good performance is DAC performance of TAS5754M, right? I would say that the very nice BurrBrown DAC is a part of the reason that it has good performance. Another part is definitely the loop architecture.

Thank you for your support!

Best Regards,

Cody