TDA7376B 2 channel power amplifier How is the internal Vref controlled? Is it a fixed, constant voltage? or is it a scaled fraction of the main power supply Vdc?

[RWagn.3]

If it is a scaled fraction then how does it get filtered from transient changes of the Vdc supply?

If it is a fixed constant voltage then the outputs will be biased at that same voltage and peak output audio will be clipped non-symmetrical. For example, if Vdc supply is 18Vdc and Vref is 5Vdc then the output can swing from 5 to 18V ( a change of 13V before clipping) and the output can swing from 5V to ground (a change of only 5V before clipping). This would severely limit the low distortion power rating.

[Peter BENSCH]

Welcome, @Community member​, to the community!

The TDA7376B is an audio amplifier with a single (asymmetrical) supply voltage. For this reason, the reference point (Vref) is set internally to half of VCC. When AUDIO MUTE is activated, the first amplifier stage is switched from IN to this Vref (i.e. zero), which mutes the amplifier.

To come back to your question: since it is a dual bridge amplifier, in which the speakers are not connected to GND but between two power amplifiers, you always have an (almost) symmetrical voltage swing before clipping occurs.

Good luck!

If the problem is resolved, please mark this topic as answered by selecting Select as best. This will help other users find that answer faster.

/Peter

[RWagn.3]

Thank you for the prompt answer, Peter!

My plan was to use the TDA7376B to drive a stereo pair of smaller speakers in a bi-amped (2.1 channel) system, with a mono class D amp for the woofer. I was going to use the input coupling caps into the TDA7376B along with its ~47K ohm input impedance as a first order high-pass filter for the active crossover at about 150Hz. The input coupling caps would be 22nF each. The DC power supply was to be an unregulated 60Hz transformer>rectifier bridge>eCaps.

Say the DC supply had 2Vpk-pk 120Hz ripple on it (when heavily loaded by the woofer power). VREF would be half of that and have 1Vpk-pk of 120Hz. That 120Hz signal on Vref would appear across the input coupling caps with about 800mVpk-pk amplitude. This would be fine if the amplitude of this Vref "noise" was perfectly matched for both the + and - inputs because the balance bridge amplifier would cancel it out for the differential signal across the stereo speakers. The 26dB (20x) gain of the amplifiers would cause about 16Vpk-pk of common mode signal at the speaker terminals, taking up virtually all of the audio signal headroom. I'm not sure how the COMMON MODE, ICM block works but assuming it is fast enough to null out this "noise" signal, I still have a problem with the + and - inputs not being equal if the 22nF input coupling caps are not precisely matched. A worst-case of +5% tolerance on one and -5% on the other, along with ?% tolerance of the Rin values could cause as much as 10% of that hypothetical 800mV common mode signal become an unbalanced 80mV signal. Amplified by 26dB results in objectionable 120Hz hum.

If input coupling caps are larger (.22uF or more) they would filter much of the noise signal on Vref. But then I would need to add a separate hi-pass cross-over filter circuit ahead of the power amps.

I like the TDA7376B because of it high output current capability. I may need to consider using the lesser current TDA7375AV which has a pin for a cap to filter the power supply noise from the internal 50% dc bias node.

Is my analysis above correct?

As a side note; why is the TDA7375V priced higher than the TDA7375AV? It seems the ...A version has a bit more power and current output than the non-A, making it a premium part.

[Peter BENSCH · Accepted solution]

You got confused by several things: the TDA7376B is not just a classic bridge amplifier, but even an AC-coupled bridge has a very good SVR (Supply Voltage Rejection). The TDA7376B has an SVR of approx. 50dB (interpolated for 60Hz). In addition, however, it is a real bridge, where this ripple has the same effect on both sides, whereby both effects cancel each other out (difference almost zero). The many millions of devices with the TDA7376B prove that it functions perfectly even with a normal, unregulated power supply. Nevertheless, it makes perfect sense to dimension the ecaps on the rectifier sufficiently large in order to sufficiently suppress hum on the supply voltage.

If you use 22nF as the input capacitor, this results in a first-order high-pass filter with a cut-off frequency of 181Hz (f = 1 / (2 * Pi * C * R)) in single-ended operation (Rin=40k). Anyway, you would not notice strong deviations in the output signal with normal tolerances of the input capacitors, because it changes in the same percentage as the tolerance.

Please remember that the specification for the impedance is the minimum value. It may well be that the impedance is significantly higher, which lowers the cutoff frequency accordingly. To be on the safe side, I would put an active high-pass filter (e.g. with an opamp) in front of it.

As for the price differences between TDA7375AV and TDA7375V: you are welcome to use the device with the lower price. 😉

Regards

/Peter

[RWagn.3]

Thank you, Peter!
I agree with your suggestion that it should be better to add an active high-pass filter instead of relying on the input coupling cap.

[Peter BENSCH]

If the problem is resolved, please mark this topic as answered by selecting Select as best. This will help other users find that answer faster.

/Peter