Coupling capacitor voltage rating - Calculating the 'proper' value?

[exclamationmark]

I've always sort of taken this for granted, but how do you actually calculate a reasonably 'safe' voltage rating for coupling capacitors? I usually just stick some 630V caps in there and call it a day. However, due to some space limitations I'm considering downsizing to 400V capacitors. I know that they should be at least rated for B+ due to cold starts, etc, but I've also heard I also have to take into account the DC voltage AND the AC voltage across the capacitor. The worst case scenario for me would be the cold clipper stage running from a 350V B+, with the plate voltage at something like 340V. Since this is only a pre-amp my unloaded B+ voltage doesn't actually increase by much (the heaters draw like 3/4 of the total power), so I would assume I would be covered in the event of a cold start, but how do I take into account the AC rating? Is this even necessary?

Any help would be much appreciated

[Tage]

As far as i know the AC voltages in tube preamps can swing around 1/2 to 2/3 of the HT, so you are still probably safe with the 400v caps.

I built my first amp using some sprague orange drops only to later realize they were from a lower, 400v, line than what everyone else uses. but I never had any issues with it, although the amp was only used for about 2 months befoe being torn apart for other progects...

[Alan0354]

You better look out for the initial B+ voltage before all the tubes warmup and start conducting. Most amps has filtering resistor going from the power stage to the front end stages. The power amp stage usually is +450V or higher, the filtering resistor drop like 50 to even 100V. So you see the spec at the front end is +350V. But that is not the voltage at first cold start. there is no current draw, there is no voltage drop and you see the full 450V at the preamp tube.

AC voltage don't matter, it is not like your load is transformer that it can fly up to double the B+. The voltage at the plate never goes beyond B+ with resistor load like in the front end. But if you drive a transformer, then you have more trouble.

I would not design with 400V cap if it is a product. But if you just tinkering around, use it.

[pdf64]

Cap data sheets sometimes mention a level of overvoltage they are tested to. Plenty of vintage amps seem to have got away with 400V coupling caps, even between the LTP and (fixed bias) power tube grids, which would be be subjected to overvoltage at every power up, but it has to be bad practice to follow that example.
My guess is that modern production methods have greatly tightened up tolerances but that may mean that they have a lower level of safety factor built-in than 50 years ago, so a modern 400V cap may not tolerate the abuse that vintage caps have.
Pete

[Enzo]

Caps are usually rated at "working voltage", which is not as high as their surge voltage. Look at many old can caps and you will even see it in print - 20uf 400WV. WV for working volts. That meant if you had a 400v power supply, this 400v cap was suited to it. It may have a surge rating of 500v.

My opinion is that modern caps would be better able to withstand abuse. Modern materials are better. And parts are more consistent. The days of filter caps marked +80/-20% are behind us.


Good practice is to have caps ready for anything the amp COULD throw at it. A lot of caps will never see the circuit max. Many amps were never fired up without their tubes, for example. And even at whatever the rating might be, caps do not pop like fuses. 400v OK, 401v FIRE!!! Nah. Not like that.

As Pete points out, getting away with something does not make it a good idea.

[Steve A.]

I like to use 600vdc caps but many of the amp mfgs use plenty of 400vdc caps- probably even more common when they use EL84 or 6V6 output tubes. But I think that they do a cost analysis figuring in the odds of the cap failing during the warranty period (too heck with the customer if it is out of warranty when the part fails.) I was indoctrinated into using huge 600vdc 418P Orange Drop caps but the Mallory 150's seem to be just as good and in a much smaller case. I also stock good quality mylar caps from Mouser when I don't need or want to use film and foil caps. I would go with a cap in a smaller case rather than dropping down to 400vdc. One exception might be an amp with very low plate voltages on the initial preamp tube stages...

Steve Ahola

[Wombaticus]

Caps are usually rated at "working voltage", which is not as high as their surge voltage. Look at many old can caps and you will even see it in print - 20uf 400WV. WV for working volts. That meant if you had a 400v power supply, this 400v cap was suited to it. It may have a surge rating of 500v.

My opinion is that modern caps would be better able to withstand abuse. Modern materials are better. And parts are more consistent. The days of filter caps marked +80/-20% are behind us.


Good practice is to have caps ready for anything the amp COULD throw at it. A lot of caps will never see the circuit max. Many amps were never fired up without their tubes, for example. And even at whatever the rating might be, caps do not pop like fuses. 400v OK, 401v FIRE!!! Nah. Not like that.

As Pete points out, getting away with something does not make it a good idea.

Yeah, my suspicion is that modern film caps are pretty darn tough. This is strongly hinted at by many datasheets. For instance, for 400V Orange Drops, the datasheets says something like:

DC Life Test: Capacitors are capable of withstanding a
500 hour life test at + 85°C at 150% of rated working
voltage. After test, capacitance shall not have changed by
more than 5% of initial value, insulation resistance shall
not have decreased by more than 50% of the initial limit
and dissipation factor shall not have increased to more
than 0.1%.

That quite a significant overload, for quite a long period of time, and at quite an extreme temperature. It's quite impressive, actually, when you consider that back in the day the rated useful life at working voltages of an electrolytic cap might only be 2000 hours.

[Steve A.]

Capacitors have many important parameters besides the actual capacitance, especially when used as coupling caps as in the title of this thread. Speaking hypothetically I can imagine the possibility of some caps sounding or responding differently depending on whether they were the 400v or the 600v version of that particular brand and line. With that in mind if you are following a schematic which spells out exactly which capacitors were used you might want to stick with the voltages specified even though they may seem like overkill and are not technically required. (If you are working on your own design by all means use your best judgment.)

Steve Ahola

[Steve Conner]

Coupling capacitors don't have any significant AC voltage across them, otherwise they wouldn't be called coupling capacitors. So yes, I say they should be rated according to the highest DC voltage they'll see, which is the unloaded B+. or in the case of a fixed bias amp, the coupling caps to the power tube grids can see the sum of B+ and bias voltage.

[Steve A.]

Coupling capacitors don't have any significant AC voltage across them, otherwise they wouldn't be called coupling capacitors. So yes, I say they should be rated according to the highest DC voltage they'll see, which is the unloaded B+. or in the case of a fixed bias amp, the coupling caps to the power tube grids can see the sum of B+ and bias voltage.

On the schematics I've seen many modern amps will use 630vdc coupling caps going to the power tube grids and 400vdc everywhere else, presumably to save money and pcb real estate. For me it costs less to stock just the 630vdc caps- I also stock mainly the 1/2W metal film resistors (every standard value from maybe 560 ohms up to 2M2- I also have selected values for 1/4W, 1W and 2W.)
I forget the technical term for "overkill" but a lot of electronic gear uses components rated at least twice the expected voltages and wattages. MTBF is just an average- I'd rather cut my chances of being the 1 in 1000 who gets burned.

[Steve A.]

Coupling capacitors don't have any significant AC voltage across them, otherwise they wouldn't be called coupling capacitors. So yes, I say they should be rated according to the highest DC voltage they'll see, which is the unloaded B+. or in the case of a fixed bias amp, the coupling caps to the power tube grids can see the sum of B+ and bias voltage.

On the schematics I've seen many modern amps will use 630vdc coupling caps going to the power tube grids and 400vdc everywhere else, presumably to save money and pcb real estate. For me it costs less to stock just the 630vdc caps- I also stock mainly the 1/2W metal film resistors (every standard value from maybe 560 ohms up to 2M2- I also have selected values for 1/4W, 1W and 2W.)
I forget the technical term for "overkill" but a lot of electronic gear uses components rated at least twice the expected voltages and wattages. MTBF is just an average- I'd rather cut my chances of being the 1 in 1000 who gets burned.

[frus]

I have Sozo standard caps that are rated for 400V in a 6L6 amp that has some 420V on the plates + (-45V) on grids. it's been working fine for a few years. do you think I should replace them with 630V ones?

[Alan0354]

What is the difference between a cheap ceramic cap vs the so called orange drop, the old blue Fender caps, Mylar and poly propylene caps? What I mean is the sound and the reason why they are different.

In my opinion, I don't believe it's the loss tangent that's the issue as tube amp is high impedance type of circuit, current is small, AND you are talking about low frequency. Even for high frequency, we use ceramic cap to 100MHz. And also it is common to parallel a larger cap with small value cap to overcome loss tangent problem. eg, you parallel 0.1uF with a 10uF. If you want really high frequency, you even parallel a 1000pF or even 100pF. So why people pay so much money for those so called "orange drop" cap?

[pdf64]

Well, I guess high value ceramic caps tend to have a poor temperature coefficient, so may not be suitable for, say, the tone stack, where the turnover frequency of the mids may change a little as the amp chassis warms up.
Also in a combo, microphonics may have an impact.
In classic designs, eg AB763, some coupling caps feed 'medium' impedance circuits, the 0.1uF into 50k trem intensity control, so variations in series resistance, inductance etc between coupling cap types may have a slight effect on the frequency response at the output of that effective potential divider.
Pete

[Steve Conner]

I personally don't believe that different brands of coupling caps make the slightest bit of difference to the sound, except for the well-known distortion caused by voltage coefficient in high-K ceramic caps. I have found real examples of this in a supposedly "pro" audio parametric EQ unit. I heard grungy distortion when using it, and on further investigation, my distortion analyser showed something like 1% THD. I opened it up and found that the capacitors in the filter network were tiny surface mount ceramics.

Now hiding behind the couch with my flameproof underwear on.