Paraphase inverter question

[Dave H]

What does simulation say? Any difference with bypassed cathode resistor?

not bypassed 100mV input
First triode 5.7V. Second triode 5.45V. Gain 57. Imbalance 4%

Bypassed 100mV input
First triode 5.84V. Second triode 5.56V. Gain 58. Imbalance 4.7%

I calculated 2% imbalance. I think that's because I didn't allow for the output impedance of the first stage being much greater than the output impedance of the second stage.

[Helmholtz]

Interesting, thanks.

because I didn't allow for the output impedance of the first stage being much less than the output impedance of the second stage.

Wouldn't you expect the second stage to have the lower output impedance because of its total voltage NFB?

[pdf64]

Yes, that thought struck me too.

[Dave H]

Wouldn't you expect the second stage to have the lower output impedance because of its total voltage NFB?

Thanks, you're right. I had it backwards and still didn't notice after reading it again
I've fixed it now.

I think the second stage output impedance would be reduced by the loop gain to something like 30k/60=500R which is negligible compared to the 1M feedback resistor. If it was 30k the same as the first stage it cancels out. There would be 30k added to the feed on resistor and 30k added to the feedback resistor so the feedback factor is still unity.

EDIT: I ran the sim. again with the feedback resistor 5% larger (1050k) and the outputs are balanced.

[catalin gramada]

What does simulation say? Any difference with bypassed cathode resistor?

"If the two triodes are really operating in push pull,the current in R3 should not contain any alternating component,and C has no decoupling function."- N.Crowhurst

[catalin gramada]

More material:

Crowhurst_Cooper_1956_High_Fidelity_Circuit_Design_text.pdf

from p.126 and the follow

[Dave H]

More material:

[ATTACH]54964[/ATTACH]

from p.126 and the follow

We think we are cleaver with our computers and simulations but it's true, we are standing on the shoulders of giants. I've only just worked out that the feedback resistor needs to be 5% bigger for perfect balance but Crowhurst had done it back in 1956 with just a slide rule.

[pdf64]

When I've tinkered with both simple and floating paraphase circuits with a type 1 master vol after it, I found that if the cathode resistor was shared and unbypassed, horrid blips of parasitic oscillation were generated when it was overdriven. Of course it could have been something else but bypassing or splitting the cathodes got rid of the issue.
Note that the Vox AC50 variants that used a paraphase had its shared cathode bypassed, also the Ampeg B15, so perhaps they noticed it too?
https://el34world.com/charts/Schemat...Vox_ac50_1.pdf
https://el34world.com/charts/Schemat...eg_b_15_nc.gif

[catalin gramada]

When I've tinkered with both simple and floating paraphase circuits with a type 1 master vol after it, I found that if the cathode resistor was shared and unbypassed, horrid blips of parasitic oscillation were generated when it was overdriven. Of course it could have been something else but bypassing or splitting the cathodes got rid of the issue.
Note that the Vox AC50 variants that used a paraphase had its shared cathode bypassed, also the Ampeg B15, so perhaps they noticed it too?
https://el34world.com/charts/Schemat...Vox_ac50_1.pdf
https://el34world.com/charts/Schemat...eg_b_15_nc.gif

I don't think so but who knows.... I have not You mentioned issue in my experiments. And think the decoupling cap is useful just to can apply the global nfb to the first section. It is a bit tricky due to the double function as bias resistor and component of voltage divider of nfb but can be done. I much prefer the Grommes solution for the moment.

[catalin gramada]

With equal values resistors, not precisely balanced but didn't looks so bad.



Of course the values should be tailored a bit...

[Helmholtz]

I much prefer the Grommes solution for the moment.

Does someone have a good explanation of the strange asymmetrical cathodes' wiring of the PI in the Grommes circuit?

[catalin gramada]

I cannot provide a good explanation, there are not, but who cares ? (meant why should be so important as time as are decoupled?). The circuit should work with separate cathode resistors as same, but think economy reasons...

[catalin gramada]

And just separate the cathode and tested: it works as same, with same degree of imbalance. Think the more important think in circuit are anode resistors, the second triode should be compensate in any circumstance for precise balanced results. Nfb can be applied in the bottom of first cathode resistor in more traditional way.
But Grommes shows very elegant simply solution

[catalin gramada]

What I found is the circuit ( in my case) support a limited amount of nfb. Over 12 db sign of instability appears. Think is more related by shifting phase and for this reason I asked at beginning of topic about. Of course think is possible to shift the pole changing the cap value but just thinking at blocking distortion opportunities...

[Helmholtz]

O.K., meanwhile I understand that the asymmetrical cathodes' wiring (additional unbypassed 100 Ohm resistor at the first tubes' cathode) is necessary to allow feeding NFB only to first tube's cathode.