Correct cap to use here? Bias

[Intergalactic Tourister]

I'm wondering the opinion of others as to what the best type and voltage cap should be used at the red question mark...

There is an old CE 1960s .022 or (.22) 1200v cap there now.

Thanks.

[Bruce / Mission Amps]

Probably more like .22uF to .47uF for that low of impedance to couple AC to a bias supply.

[Gingertube]

I would be useing a cap designed for mains filter duty, that is one with an X2 safety rating. These caps are self healing and guaranteed to not go short or burst into flame. Metalized Polyproylene or Metalized Polyester.
Cheers,
Ian

[Intergalactic Tourister]

Thanks Bruce, it does kinda look more like 0.22... its very hard to read though but wanted to confirm.

I would be useing a cap designed for mains filter duty, that is one with an X2 safety rating.

Initially I thought to put a X1/Y2 but they're only rated at about 250v. I'm not sure why an X2 would be better here. The original cap is a 1200v CD brown drop from the 1960s... and its on the high volt secondary. Seems like safety caps are better suited at the mains rather than the secondary. But I'm just not too sure.

I'm about to make a parts order but I'm stuck on what to get. The X2 is also the only cap that comes close to being the correct value at 0.25uf. But at 250V- Just not sure about that.

Thanks.

[g1]

Why do you want to replace it? I don't think those CD brown drops are prone to leakage or failure are they? If you feel you must, I would suggest .22 at 600V or better. (suggest checking the voltage across the original cap in both standby and run modes).
For myself I would not be too concerned about X or Y safety ratings. You are on the secondary, and not going to the chassis or anything. Guaranteed not to short is nice, but in a bias circuit, isn't it just as important to not go open?

[Intergalactic Tourister]

I don't think those CD brown drops are prone to leakage or failure are they?

They are 50 years old... figure I'd replace while I've got the amp open.

I would suggest .22 at 600V or better. (suggest checking the voltage across the original cap in both standby and run modes). For myself I would not be too concerned about X or Y safety ratings. You are on the secondary, and not going to the chassis or anything. Guaranteed not to short is nice, but in a bias circuit, isn't it just as important to not go open?

Thanks. That sounds good.
It is grounded through a 27K resistor. But I guess I'll use a higher voltage orange drop or so.

[J M Fahey]

Funny that cap values were suggested without considering the resistor value.
I wonder why.
That particular circuit is a voltage divider, made out of the cap and the resistor.
If both were resistors we would be talking about the resistance ratio needed to turn, say, 350 VAC into around 50VAC or whatever's needed, but since the upper leg is a capacitor, we must work with the impedance ratio.
The resistor part is straightforward and the capacitor part is the cap impedance at 60Hz (USA/Brazil) or 50 Hz (Europe/Latin America).
I always use such RC dividers to get negative bias voltage out of my bridge rectified single HV secondaries.
Simplifies my winding them a lot.

[Chuck H]

Funny that cap values were suggested without considering the resistor value.
I wonder why.
That particular circuit is a voltage divider, made out of the cap and the resistor.
If both were resistors we would be talking about the resistance ratio needed to turn, say, 350 VAC into around 50VAC or whatever's needed, but since the upper leg is a capacitor, we must work with the impedance ratio.
The resistor part is straightforward and the capacitor part is the cap impedance at 60Hz (USA/Brazil) or 50 Hz (Europe/Latin America).
I always use such RC dividers to get negative bias voltage out of my bridge rectified single HV secondaries.
Simplifies my winding them a lot.

Since the caps knee, and I assume therefor what it's impedance will be @ frequency, is determined by circuit impedance, how do we figure the circuit impedance? It seems like it would be complex due to the involvment of diodes, transformer winding resistance, main filter ESR, etc. Or am I just over thinking this?

[Merlinb]

Since the caps knee, and I assume therefor what it's impedance will be @ frequency, is determined by circuit impedance, how do we figure the circuit impedance? It seems like it would be complex due to the involvment of diodes, transformer winding resistance, main filter ESR, etc. Or am I just over thinking this?

In that situation you just want the biggest cap you can readily get- it's coupling everything, rather than being used to get a particular voltage. 220n to 470n is indeed the usual range, since you don't get much bigger in plastic caps. Should be easy enough to buy a 1kV-rated cap in that range.

[J M Fahey]

It's deceptingly simple.
I also had one of those aha! moments after
1) I thought about it
2) calculated some rough values and
3) IT WORKED !!!!

The method is as follows:
4) decide what AC voltage you need . Say, the classic 50 VAC.
5) check what AC high voltage you have. Say, 350VAC
6) your *actual* load will be the DC resistance of the elements in the bias net, which usually is the bias pot, plus any other resistance to ground you have there.
The (typically 220K) resistors to grids do not count, since grids do not take DC current in normal operation.
One apparent problem is that the bias pot does not have a predefined value, but is set acording to needs.
Really it is not a problem, we just decide which is the maximum bias voltage we'll need (plus some extra just in case) which will correspond to highest pot resistance.
We'll get anything from that one down, what *any* bias circuit does, by the way.
To further protect ourselves from pot tolerance, we put it in parallel with a resistor of similar value, to somewhat absorb differences.
Imagine a 50K pot, we may use a 47K resistor in parallel, we´ll be close to the 27K used in the example.
7) 50V/350V=7
so the capacitor impedance at 60 Hz must be 6X (7-1) 27K= 162K
Now where did I put that f*ckng calculator?
No worry, Mr Google to the rescue:
Opamp Labs Inc <> RFC Calculator <> www.opamplabs.com
there you plug any 2 values (in this case frequency and impedance) and it solves the third (capacitance)
Capacitance: 0.01637uF=0.016uF Use .022uF
Why? your possible adjustment is down so err on the high side, if necessary.

And what about transformer winding resistance, ESR, etc?
They will be on the order of magnitude of, at most, tens of ohms; compared to the 150K ohms needed= nothing.
And what about the diode, capacitor, etc?
Same considerations: they will matter for the few seconds while the capacitor charges up; a shorter time than warm up anyway; after that it's a "static" load, meaning it does not change.
*If* we had to get , say, +/-15V to feed a preamp (I have done that too) calculate the needed current and try to minimize it to avoid using a big expensive capacitor.
I love TL062 because they "eat" 1/5 or 1/10 of what a TL072 does.
As you see, I'm cheap and lazy so I am all day long imagining these kinds of solutions. Oh well.

[g1]

220n to 470n is indeed the usual range

Capacitance: 0.01637uF=0.016uF Use .022uF

So we're full circle back to the .22 vs .022 question?

[J M Fahey]

Well, I provided the analysis and the Math.
If you still need a commercial example , or feel safer in the answer of: "what would have Jim/Leo done?", here you have 2 examples of bias voltage derived fro the HV winding.
The first, usable *only* if your winding is referred to ground, as in the so called "full wave rectifier" , you can simply use a resistor to both feed and *attenuate* the voltage.
It's not just "coupling everything" since you usually do not need, say, -350V bias but much less.
Here a 220K resistor is used (in the general area of the 160K I calculated in my example)


Here we have the more "universal" case, where you may have either a "full wave rectified" center tapped HV winding *or* an always floating winding, bridge rectified, which is *always* positive respect to ground, so a resistor will not do.
The capacitor they used is .047uF , still in the ballpark of what I calculated and necessary because the waveform (referred to ground) at any of the winding ends is not a sinewave but a funky rectified sinewave so some empirical correction is needed.
The calculation is precise for sinewaves.


In my own amps, to protect myself from different line frequencies and any other possible variations, I aim high (say, -70V raw bias voltage) and, excuse the heresy, (Jim and Leo will rise from their tombs), I use a Zener to stabilize that.
Works like a charm.

[g1]

Thanks JM, going through this thread I had the JCM900's in the back of my head. They also use the .047 as in your second example. Ampeg V4 is also .047.
So I'm wondering if closer inspection will show the cap in the original post to be .022 ?

[J M Fahey]

Try it, without power tubes .
Start small as not to kill the bias supply and increase if needed.
Meaning: start with a .022 and measure bias voltage, I don't know how much you need nor if it's adjustable, I only see a small part of the schematic.
I guess the range from -38 to -52V covers many bases.
If less, add another .022 in parallel.
600V WV should be about the minimum safe, 1200 or 1600 V much safer.
When they mention "250V" "X rated" caps, it really means "approved for use in Power lines *up to* 250VAC " Which includes 220 to 240V European lines.
I *think* that they stand at least 1600V DC , if not more, because HIPOT transformer tests routinely use 3000V and it would be stupid to have Caps standing 1/2 what the PT does, but I'm only guessing here.

[Chuck H]

FWIW... What are the disadvantages to using a capacitor coupled bias supply? I think it was discussed here once, but I can't locate the thread. I actually have a project I'd like to use it on. This assuming that the cap coupled bias is the ONLY way to make a bias supply with a bridge rectifier.