Grant RIAA phono with 12AX7/ECC83

[Brian Beck]

The 12Meg and 1000pF feedback parts are a bit perplexing. The 12Meg will provide only a slight degree of feedback in the very lowest frequencies. If the open loop gain were infinite this resistor would need to be about 1.2Meg to give the requisite 60dB or so of gain in the low end, but since open loop gain is finite, this resistor has to be larger than 1.2Meg, hence 12Meg. Besides, above about 13Hz, the 1000pF will dominate, so that most of feedback in the audio band runs through this cap. This large resistor may also participate, along with the 1000pF cap, in adjusting loop phase shift to prevent low-end instability due to too many coupling time constants enclosed in one feedback loop. Simulation would provide more insight if you are really interested.

This is conjecture, but perhaps the designer also felt that biasing the feedback caps with DC would improve their performance. So the 12Meg would provide a DC path to keep the 100pF, the 270pF and the 27pF biased above 100 or 150 volts. Also, at the top of the audio band the 0.47uF output coupling cap is enclosed in a feedback loop by the presumably superior 1000pF cap which is 470 times smaller. Possibly the designer figured any distortion the 0.47uF cap might make would be eliminated by this feedback.

Or, maybe the designer just threw parts together until it all seemed to work OK

[fdegrove]

Originally posted by Brian Beck
The 12Meg and 1000pF feedback parts are a bit perplexing. The 12Meg will provide only a slight degree of feedback in the very lowest frequencies. If the open loop gain were infinite this resistor would need to be about 1.2Meg to give the requisite 60dB or so of gain in the low end, but since open loop gain is finite, this resistor has to be larger than 1.2Meg, hence 12Meg. Besides, above about 13Hz, the 1000pF will dominate, so that most of feedback in the audio band runs through this cap. This large resistor may also participate, along with the 1000pF cap, in adjusting loop phase shift to prevent low-end instability due to too many coupling time constants enclosed in one feedback loop. Simulation would provide more insight if you are really interested.

This is conjecture, but perhaps the designer also felt that biasing the feedback caps with DC would improve their performance. So the 12Meg would provide a DC path to keep the 100pF, the 270pF and the 27pF biased above 100 or 150 volts. Also, at the top of the audio band the 0.47uF output coupling cap is enclosed in a feedback loop by the presumably superior 1000pF cap which is 470 times smaller. Possibly the designer figured any distortion the 0.47uF cap might make would be eliminated by this feedback.

Or, maybe the designer just threw parts together until it all seemed to work OK

Hi,

As I don't have sufficient knowledge to figure out feedback loops by just looking at them I was thinking along the same lines as you point out in the second paragraph. I do doubt however that any designer from that era was concerned about polarizing caps in the FB loop but you never know....
Either way it's a neat little trick.

Cheers,

[soulmerchant]

Ok.

I figured out what the humm was. If you physically move the power supply away from the pre-amp the humm goes away. The 220 uf B+ caps are probably not low esr and are getting close to 20 years old.

I will replace these caps in any case.

What is really interresting is the power supply itself though.

Two separate voltage doublers (one for each channel) - ok not so special... but a highly regulated filament supply using solid state.

I will post the schematic if there is interrest, but why regulate the filament supply so much?

The transformer is dated Aug. 1986. and has no other markings (made in england though you can be sure of that).