Musings on amp design... a thread split

[Bernhard]

Look at the first schematic with mosfet output stage post #1106

620 ohms to 33 ohms.
620 ohms to 33 ohms.

If you change one resistor to:

620 ohms to 33 ohms.
626 ohms to 33 ohms.

Amplitude is asymmetric.
THD rises.

I verified that in simulation

 

[Hans Polak]

N and P mos have different saturation properties hence clipping is asymmetric (shows best at virtual earth, see pic)- no amount of feedback could correct that.
It will be most interesting to see the IM your amp produces at maximum output, with 1:1 11.5 KHz and 950 Hz.

I think that you are ruining the high impedance of the Vas with a low end stage input impedance, thereby massively decreasing loop gain and thus distortion reduction.

Hans

 

[Bernhard]

The big cap in the second schematic seems to fix the symmetriy problem but now you have the big lytic there, maybe two big lytics back to back…

 

[Aridace]

I think that you are ruining the high impedance of the Vas with a low end stage input impedance, thereby massively decreasing loop gain and thus distortion reduction.

Hans

The ECW mos series used as source follower has a rather high input impedance, to the extent that a source follower doesn't result in THD / IMD improvement.

 

[Aridace]

With just 63 mV DC over C3, back to back won't be much of an improvement.

 

[Hans Polak]

The ECW mos series used as source follower has a rather high input impedance, to the extent that a source follower doesn't result in THD / IMD improvement.

I see two BJT output transistors in your circuit diagram, so what do you mean with ECW mos series ?

 

[Aridace]

The ECWs in post

• #1,069

 

[Hans Polak]

O.k.
Now to which circuit diagram does the mediocre IMD plot in #1113 belong, the one with BJT’s or with ECW’s ?
Rather confusing.

Hans

 

[Bernhard]

To my understanding we have 3 topics here.

First schematic from Petr with mosfet output and 2 feedback path that can introduce errors.

Second schematic from Petr with BJT output from post 1113 where problem in nr.1 is fixed by a big cap and that has mediocre IMD.

3rd schematic from Aridace where he states that P and N devices are different which results in asymmetric clipping and feedback does not fix it.

 

[Hans Polak]

So the text in #1113 doesn’t apply to the accompanying circuit diagram.
I’m lost.

Hans

 

[Bernhard]

Why? He loaded that bjt schematic in MC for his IMD test and found that it is rather poor but low order, so the amp could still sound ok.

 

[Hans Polak]

Why then the reply with ECW’s on my comment that the Vas output impedance was ruined with the BJT’s low input impedance ?
With additional pre emitter followers result would have been much better.

 

[Bernhard]

Schematic from post 1112 and 1113

THD with cap is worse than without.

With cap:

Without cap:

 

[Bernhard]

Why then the reply with ECW’s on my comment that the Vas output impedance was ruined with the BJT’s low input impedance ?
With additional pre emitter followers result would have been much better.

I don´t know.
Maybe he will clear that up.

Small update:

I made a simplified line level buffer from my schematic, THD is -225 dB, for the full version I expect -250 dB or better.

Problem: MC12 transient analysis seems to be limited to -260 dB. Any chance to improve that with settings ?

Black is the sine source, red my amplifier.

 

[Hans Polak]

With just a bit fiddling and increasing the loop gain, I could lower IMD by 30dB still without any optimization.
Left the original and a the right the modified version, both having 850mA bias current.

Hans

 

[Hans Polak]

And when removing the 1m cap over the two 33R resistors, IMD spectrum becomes even better.

Hans

 

[Bernhard]

And when removing the 1m cap over the two 33R resistors, IMD spectrum becomes even better.

Schematic from post 1112 and 1113

THD with cap is worse than without.

Same for THD 😊

Without the cap, try IMD with one of the 330 ohms feedback resistors changed to 333.3 ohms (+1%)

 

[Aridace]

I see two BJT output transistors in your circuit diagram, so what do you mean with ECW mos series ?

So many amps are being discussed that confusion about "which amp?" is likely. I only simmed IMD (950,11500 1:1) for that Hiraga amp with bjt.
In my amps I would only use bjt as EF, when driven by low distortion opamp.

 

[Hans Polak]

Same for THD 😊

Without the cap, try IMD with one of the 330 ohms feedback resistors changed to 333.3 ohms (+1%)

Well, that's the easiest problem to solve, just replace the four 330R and 33R resistors by one 160R and one 16R resistor.
The four 10R emitter resistors are just there to linearize non ideal diamond transistors, but don't affect gain.
With this mod, resistor matching is no longer an issue.
It simply shows that this Hiraga design is not well developed to its ultimate potential.
There is probably even more to improve, but a few steps already improved IMD by 44dB to below -125dB, not that bad at all.

Hans

 

[Bernhard]

I try to run your IMD test, it is over night and still PSS iteration = 1 after 12 hours

Do you have the steady state box checked ?

On another computer box unchecked -144 dB at 1W Pout
IM2 / IM3 are equal