Poor amplifier frequency response.

[sgrossklass]

If C9 is 220pF and R49 is 12 k as in post #12, the bandwidth is limited to 60 kHz; this appears to be low to me...

Sounds plenty sufficient for an audio amplifier. There is no reason for closed loop bandwidth to extend into the longwave range. Values around 50 kHz actually aren't uncommon.

Sandrov and Knutn the feedback components are: C9 220pf and R56 8k2.

That's 88 kHz then. -1 dB at about 44, -0.1 dB at about 22 kHz.

 

[Gofer]

Maybe it's only me, but I see another signal riding that square wave shown on first post. That will give poor bass response in practice.
I suspect oscillation.
My 2 cents.

 

[knutn]

That's 88 kHz then.

Yes, and after what I can see, it also matches the 10 kHz square way in post #1.

But I still think it's low, the phase lag is more than 13 degrees at 20 kHz...

 

[sandrohv]

Hi Sgrossklass, you are correct about GM_A1. The first diff pair is just a gain block, and the GM comes for the second diff pair (Q8 and Q9). The overall GM is then

GM_A1 = (A_V diff pair Q1,2) x (GM diff pair Q8,9) = ~ (3.9k/2.25k) x (1/(750/2)) = 1/216

In reality it will be a bit lower than this since I did not take re's into account, but it is in the neighborhood.

Now, if CF_A2 is 220p as you mention, the BW of the unfiltered amp is 3.3MHz. That is plenty of BW, so the overall BW will be dominated by the input filter.

Oops... made a mistake here. The BW is not GM on CF, but RF on CF like others mentioned.

I agree with Sgrossklass: "That's 88 kHz then. -1 dB at about 44, -0.1 dB at about 22 kHz." -0.1dB at 22KHz is quite ok.

As an experiment, pull c9 out and see what happens. Just watch out for a potential oscillation on A2.

Best, Sandro

 

[bkneale]

Gopher you are right. Both channels with a slight mismatch are shown in the first image.


Sandro I can get to C9 but am reluctant to pull it out. With double sided soldering it's not that easy.

 

[bkneale]

Sorry Gofer for spelling your title wrongly.

 

[sgrossklass]

Does this amplifier have an output inductor? Zobel network? If so, do the parts check out?

I think the closed-loop frequency response is nothing to worry about per se, but pulling down high-frequency closed-loop gain that much is quite likely to negatively impact stability margins. You generally don't need more than a few pF to counteract the effects of input capacitance.

 

[bkneale]

There is a Zobel network as shown in image 3 which I think covers my amplifier. I haven't checked them out. There is a coil wound around a resistor but I cannot identify it.

 

[Mooly]

There is no coil in your diagram. The Zobel network is also slightly unusual in its choice of resistor value although it will have no impact on the amplifiers response.