Cyrus two phono stage oscillations

[Bonsai]

Not sure why Bruce Hofer went for the split series 47pf , but 22 pF always worked for me with ne5534 and the mc preamp has a lot of gain anyway.

Two pole compensation aka TPC. This will raise HF loop gain by 20-30 dB, dramatically reducing distortion.

You can take the resistor to either of the rails or to ground.

A better more sophisticated technique is ‘transitional Miller compensation’ aka TMC that’s would return the resistor to the output rather than one or the rails or ground. The benefit is you don’t have the closed loop gain peaking you see with TPC.

🙂

 

[mpa2]

There are other issues that give me pause. The regulators and their output caps seem to be under suspicion; are they innocent, and only the MC stage(s) at fault? The supply bypass caps on the MC stage seem unusually small, and I don't see any bypass for the other opamps indicated on the schematic.

Well, the regulators are another PITA. They run pretty hot, reducing the voltage from appr. 40V to 18V. I'm considering adding a series resistors to help them a bit, although there is not much space there to do it clean. The same is with additional bypass caps. One could stay on the underside of the pcb, but a tidy way it isn't🙂

 

[MarcelvdG]

The MC stage also has a lot of extra loop gain due to the extra input stage.

If the extra input stage had infinite bandwidth, C7 should be 22 pF times the gain of the extra input stage divided by the attenuation of the feedback network, which should boil down to about 33 pF. I think there is a good chance that the phase shift of the extra stage is small and 33 pF will work fine, but just to be sure, I would try a higher value first.

I realized today that my calculation was wrong. 22 pF is the usual value for the external compensation capacitor when you want to make an NE5534 unity gain stable, but that comes in parallel with an internal capacitor of nominally 12 pF. This internal capacitor makes the NE5534 stable for gains of 3 and greater, that is, with a feedback network that attenuates at least three times.

I had estimated the feedback network and external input stage of the MC amplifier to together have a gain of about 1.5, so compared to the unity-gain case, the total compensation capacitance would have to be increased by about that factor if the phase shift of the extra stage is small enough. 1.5 * (22 pF + 12 pF) = 51 pF. Subtract the 12 pF that is already inside the op-amp, and the external capacitor should be 39 pF.

Fortunately, manufacturers normally take rather conservative values for compensation capacitors, which explains why it also works with 33 pF.

 

[mpa2]

It seems to work with 33 pF, so I'd probably leave it so. Or is 39 pF the way to go? Btw, NP0 is OK there (I used a 1206 smd) or should I go for a film type?

 

[MarcelvdG]

I'd also leave it so. NP0 is fine.

 

[b_force]

In general NP0 performs very similar to film and is also extremely stable in sense of temperature stability.

 

[mpa2]

Well, the regulators are another PITA. They run pretty hot, reducing the voltage from appr. 40V to 18V. I'm considering adding a series resistors to help them a bit, although there is not much space there to do it clean.

Found two very convenient zero-ohm links in the supply path of the regulators in front of the rectifiers. Two 47 ohm resistors here have about 4.5 VAC drop across them, the regulators feel better now .A better way would probably be a smaller transformer with 20V secondaries, since the guy uses a PSX (an external power supply for the power amp section), but I don't want to modify this old beauty so heavily🙂