20/10/10 cap can for Supro 1624

[Tone Meister]

I literally watched them fail incrementally in about 5 minutes' time. I know lots of folks who use those CE cans, but geez, I can't take much of that.

[Leo_Gnardo]

I literally watched them fail incrementally in about 5 minutes' time. I know lots of folks who use those CE cans, but geez, I can't take much of that.

Late 90's CE/Antique had some good multisections, and they were covered with a transparent blue plastic film. I bought 10 or 12, they worked flawlessly, and one continues in my Heathkit W4-M amp modified to test tubes. 14 years on the job daily, no problems.

Since then, CE/Antique offered multisections without the blue film, nothing but trouble. I quit buying them 12+ years ago, and most times I find one in a repaired amp, it's failed. Drat and double drat. For all that cost, one would hope they're good but experience proves otherwise. One of the few dodgy items from CE. When there are can caps, either I install 50+50 uF 500V Ruby cans when appropriate (F+T, JJ also make these) or leave the bad can in place, disconnected, and run individual caps inside the amp.

[Gingertube]

Gingertube: I was warned strongly against using more than 10 uF on the screens in this particular amp because the extra capacitance would cause the amp to "eat tubes." If your Supro was a 1624, perhaps you should verify for yourself that 22 uF isn't too much filtering for its screens.

Tone Meister,
I don't understand this - The screen voltage will be established by the series resistor size and the screen current. The size of the filter cap (within broad bounds) is irrelevant, particularly when electrolytic caps of the vintage of this amp were typically +100% - 50% tolerance anyway. Did the guy warning you give any other detail on why he believed this.

All,
If too much capacitance in the screen supply causes the screen voltage to hold up at switch off time while the anode supply drops will that cause excess screen current? - my view is NO but if wrong it won't be the first or last time.

Any facts / opinions / WAGs

Cheers,
Ian

[Leo_Gnardo]

Tone Meister,
I don't understand this - The screen voltage will be established by the series resistor size and the screen current. The size of the filter cap (within broad bounds) is irrelevant, particularly when electrolytic caps of the vintage of this amp were typically +100% - 50% tolerance anyway. Did the guy warning you give any other detail on why he believed this.

All,
If too much capacitance in the screen supply causes the screen voltage to hold up at switch off time while the anode supply drops will that cause excess screen current? - my view is NO but if wrong it won't be the first or last time.

Any facts / opinions / WAGs

Cheers,
Ian

Sure thing Ian, I'll give 'er a go.

It's an interesting argument. On one hand we're told "it's important to have a well-filtered screen supply to keep hum out of the audio" and now for the first time I've ever seen it, "too much capacitance on the screen grid supply will wreck tubes." Can't see why one particular amp would be more susceptible to this kind of damage than any others. Notice old fashioned 100W Marshalls have the same capacitance on the screen supply as on the main B+. Plenty of old Fenders I've doubled up and more on the screen filter - 47 uF - to suppress hum and I haven't seen these amps return with prematurely failed output tubes.

On one hand we have "don't stress the screens" and on the other, lots of amps that don't seem to be prematurely wrecking their output tubes. What a quandary. What's a mother to do??? Perhaps put a whopper choke between first and second stage B+ filters and reduce the second stage capacitor to accomplish a filter that has enough hum suppression but not enough capacitance to stress the screens? Maybe that's the solution. Caps being cheaper, smaller & lighter than chokes it won't be popular.

For now I'll elect to stick with the "usual solutions." But it does get me thinking along the lines of building an amp with big henries and small microfarads in the screen filter to find out what it sounds like & what kind of screen stress I see. It would take a long time to assess whether this sort of design enhances tube longevity.

[trobbins]

Ian and Leo, I also can't see an issue with the screen cap. With loss of mains, the anode and screen currents both continue and drag down their respective caps. Even if the rate of discharge of the anode voltage is faster, the energy in the screen cap still gets to discharge in to the screen (unless some energy weeps back in to the anode supply). The grid -cathode voltage could fly high if the bias voltage discharge is relatively rapid - but again that just dumps the available energy in to the screen in a different dynamic profile.

But screen power capability is a temperature related limit, and although the peak power overload capability of a screen is a lot less than a plate, I'd be thinking it would cope. The only concern would be if the grid-cathode voltage was forced to zero or positive, and the peak screen current went up to a such a high level that I*I*t overload occurred.

[Gingertube]

Tim,
One more detail for this particular amp - it is cathode (auto) biased.

It is, however, worth thinking about what happens if the screen supply holds up while bias (in a fixed bias amp) and anode supplies collapse. Of-course, in most cases the collapse of the anode supply will drag screen supply down with it. (as you say).

Have we put a stake thru' the heart of this one or is there more to say?

1 hour to "knock off time" for this week. Weekend job is set to work of a Healing "Church Chimes" PA Amplifier (made here in Adelaide). Push Pull 807, Interstage Tranny phase splitter driven by a triode strapped 6J7 and more 6J7 for mic and pickup preamps. Done all the repairs, tubes tested, recap etc. Ready for first power up in 35 years.

Cheers,
Ian

[trobbins]

For this amp I'd be thinking the cathode bias would head to 0V at a lot faster pace than the HT's. The nominal ratio of screen to anode current would initially remain substantial, but then rise substantially as anode voltage falls below the knee region. So sounds like the screen voltage would hold up longer and bleed back in to the anode, and then sag. KT88 data shows peak screen current reducing as screen voltage sags.

Are you aiming for a tubebuster tube busted statement?

I have one driver transformer 807 PP to restore too (but not this weekend), Philips 952, circa 1950 - lets keep in contact on that topic :-)

Ciao, Tim

Edit : oops, the cathode bias won't go to 0V fast, because it is pinned to the anode and screen currents, so basically it will come down in sync with anode current, and hence I'm thinking screen current won't peak too much even if screen voltage is relatively high.

[Tone Meister]

Here is specifically what I was told:

"DO NOT use more than 10 uf on the screens - that sag is necessary to keep the tubes from completely self destructing at their 100% plus bias and only a 1K dropper. A dual 20, with 20 on the screens, will eat those tubes in very short order as the plate current will run away when the amp is 'dimed' - which it inevitably will be, as that's the only reason people love these amps! There's a very practical reason that Valco consistently used very low screen capacitance on all their amps."


I have no evidence supporting or refuting this advice, but this person has a great deal of experience with these amplifiers and I trust his advice enough that I followed it and passed it along to you to verify for yourself. Only time will tell, I suppose. The amp I did is back in the studio after being fitted with 20/500 and 2 x 10/500 F&T axials.

[Enzo]

That sounds mostly like after the fact rationalizing. When Valco was making amps, no one used them running flat out to get distortion. AMps were made with the smallest caps they could get away with for cost reasons. The screen node cap could be a lot smaller than the plate node cap because the ripple was far less at that node.

[Tone Meister]

Well, nothing like a real-world test to see what is and what is not fact. The Supro here has 10 uF on the screens and Gingertube has one with 22 uF on the screens. If the advice I received is accurate, it shouldn't be too awful long before the tubes self destruct in his amp. Otherwise, it's much ado about nothing. But, I did feel compelled to bring it to his attention.

Being the least experienced of either group, I chose the safe route and went with 10 uF because the advice came from a trusted source, plus I was dealing with a bonafide time capsule.

[trobbins]

Screen dissipation resulting from overdrive circumstances should certainly be assessed differently from the 'turn-off' event (which is what I was previously positing on). Perhaps the following is a reasonable view:

If screen voltage cannot slide sufficiently for a gross overdriven scenario (grid-conduction conditions for 50% of the time), then screen average power could exceed spec level. Any increase in cap value will relatively cause an increased power dissipated in the screen during the time that the screen cap is sagging, and cause a slightly higher power dissipation when the cap voltage is rebounding, but would cause no change under steady state conditions, so on average would increase the screen dissipation level somewhat for overdriven conditions.

So if the screen dissipation was close to suspect for starters, then that sounds reasonable that more failures could occur when screen cap is increased.

[fishergary539]

isn't there only one 35/50 electrolytic in the amplifier ? grounding the power tubes?

OK, here's the short version: This morning I installed a 20/20/10/10 @ 475 cap can from CE Distribution and within minutes 3 of the 4 sections had failed miserably. Both 20 uF sections let through more than 5 VAC ripple, and one of the 10 uF sections has nearly 1 VAC. I could literally smell the can as it disintegrated from within.

So ... I fitted a 20uF/475 and two 10uF/475 F&Ts and snugged them in there like they should have been there all along. After all, I distinctly remember telling Enzo that there was plenty of room in the chassis for discrete caps.

Anyway, the complete restoration consisted of just those three power supply caps, two 35/50 electrolytics, one 1k resistor, and one 0.02 polyester cap. The original RCA 12AX7s, dated 1962, tested like new and so did the original RCA 5Y3. We did install a fresh pair of 6973 power tubes.

I believe I am the only person to have ever turned a screw on this amp since it left the factory. It sounds great, with a floor noise so low I can barely hear it. I feel good knowing that not one single part was changed unnecessarily.

[Jazz P Bass]

Not too sure what you are asking.

Yes there is one 35uf/50 V capacitor.

But there are a total of four electrolytic capacitors utilized in the amplifier.

The 20/10/10 configuration is directly after the tube rectifier with there repective dropping resistors.

The 35uf/50Volt capacitor that you mentioned is on the output tube cathode pins.

It is refered to as a cathode bypass capacitor.

Here is the schematic if that will help.

Supro 1624T.pdf

[52 Bill]

There are a total two 35uF 50 volt caps in the amp. Both are cathode bypass caps.

One is for the power tubes and the other is for the 2nd triode of the first 12AX7.

[Jazz P Bass]

oops!

Thanks for the clarification.