Filter caps over voltage on startup

[trobbins]

Imho, a bleeder is a must for somewhere in the HT line, to cover the situation of all tubes pulled out. So if there have to be bleeders for series capacitor use, then their use for a global HT bleed function and their power loss isn't wasted per se.

Edit: The bleeder has the function of discharging the caps in a timely manner after mains is turned off (not with the function of somehow restraining B+ level).

[Gingertube]

Not necessarily promoting it as a great idea but I have seen HiFi Amps where there are series connected zeners to limit the turn on peak (to earlier stages) down stream of the normal dropping resistors, that is in the preamp or driver section. Once the B+ has dropped to normal the zeners are non-conducting (OFF).
One particular instance was in a Chinese Music Angle 845 SET Amp. Of-course in that instance, with our 250V AC mains here in OZ, there was 915V across 2 off 450V rated 470uF capacitors with 330K share resistors. Those share resistor are at least 3 times too high for that case. It handled the turn on peak on the main B+ with a 555 timer switching a standby relay such that the B+ did not come up till the 845 filament had been on for 45 seconds or so. It lasted about 2 years before blowing up spectacularly, complete with smoke and flames. I think one 845 developed and open filament so there was nothing to hold down B+ once the standby relay kicked in.

I guess what I'm advocating is that the filter caps should be rated to handle full peak voltage such that you can work on the amp safely with all tubes pulled out. That is make the design "bulletproof".

Cheers,
Ian

[jamesmafyew]

Another big no-no that I've seen is where there's a standby switch downstream of the first filter node. Saw this in a boutique 2x6550 bass amp, where the reservoir was a 500V 50uF x 2 can cap wired up as a pi filter. The standby switch was AFTER it. So with a nominal idle B+ of 480V, it meant that the reservoir would see about 580V whenever the amp was in standby. It came to me for repair because it was 'blowing fuses at turn-on,' because that first cap was just straight-up shorted. I told the player to take it back to the builder and ask for either a refund or a revision.

[Rhodesplyr]

Another big no-no that I've seen is where there's a standby switch downstream of the first filter node.

Maybe I'm misunderstanding what you wrote, but isn't this how most Fender amps are built (with series caps for the first stage to avoid exceeding the voltage rating)?

[jamesmafyew]

(with series caps for the first stage to avoid exceeding the voltage rating)?

That right there -- that's fine, obviously. In this case, I'm talking about where the pi filter was not rated for the unloaded standby voltage, only rated to handle the loaded idle B+. It made me pretty mad, because it was an amp made by one of those snooty booteek builders and it had cost a lot of $$$$.

[Mick Bailey]

Mick, for a ss rectified B+ the initial overvoltage on the caps would still be the same wouldn't it? The initial load on the B+ is effectively zero, so the B+ caps will peak charge, and I'd have thought any residual increase in mains side resistance would have minimal effect. If the amp was valve rectified then from what I've observed, the time of over voltage is typically much less, depending on the ramp up timing of diode conduction, which a thermistor may have a little more affect on.

You're right. I hadn't thought it through! A thermistor gives a soft start at switch-on with SS rectification, and as the caps charge and current draw levels off the voltage drop across the thermistor decreases. With a long time delay thermistor it would only be of benefit if the amp was flipped off standby before the caps had fully charged. With the Bogner I mentioned, after a minute on standby there was still an appreciable ramp-up to full power when switched to 'operate' so by that time the caps were still under-charged. The amp has gone back to its owner so I can't take any measurements on that one. I suppose for this particular problem a thermistor (at best) could only be a partial answer.

[R.G.]

I thought a bit about the meta-issues here.

Back in the Golden Age of Tubes, electronic parts were so expensive that anything one could do to avoid using a part was a good idea, even using a higher-rated part. Except for cost-is-no-object units, it did not make sense to use additional parts for protection, and in fact all teaching and thought about design was to use the inherent ratings and overload ability of the basic parts to withstand surges and such.

Today, all parts except transformers and tubes are very cheap on a relative basis. The thoughts underlying design in today's world is to sense the conditions in the unit and selectively limit things to withstand overloads, not to simply endure them, and occasionally die in the attempt.

I'm guessing that most boutique amp makers have little exposure to modern design as a discipline, their education being gleaned from the internet and (the better ones!) from working on actual amps. So their orientation to "design" is to do it the way it was done back in the Golden Age - let the parts take it where they will. And this approach can turn out good-sounding amps. But the good-ness of hand tweaking every part for sonic perfection gets lost when something fails and is replaced in a repair shop with parts from a bin. I suspect that this idea doesn't get much thought, either.

So it doesn't surprise me much that a high priced boutique amp should be deficient in the simple stuff like inrush surges, overvoltages, and so on. It is somewhere between probably and certain that the designer gave zero thought to the conditions we're talking about. Boutique amp designers (as a broad, broad brush; there are certainly exceptions) will design for a given sound and gain/color/distortion characteristic, not for the dull considerations of what happens at power on, standby, tube failure, and so on.

I realize that my "Immortal Amplifier" ideas are reflective of the design world as it is today. That is, if there is a surge or overload condition, or an expected wearout or failure mechanism, use more of the readily available and inexpensive parts we have over a half century later to make the circuits do what we intend, not to use the side-effects of the basic operational parts to endure odd conditions.

To me, it's absolutely natural that if one wanted to limit surge voltages, one could either turn down or off the power input in a smooth, clean way, or add load through some active bleeder for the time that there was an over-voltage. Both analog and digital ways to do this could be whipped up in a jiffy.

[Chuck H]

Agree. And, just to clearly state the second best reason to limit inrush current and voltage peaking, nobody but amp builders and designers ever uses a standby switch properly. Not that the standby switch will stop the inrush problems (since they still happen when the switch is flipped, as noted) but a circuit that automatically managed this parameter would also force users into allowing the filaments to heat properly before HV is applied to the tubes. I'll guess that 90% of non technically aware players just flip the power and standby switches up simultaneously and start blanging away until sound comes out. That, or they just use the power switch and leave the standby switch in the play position at all times. Even after being told many times by someone educated that it shouldn't be done. The only way to stop this behavior is to remove the option to do it. I use adequately rated filter circuits, but I've been considering implementing and automatic soft start system in my designs for this other reason.

[Merlinb]

Not that the standby switch will stop the inrush problems (since they still happen when the switch is flipped, as noted) but a circuit that automatically managed this parameter would also force users into allowing the filaments to heat properly before HV is applied to the tubes.

I don't think this goes far enough. The whole point of allowing the heaters to warm up is so the PSU caps are not subjected to overvoltage when the HT is finally applied.* But even if the standby warm-up is automatic, it would not protect the caps against the circuit being switched on with no tubes installed. You would need something that either prevents the circuit from being powered up without tubes installed, or else have a voltage limiting circuit that works irrespecitve of the tubes (in which case there's no need for the standby function at all).

*Before anyone aks: no, the standby switch doesn't improve tube life.

[jamesmafyew]

I'll guess that 90% of non technically aware players just flip the power and standby switches up simultaneously and start blanging away until sound comes out.

My experience has been just the opposite -- that most players I work with (and even many recording engineers) treat standby switches superstitiously. I don't even think they're all that necessary, but I see lots of users treating them with deep deference. "No, man, I heard you need to let the tubes warm up for like 10 minutes before you bring it out of standby, dude. And then you're supposed to put it on standby for at least like 5 minutes before you shut it off." I hear this stuff all the time.

Fuses, speaker cables, safety grounds, on the other hand..... things with much more bearing on safety/reliability, are disregarded.

But I do build soft-start arrangements into all of my amps.

So their orientation to "design" is to do it the way it was done back in the Golden Age - let the parts take it where they will. And this approach can turn out good-sounding amps. But the good-ness of hand tweaking every part for sonic perfection gets lost when something fails and is replaced in a repair shop with parts from a bin. I suspect that this idea doesn't get much thought, either.

I suspect it gets more thought than we might realize. Because of the internet, more guitar players are looking inside their amps than ever before. I know from experience that no matter how good it sounds, I can't put a diode-clipping stage inside one of my amps, because NO ONE will buy it. However, that exact circuit, when built inside a pedal and placed in front of the amp, can sell very well. I wonder if boutique builders are hamstrung not by ignorance or slavish following of "how Leo did it," but by the realities of a market made up of irrational consumers. These irrational consumers aren't satisfied with a pretty panel and good sound, they want the 'outside of the inside' to be beautiful and vintage-looking. That's one of the central appeals of tubes in the first place: they're big, glass, and they light up.

I'm not trying to excuse the making of unreliable, overpriced crap. All I'm saying is that in the amp that I worked on with the shorted cap, the pi filter was built with a JJ can cap! It would have been not only more reliable but cheaper to build that filter with 4x 350V 100uF discrete parts. Why did the builder use the can cap? Almost certainly because it meant that there would be big cylinder that would visibly stick up inside the amp and have a logo on it that's recognizable by guitar players. His ignorance of the unloaded voltage problem was just an unhappy accident.

My personal standard is that reliability comes first. There's no point in making an amp I build "vintagey" and "collectible" -- because it will never become a revered sound anyway if it fails constantly. Good sound is entirely nullified by unreliability, but not vice versa. So reliability is the foundational priority in design and building. Basic engineering! I build amps for studio backlines in other cities -- they have to work reliably because I can't be constantly taking a bus somewhere in another state to fix the thing. But I can afford to do it this way because building and marketing amps is not my life's work. So while I hate that it happens, I can understand the pressure that full-time boutique builders feel to produce something that's made "the old fashioned way," and would not entirely attribute these kinds of problems to hackery.

[Zozobra]

I'm interested in making things as bullet proof as feasible as I don't want things to break if it can be easily avoided. I think beyond the initial step of people being ignorant to precautions that can be easily taken the next step is how they're going to do it. Most hobbyists, like myself, are limited to what they can do with eyelets ant turrets as etching PCBs is seen a step too far. The difficulty of laying some of these things out with turrets/eyelets is what puts people off!