B+ capacitor question

[Johntb]

The amp I am building calls for B+ 2-5 to be 30 uf each.

Can those 30 uf be replaced with a 20 uf? If so what types of affects could be expected? Godd or bad.

I have 33 uf axials I can use but I want to save space.

B+ 2 is for power tubes
B+ 3 is phase inverter, loop and 1 of the preamp tubes
B+ 4 & 5 are the other preamp tubes

Thanks

John

[92Volts]

This would reduce the amount of power supply filtering, allowing more power supply hum/buzz to reach your tubes. 20uf vs 30uf isn't a huge change but when you add this change across multiple stages of filtering, it could mean a noticeable amount more noise at B+ 4/5 which are supposed to get the most filtered power (because noise introduced near the input will be amplified by the circuit).

If you increase the values of the resistors that separate the B+ stages this will improve noise filtering. If you divide cap value by 1.5 I believe multiplying the resistor values by 1.5 would bring noise filtering back to what it would have been on the stock circuit. The downside of doing that is that the resistors burn off some B+ voltage and you now have less voltage remaining, especially at B+4 and B+5. This could cause distortion earlier (at lower volumes) or other changes to tone, but in my opinion this is probably fine/not noticeable in most circuits.

Be careful if you increase these resistor values that they are rated for enough power dissipation, which will be higher than the stock values would handle, and might require a larger size of resistor.

[Willabe]

The amp I am building calls for B+ 2-5 to be 30 uf each.

Can those 30 uf be replaced with a 20 uf? If so what types of affects could be expected? Good or bad.

Really hard to for anyone to say without knowing what amp and seeing a schematic.

Can you post a schematic?

[tubeswell]

The chosen cap value practically depends on the value of each dropping resistor in the PS rail. Each resistor (R) and filter cap (C)  forms a low-pass (R/C) filter, and the goal is to make the pass frequency as low as possible, in order to shunt every last bit of noise out of the power rail to ground. If you aim for 1Hz rolloff, that is about the best you can do. But anything below 10Hz is also great.

In 99% of geetar amp circuits, the screen supply (smoothing) cap node is the first R/C node in the power rail (after the reservoir cap), so this is where we start looking at the efficiency of the filter caps at removing ripple current. Taking your typical 1k Princeton Reverb dropping resistor going to the screen node, if you had a 30uF cap vs a 20uF cap with this 1k, you can work out the pass frequency thus: f = 1 / (2 pi R C) . (C is in Farads, and 1uF is one millionth of a Farad i.e. 0.000001F)

For 20uF; 1/(44/7 x 1000 x 0.00002) = 7.95Hz
For 30uF; 1/(44/7 x 1000 x 0.00003) = 5.3Hz

30uF is marginally  better than 20uF, but in a geetar amp, you probably won't hear the difference amongst everything else the affects the signal.

As resistance increases, the amount of capacitance you need to get to sth same point decreases.
At the next supply node in the PR circuit, you have an 18k dropping resistor, so:

For 20uF; 1/(44/7 x 18000 x 0.00002) = 0.44Hz
For 30uF; 1/(44/7 x 18000 x 0.00003) = 0.29Hz

So - even less of a difference.

And so on...

In every case here, there is going to be no practical difference between 20uF and 30uF. But if you had 50uF vs 20uf, the frequency shelf would be halved (for the same resistance), although after the screen supply node, the practical result of this is hardly any difference in actual freq rolloff.

[Johntb]

Thank you for all the information. It is very helpful.