Which is quieter? (Filter node question)

[jordan86]

Replacing a cap can in an old Silvertone 1482. It currently has only three filter nodes.

Stage 1 6v6 plates (20uf)
Stage 2 6v6 screen grids (10uf)
Preamp and paraphrase PI consisting of two 12ax7s. (5uf

The amps not inherently noisy. Fairly quiet actually. Replacing the cap can just to be safe. My replacement cap can comes with 4 caps: 20/20/10/10uf.

I could copy original schemes with a tad better filtering 20/20/10

Would creating an extra unused node between the preamp and power section improve filtering in the preamp? Would simply adding a small (1K or less) resistor between node 2 and 3 give better filtering? Maybe decrease final power rail resistor going to preamp to keep voltages stock at v1. 

20 6v6 plates, 20 6v6 screens, 10 unused*, 10uf preamp/pi.

UPDATE:*when I say "unused" I mean, in the filter path but not tied to any gain stages.

[AlNewman]

You could also run 2 caps in parallel, and do 30/20/10, 20/30/10, or 20/20/20.

[jordan86]

Post #1 updated: schematic added.

6x4 data sheet says 10uf max at first node. Obviously the 20uf is stock and proven to be acceptable. I'd be hesitant to go up to 30uf unless someone has some authoritative word on that.

Besides the higher capacitance vs stock scheme. Would one of these be better than the other? Better than stock concept of only 3 filter nodes?

A)20uf>2.2K>20uf>25K>10uf(unused)>100R>10uf

B)20uf>2.2K>20uf>15K>10uf(unused)>10K>10uf


These values are somewhat arbitrary. I’m not wondering if having more even values between node 3 and 4, or one higher and lower affects filtering. I’d probably need to change values to get voltages correct at V1.

[jordan86]

You could also run 2 caps in parallel, and do 30/20/10, 20/30/10, or 20/20/20.

Yeah, I considered 20/20/20. That's a good suggestion. Wondered if that would tighten up the preamp too much though. Mostly just curious if say 20/20/20 filtering would be the same as 20/20/10*/10 in terms of noise filtering, but leave stage 1 a tad looser like stock.

*when I say "unused" I mean, in the filter path but not tied to any gain stages.

[AlNewman]

I don't see a maximum capacitance on the datasheet, I see a typical operating point of 10uf.

6x4 can be a problematic tube, and it would likely be beneficial to add safety diodes at the input of the plates.
You could use the 10uf cap for the reservoir, if you're worried.

[jordan86]

You could also use the additional stage of filtering before the node that the OT connects to. Less ripple at every stage.

I like this idea. Though I would like to keep 20 on plates and screen grids. Would it be to weird to do 10uf>100R before the OT. Then do 20/20/10 with stock resistor values?

The 6X4 might like that too?

[tubeswell]

You could also use the additional stage of filtering before the node that the OT connects to. Less ripple at every stage.

I like this idea. Though I would like to keep 20 on plates and screen grids. Would it be to weird to do 10uf>100R before the OT. Then do 20/20/10 with stock resistor values?

The 6X4 might like that too?

Zero benefit in a push-pull amp - all you will do is lower the B+ unnecessarily. Better to increase your filter capacitance at the OT supply and/or screen supply nodes.

You can work out the freq shelving point for R/C filtering thus:

Freq = 1/(2pi.R.C)
where;
2pi = 44/7
R is resistance (in Ohms)
C is capacitance (in Farads) - note 1 Farad = a million microfarads, so 20uF would be 0.00002F, 2uF would be 0.000002F when doing this calculation etc)

If you do this calculation for each node in the power rail, you'll see that there's hardly any Hz difference in going from 10uF to (say) 20uF or even 30uF (because the frequency shelving point is a factor of capacitance, resistance and the phasor/radian equation.

[jordan86]

Zero benefit in a push-pull amp - all you will do is lower the B+ unnecessarily. Better to increase your filter capacitance at the OT supply and/or screen supply nodes.

Hmmm. K. Then back to original idea. Would adding a stage between screens and preamp help? I’m hesitant to add more capacitance to the first reservoir cap but maybe I’m misguided in that.

[tubeswell]

To get it quieter/stiffer, I'd just slug a 47uF in parallel with the screen node and leave the reservoir capacitance alone

[ac427v]

The extra filter cap gives you the opportunity to fix some problems. But does your amp have a problem?
You already are happy with the noise level. How does the amp sound?
How do present day voltages compare to the schematic? What is the bias on the power tubes?
Let's look at this opportunity in a systematic way