Class D BTL noob question

Sorry for this basic question, but I can't seem to find a simple answer:

when running typical class d chips (like those TI often used here, like tpa3116, tpa3118, tpa3122or even the mono tpa3106 ecc...) if they are run as BTL amps, they should receive a differential input, right? So a normal signal and its inverted polarity signal. Or do they run also on normal signal and ground?

Thanks
 
Mostly classD chips do accept differential inputs yes, but they're not bound to that by being BTL. For example the NXP TDA8932 can be used BTL or SE and it also has differential inputs.

Having a differential input by no means requires input plus inverted polarity input. They can run just as normal single-ended inputs with the inverted input acting as ground, carrying no useful signal. This is generally necessary because the ground local to the amplifier chip is going to be rather noisy.
 

Spaiz

Member
2016-02-01 1:38 pm
Bergamo
thanks for the answer!
In other words: if I use a stereo amp in a BTL mono config, I don't strictly need to feed it a balanced input; being differential it subtracts the negative (zero) from the positive (normal signal) and it amplifies it.

I'd guess a balanced input would be nonetheless better though - better noise rejection, higher "differential" signal and so on...


Mostly classD chips do accept differential inputs yes, but they're not bound to that by being BTL. For example the NXP TDA8932 can be used BTL or SE and it also has differential inputs.

Having a differential input by no means requires input plus inverted polarity input. They can run just as normal single-ended inputs with the inverted input acting as ground, carrying no useful signal. This is generally necessary because the ground local to the amplifier chip is going to be rather noisy.
 

Spreader

Member
2010-04-01 7:58 pm
tpa311x, mono tpa3106, etc.
they all have a differential built in for btl setup and can work with SE connections;
others: TAS 56xx family or tpa3122, tpa3123 they need an extern SE to diff for btl setup
you can use opa1632 or if that is too expensive; you can build that with 2 normal OP-amp for one channel (4 for stereo) .(if you intend to power them with single supply you need to be more carefully with designing that)
 

Spaiz

Member
2016-02-01 1:38 pm
Bergamo
tpa311x, mono tpa3106, etc.
they all have a differential built in for btl setup and can work with SE connections;
others: TAS 56xx family or tpa3122, tpa3123 they need an extern SE to diff for btl setup
you can use opa1632 or if that is too expensive; you can build that with 2 normal OP-amp for one channel (4 for stereo) .(if you intend to power them with single supply you need to be more carefully with designing that)

Thanks, very interesting informations. It's either they don't write it in the spec sheet or I'm way to dumb and blind to read it, as I went through those sheets many times in the last days.

Yes a balanced signal sent to the amp would be the ideal case, it maximizes the SNR. Other than that though you're fine sending in an SE signal.

In terms of subjective sound quality, I and not a few others have found input transformers to be the best way of getting a signal in.

That would be indeed cool as ever, I love the sound of transformers and I've used some in preamps. Pricewise though they are heavy. A good trafo would also be about 10 times more expansive than a That chip, 20 times than an opamp, and some 100 times than an inverting NPN thingy.
 

Spreader

Member
2010-04-01 7:58 pm
if you look carefully on block diagram you can spot a fully diff amplifier like opa1632 (i'm guessing is the same but built in)
 

Attachments

  • tpa311x.jpg
    tpa311x.jpg
    159.4 KB · Views: 185
  • tpa3106.jpg
    tpa3106.jpg
    145.6 KB · Views: 183
  • tpa3123.jpg
    tpa3123.jpg
    102.6 KB · Views: 177
  • opa1632.jpg
    opa1632.jpg
    232.7 KB · Views: 175

Spreader

Member
2010-04-01 7:58 pm
if you need an alternative to opa1632 for tpa and tas family you can use my circuit (2 opamps with single supply);the exit capacitor is optional because you already have input caps on tas or tpa;
you find 2 identical opamps in 8 pin capsule(VERY cheap) and can use one chip for channel.
the only downside for my circuit in contrast to opa1632 is that at full input signal the background noise will rise a little for some soundcards.
Also on start-up (tas or tpa) you will have some pop up bum ; opa1632 is far better at this :)
 

Attachments

  • se to diff.jpg
    se to diff.jpg
    188.2 KB · Views: 169

Spreader

Member
2010-04-01 7:58 pm
Your calculation of the +in resistor in the inverting case ignores the impedance (at DC) of your virtual ground (2.5k). It also looks to have rather a too high AC impedance leading to crosstalk between the two opamp elements, particularly at LF.
we have interest only in ac impedance for calculating the input capacitor; the virtual ground will not affect the frequency response;yes we have a high ac impedance because i made this to be cheap with low parts count in response to opa1632 implementation of ti in tas56xx boards where they have a 5k input impendance; of course you can lower the gain resistors to small value for a lower impedance.