The Immortal Amplifier

[nuke]

How much life is enough life?

I've been using these Kemet long-life electrolytics: PEG124YL2470QE4

They're rated 27,500 hours at 105c operational lifetime. And 10 years of shelf-life at 40c with 0vdc applied. The shelf-life probably goes up amazingly as the temperature is reduced. It is probably overkill, of course.

Perhaps guitar tube amp immorality is better addressed by repairability, combined with the use of robust, commodity-components and construction techniques?

I've got tube amps that are older than me, that are completely serviceable, easy to repair and maintain.

On the other hand, I just picked up a complete, perfectly functioning and clean Peavey Vypyr 30 head, for under $100, just so I can scavenge the head cabinet for a homebrew I'm building. But the amp is largely unrepairable since it is beyond factory inventory support (though Peavey will try) and the components are no longer commercially available.

(and < $100 is a pretty good deal for a decently-sized head cabinet that's ready to use)

[R.G.]

How much life is enough life?

Indeed. I think that most guitar players would not like their amps to last less long, though, no matter how old they are. There is a vocal "older is better" lobby too. That's hard to square with the people who seem to think that replacing even one original-manufacture part with a modern equivalent will subtly damage the amp's tone. I mean, a purely original amp is not going to sound good if it's non-functional, is it?

There is a "too much is not enough" bravado at work as well.

Perhaps guitar tube amp immorality is better addressed by repairability, combined with the use of robust, commodity-components and construction techniques?

The engineer's mantra - good ones at least - is that if you can't make it last forever, make it easy to repair/replace.

I'm with you philosophically on there needing to be sincere efforts to make amps (and -insert your favorite whatever here-) easy and cheap to repair and maintain. Good engineers love that. But the companies that make amps hate it.

I spent some time with a F3%Drr design guy at a trade show once. He privately admitted to the Company culture holding that the biggest competitor they have in amp sales is not other companies - it's the company's own amps from decades ago, carefully kept running by decades worth of attention from techs. Every vintage original is a new amp not sold. It's an MBA-disease kind of opinion, but business guys - well, they're business guys.

Then there is the whole issue of what I think of as "component genetic drift". It may not be possible to get robust, commodity components for decades. The market nearly lost tubes, carbon comp resistors, tube sockets, and others. There may be robust, commodity components, but not the same as the original; for instance, we may some day need to use metal oxide resistors instead of carbon comp. High voltage electrolytics nearly became rare items; twist lock cans died and had to be revived; switching power supplies came along and needed high voltage caps for rectified line, so now they're cheap. We ride on the economic back of the bigger electronics industry.

[nuke]

I spent some time with a F3%Drr design guy at a trade show once. He privately admitted to the Company culture holding that the biggest competitor they have in amp sales is not other companies - it's the company's own amps from decades ago, carefully kept running by decades worth of attention from techs. Every vintage original is a new amp not sold. It's an MBA-disease kind of opinion, but business guys - well, they're business guys.

Then there is the whole issue of what I think of as "component genetic drift". It may not be possible to get robust, commodity components for decades. The market nearly lost tubes, carbon comp resistors, tube sockets, and others. There may be robust, commodity components, but not the same as the original; for instance, we may some day need to use metal oxide resistors instead of carbon comp. High voltage electrolytics nearly became rare items; twist lock cans died and had to be revived; switching power supplies came along and needed high voltage caps for rectified line, so now they're cheap. We ride on the economic back of the bigger electronics industry.

Don't forget the EV industry, something I've had an engineering role in. (yeah, software side, but I'm still an EE at heart....)

The whole "circuit boards and surface mount can't do high-voltage" is poo. I've seen plenty of 900vdc power systems (yeah, real "he-man" 900v electricity, not no girly-man electricity), the great majority of which is surface mount. But EV have contributed to the availability of high voltage capacitors and resistors (don't forget those!) and some pretty groovin high voltage components for power supplies and what not that could be a lot of fun to play with.

Fender's "we're competing with our old-selves" thing has been around a long time. I heard it back in the early 80's working a music shop fixing amps. Fender rep liked to say that, mostly about guitars. I think in response to the store owner griping about the 1979 Strats still hanging on the wall with their 10-lb bodies and 3-bolt twist-and-flex necks.

I keep going back to my old tube amps, because they're good, and simple. I have a ToneMaster Pro board and matching FRFR speaker. I even have a Blues Junior IV, that I bought at a good deal price, in like-new condition, with a broken input jack. I fixed that, and stuck a Celestion Neo in it to make it lighter. It's light, loud and kind-of-a POS in many ways, but it sounds good-enough, and I don't mind carrying it around. It weighs about half of my 79 Deluxe Reverb that I've had since I was a teenager.

Fender don't have to fear their old gear. It's a great tool to sell new gear. Pretty much like all big companies, they're shooting themselves in the foot and blaming everything else.

[solderhead]

I like the oil can/film motor cap idea, but the problem is that they get to be huge. In a tight chassis they just can't compete with the high storage density of today's electrolytic caps, which have gotten quite small. Then there's the problem of sourcing motor run caps with adequate voltage ratings. High voltage motor run caps aren't all that easy to find, IME.

[LOUDthud]

Aren't some of the motor run caps rated for AC Voltage ? This creates confusion when trying to select caps for DC operation in a guitar amp.

[solderhead]

Indeed, they are rated for AC voltage. Here's a sample from CDE:

https://www.cde.com/capacitors/film-pap ... ing-round/

[R.G.]

In a tight chassis they just can't compete with the high storage density of today's electrolytic caps, which have gotten quite small.

Yes - they are much bigger than electros, although both film and electros have gotten much better with materials from the last couple of decades. I did ... um, autopsies on some 1960s military gear that used only the metal-case films of the time, and the caps ate up about half of the top of the chassis. This is, after all, what caused the stampede to electros in the first place in the 1960s.

The idea isn't that motor run films are a choice for run-of-the-mill amps; instead, the concept is that compared to electros, they last nearly forever. A hard working tube amp with electros will need a re-cap every decade or so, while an all-film-cap amp will need new caps ever century or two.

Then there's the problem of sourcing motor run caps with adequate voltage ratings. High voltage motor run caps aren't all that easy to find, IME.

See https://temcoindustrial.com/electrical/ ... mz3bhIUx-d, and many others. The link is to the first google result I got for "motor run capacitor". Notice that the caps are specified for 370VAc and 440Vac. 370Vac has a peak of 523V, so 523Vdc will not break it over. 440Vac has a peak of 622V, so this rating should be good for DC supplies up in the 550Vdc range.

My experience is that they're cheap (on a relative basis anyway) and widely available. Could be my location in the southern USA. They might be harder to find in other places.

[nuke]

It would be nice if someone told my air conditioner compressor that motor start/run caps last forever.

Went out on a July 4 weekend several years ago. I ran down to Fry's (may they RIP) on a Saturday night after 8pm, and bought a bunch of 600v film caps, soldered in parallel to make a suitable value, balled them up in electrician's tape and got the AC running for a few days till proper caps showed up.

[solderhead]

The idea isn't that motor run films are a choice for run-of-the-mill amps; instead, the concept is that compared to electros, they last nearly forever. A hard working tube amp with electros will need a re-cap every decade or so, while an all-film-cap amp will need new caps ever century or two.

I just don't see the value of going to all the trouble. I'm not going to be around in 50-100 years and I really don't care if the amp breaks when I'm dead.

The problem, at least as I see it, is that the footprint problem is significant enough to preclude the use of motor run films/oil cans in retrofitting an amp that wasn't designed to accommodate them from the beginning. Sure, they can last forever if you design an amp around them, but we can forget about making them fit in a Bassman or a Showman or a Plexi head, they just won't fit. So in some respect then, I think we're having an academic conversation and not one based upon practical application. To me, it's going to take a monumental effort to design a one-off amp that's intended to last forever. My time is better used not re-designing an amp that's not going to go into production, and just fixing a conventional design when it breaks. Service intervals of 10-20 years don't bother me.

Notice that the caps are specified for 370VAc and 440Vac. 370Vac has a peak of 523V, so 523Vdc will not break it over. 440Vac has a peak of 622V, so this rating should be good for DC supplies up in the 550Vdc range.

yeah, there's that sqrt(2) conversion factor when calculating Vpeak from VRMS.

To find caps like these I guess I'd have to start shopping at the local HVAC jobber's supply instead of the electronics store.

The problem though, is that if you want to build an entire B+ rail out of caps like these, they're so big that you're going to need a separate chassis for the power supply. In the big scheme of things I like the idea of bulletproofing the power supply rail, but in light of the footprint problems these ginormous caps just don't seem practical to me -- there's nowhere to put them in a conventionally designed amp.

I remember having a conversation about doing this with Bruce Collins a long, long time ago. He thought I was crazy for considering the long-life cap approach. Instead he just recommended just taking the easy way out by following the existing amp design and recapping the amps every 10-20 years when they needed it. His argument makes sense.

My problem is that given the number of amps I'm sitting on, I've always got a couple that are in the queue to be recapped, and as soon as they get serviced another one jumps into the queue. It's a never ending battle.

It gets worse if you leave amps out of rotation for a while -- it seems that they always start acting up after they've been left unused for too long without forming voltages on them. You know how it works, you switch an amp off when it's fine, then you let it sit, and then when you finally get around to using it again it starts humming. Aargh. I've got a lot of amps that start acting up when they're brought back into rotation after a long rest period.

I've got a Mesa Mk IV that I've brought out of hibernation and now it's humming. It wasn't humming when I put it up, but I haven't powered it on in several years and now it hums. I've been inside of that amp before. There's absolutely no room inside for immortal cap mods. There's not even enough room to service the electrolytics without a full disassembly of the amp because it's got components that are stacked in layers. There's barely enough room in there for the vactrols that are arranged in stacks. I don't see a practical way to solve this problem other than to take-on the drudgery of recapping every decade or so. The problem is that a decade sneaks up on me faster than I would expect, and I hate opening up that amp so much that I'm thinking that I'd rather get rid of it than recap it.

I'm resigned to the fact that amps require periodic maintenance, and that some of them are a real PITA to work on. It's the nature of the beast. Being a PITA to work on is why you really don't want to own an old Boogie or drive a Jag.

[martin manning]

I've got a Mesa Mk IV that I've brought out of hibernation and now it's humming. It wasn't humming when I put it up, but I haven't powered it on in several years and now it hums. I've been inside of that amp before. There's absolutely no room inside for immortal cap mods. There's not even enough room to service the electrolytics without a full disassembly of the amp because it's got components that are stacked in layers. There's barely enough room in there for the vactrols that are arranged in stacks. I don't see a practical way to solve this problem other than to take-on the drudgery of recapping every decade or so. The problem is that a decade sneaks up on me faster than I would expect, and I hate opening up that amp so much that I'm thinking that I'd rather get rid of it than recap it.

Even an AB763 Fender is relatively easy to restore. That, and the music that was made with them have given them another kind of immortality.

[R.G.]

I just don't see the value of going to all the trouble. I'm not going to be around in 50-100 years and I really don't care if the amp breaks when I'm dead.

OK. It's not for you. This was never meant to be for every amp and for everyone. You may remember the auto industry's concept cars. These were a showcase for new ideas. Most, maybe all of them never made it to production, but the concepts in them certainly did.

The problem, at least as I see it, [...] I think we're having an academic conversation. Which is the point: think about amplifiers, how and when they need serviced, and see if there

There isn't a "the problem"; there are several problems with the full set of concepts, as illustrated by the existence principle, and embedded in my comments. That is, if it were no problems with it, or only minor ones, some company would be making them. This indeed makes it into an academic conversation. I think this is a good thing to do.

[...] and not one based upon practical application.

That depends. Parts of the whole concept are IMHO very, very practical.

For instance, the idea of using silicon diodes as backups for a tube rectifier is deadly practical. I think every amp that uses a tube rectifier should do this. It's just too cheap and easy compared to the cost of a power transformer and first filter cap (even electrolytic! ) to not do. I have this in every tube rectified amp I own. The statistics say that your rectifier tube WILL fail, perhaps from wear-out, perhaps from failure of that electro filter cap, perhaps from a shorted output tube. I think many amp owners would prefer the SS rectifiers to replacing a power transformer, as a for-instance.

The concepts overlap and reinforce in some cases. A PT with fused winding sections probably would not die even if a rectifier tube shorted. But using the back-up diodes in addition might save a gig, because with the backup diodes, you might not even notice that the rectifier tube died while you were playing.

Fuses on transformer windings are a bit more of a pain, but still very practical. It's a bit of a pain figuring out where to put the several fuses. But someone who's had to pay for finding or rewinding a vintage [whatever magic amp] power or output transformer might be very happy to replace a fuse instead.

To me, it's going to take a monumental effort to design a one-off amp that's intended to last forever. My time is better used not re-designing an amp that's not going to go into production, and just fixing a conventional design when it breaks. Service intervals of 10-20 years don't bother me.

That's fine for your preferences. As the Brits say, horses for courses; in Latin, "de gustibus non disputandum est"; or more today, "you do you".

To me, the monumental work is already done. Coming up with concepts is a different issue than picking and integrating parts, and it doesn't have to be a one-off amp. A bog-stock Fender or Marshall or [whatever] is very amenable to subbing in film caps, adding diodes, using diodes, and do forth. A couple of hours with the Mouser catalog to pick replacement parts and some real thought about where you'll put those big film caps and you're done. Subbing a film cap into an existing amp design is much less monumental. The big task is picking available parts at reasonable prices and sizes. This can be a grind, but one measure of the practicality of a design is how hard you have to look for parts. If you have to spend a long time searching for something that will work, maybe your design approach is too restrictive.

Speaking of which, the first filter cap is maybe the only one we would have problems with. Box-style film caps have improved since I first came up with the concepts back in the early 2000's. You can now get a 10uF/500V film cap that's 15.0x24.5x31.5 mm, 0.6 x 0.96 x 1.24 inches https://product.tdk.com/system/files/da ... h_718h.pdf in a plastic box style that could easily be silicon-gooped to the chassis.

Then there's that first filter cap again; maybe that's not too bad. As an example, Amazon lists a 60uF + 12.5uF dual motor run cap https://www.amazon.com/Motor-Dual-Capac ... B00C0YT11S that would be a fine replacement for a first filter cap and a screen filter cap all in one. That's just the first one I ran into; there are others at lower prices and mixes of size and price. It's 6" long by about 2.5" diameter, which is big, but not unreasonable to hide under a chassis. This, plus a few 10uF/500V box caps starts to look more practical.

As an aside, I find that designing by looking for what parts are available while I'm picking parts can be very, very effective; picking available parts and reasonable prices during the design is a key engineering skill that most schools don't teach.

To find caps like these I guess I'd have to start shopping at the local HVAC jobber's supply instead of the electronics store.

Maybe. Although I find that HVAC supply houses today have nearly their entire inventory on their web site; I use Supply House for many of my HVAC and plumbing/appliance repair parts, for instance. Then there’s amazon, where I got the example links. Say what you will about amazon’s evil-ness, they have a lot of stuff available, and deliver fast, even if they’re not the cheapest.

The problem though, is that if you want to build an entire B+ rail out of caps like these, they're so big that you're going to need a separate chassis for the power supply. […]
In the big scheme of things I like the idea of bulletproofing the power supply rail, but in light of the footprint problems these ginormous caps just don't seem practical to me -- there's nowhere to put them in a conventionally designed amp.

As I’ve saying, maybe it’s not that bad. That 60uf/12uf cap is very available and although its big, it’s not a disaster. Motor run caps were what was available about 17 years ago when I first write this idea up. Today’s higher power DC power practices have made high voltage films available in plastic box style, and worked HARD to make them small. We ride on the backs of the more general industry.

The metal can things are probably not needed for the dropping string, so the box style films newly available for high voltage DC practice would do fine for a few caps there.

My problem is that given the number of amps I'm sitting on, I've always got a couple that are in the queue to be recapped, and as soon as they get serviced another one jumps into the queue. It's a never ending battle.

It gets worse if you leave amps out of rotation for a while -- it seems that they always start acting up after they've been left unused for too long without forming voltages on them. You know how it works, you switch an amp off when it's fine, then you let it sit, and then when you finally get around to using it again it starts humming. Aargh. I've got a lot of amps that start acting up when they're brought back into rotation after a long rest period.

And this IS the point. If every amp gets a film cap treatment, they never come back into the rotation, at least in our lifetimes. You pay some money, sweat some location tinkering, and that one’s done for a long, long, long time.

I happily admit – it’s not for everyone or every amp. But the concepts, perhaps not as the full suite (the size of the film caps is the big sticking point to you, it seems) are applicable in bits and pieces.

I've got a Mesa Mk IV [...] There's absolutely no room inside for immortal cap mods. ...] The problem is that a decade sneaks up on me faster than I would expect, and I hate opening up that amp so much that I'm thinking that I'd rather get rid of it than recap it.

Yeah, Mesa has always been a special case of special cases. It’s horses for courses – sometimes you can’t do the perfect thing because we live in a universe with only three spatial dimensions.

[solderhead]

Even an AB763 Fender is relatively easy to restore. That, and the music that was made with them have given them another kind of immortality.

The hardest part about restoring an AB763 -- for me anyway -- is that the kinds of caps that are commonly available for PSU uses keep on changing their footprint. The old Fender amps were designed to use axial caps, but axials aren't all that common today and when you can find good ones today they can be expensive or hard to source. I have to end up shopping for them with specialty-interest suppliers.

When axials became hard to find I started sourcing snap-in caps because they were readily available. In the modern era of high voltage devices (think CRT TV) snap-in caps were available in a wide variety of voltage and capacitance ratings and they were pretty easy to adapt under a BF/SF cap doghouse. Now that the industry has moved away from high voltage devices (eg: CRT TV) in favor of low voltage devices (eg: LED TV) we don't have as good a selection of high-V snap-in caps anymore. So we end up looking for whatever we can find that has the right voltage rating, capacitance rating, and size to fit in the amp. It's no surprise that sometimes we end up having to adapt radial caps to the task at hand, but even radials are getting tougher to find in the right voltage and capacitance ratings unless you shop at specialty suppliers.

You've made a good point that those old Fender amps are easy to service and they have earned the status of "immortality" in more than one way. The music that was made with them have given them yet another kind of immortality. That makes it worth the effort to deal with the changing nature of capacitors and figuring out ways to adapt to their service needs.

Getting back to the immortal cap idea, I always choose the longest life and highest temperature rating that I can find when shopping electrolytics. It's always worth considering extra long-life caps if you can fit them, but every time I look inside of an AB763 chassis I see a tight layout that makes the idea of the big film or oil can caps difficult to give serious thought.

Retrofitting an existing amp is going to be a challenge if you have to adapt to huge caps that won't fit in an existing amp's footprint. If a big cap won't fit inside of a chassis and it needs to be placed outside, now you've got electrical safety issues to deal with. The changes are never as straightforward as we'd like them to be. FWIW I like open-back combos and I hate the fully enclosed design of modern amps that have ventilated access panels that close off the back.

If you really want to immortalize a PSU then it makes the most sense to design amps to accommodate "immortal" caps from the beginning of the design stage. The biggest impediment to taking an immortal amp to production would have to be the bean counters. There's always been a never-ending fight between design engineering and production engineering, where the design people try to design the best amp and the production people change it to make it cost-effective to produce. In a world of 1-year warranties it's hard to imagine anyone building an amp to last 100 years, and it's even harder to imagine cash-strapped customers who would be willing to pay for something built to that degree of reliability. It's been my experience that people don't want to spend more than they absolutely have to spend -- and that includes both the people building the amps and the people buying them. This is even a problem with "high end" and "boutique" amps that aren't built as well as they could be built. Have you ever seen how badly designed for serviceability the low-voltage supply is in a Bogner?

[nuke]

Even film caps don't last forever.

I have a small bag of NOS "Astron" the yellow ones with the blue-epoxy end seals. They're mylar and foil (pretty sure mylar, they're some kind of film anyway). However, they've all degraded over time and are not usable at voltage anymore. They are the ones everyone wants to see in a pristine, vintage Fender tweed amp. Worth a bundle, even in this condition. Still kinda function as tone caps in guitars.

But in the 30+ years I've had them, they've degraded. When I got the bag, a few were (electrically) leaky, but most were fine. Now they're all leaky. Seems the film is not chemically stable.

[R.G.]

On a more tech-ish turn, I decided to see how bad a real test case for film caps in an amplifier would be. Like many others, I had a premonition of metal cans coating the whole chassis. Turns out, not so much.

I picked a Bassman 6G6 as a test case, for no reason other than I had the schematic. It uses a 40uF (two 20s in parallel) and three more 20uF caps for the dropper string. Setting up a sim with loading on each section to fake the actual current used, I found that the only cap that even approaches the need for metal cans as used for motor run applications is the first filter cap. That position in the circuit has an AC/ripple current of about 270ma RMS. The second cap, after the choke, has a ripple current of 1.6ma RMS, and the third and fourth caps are 30uA and 1.4uA respectively.

I spent a couple of hours trying to find some manufacturer of motor run caps that would specify a current. None of them did that I found. I suspect that the actual current handling is hidden inside how motor run caps are picked: compute-guess a value that makes the current phase on the capacitor winding be as close to 90 degrees as you can. Near as I can tell, this current is near the motor run current - that is, amperes in many cases. The power generated inside the caps is what heats them up and makes them wear out.

A motor run cap with a rating of ... um, guessing, 2A, then has a self-heating that is proportional to the current squared, without even knowing the ESR. Call it 4Amps-squared. The rectifier rms current in the "victim" circuit is, say, 0.3A, or 0.09 amps-squared. That's a self-heating ratio of 0.09/4 = 0.0225 in the first-filter-cap application - 2.25% of what it would have to stand in a motor run application. My conclusion is that there will be substantially no self heating internally, and a nominal motor run cap would last a very long time, essentially its shelf life. It's vastly over-qualified for the job.

The second through fourth caps in the chain get even more minuscule RMS currents, because the first filter cap and choke are keeping the rectifier filtering away from them. So I checked for how much current a box style cap can carry. Kemet thoughtfully provided info on their high-voltage DC link film caps. As an example, a 20uF/650V cap from this line has a maximum current limit of 15A. 15 AMPERES!?!? I think it can handle the 16ma of the second filter cap. Caps 2, 3 and 4 just need to have enough voltage rating. A 20uF/600v film cap is about 0.8" by 1.6" by 1.6", which is bigger than a similar electro, but not grossly so. Call it 3x to 4x the volume of the electro.

That turns out to be the big issue voltage rating and the concomitant cost. I found 40uF/440VAC motor run caps for as low as $4.16 each, but most were clustered in the $7-$8 range.
Mouser wants ~ $6.50 for an axial 40uF/500V electro, and $3.60 each for axial 20uF/500V electros. Film caps can get very expensive, but the lower cost 20uF/600V film types were in the $6.50 range. (I know full well that some amp-parts specialist might be cheaper; I just use Mouser for apples to apples comparisons.)

So in a 6G6 Bassman type circuit, you'd need one 2" diameter by 4" long metal can cap for the first filter, and you could use 1.6 x 1.6 x 0.6 inch box caps for the other three. They're bigger than electros, but not grossly so.

I'm hacking on a 3d model of this setup for visuals.

[soundmasterg]

Are the Solen Film caps smaller than the motor run film caps? I was going to use some in a Sunn 2000S that I have because the chassis is huge and has the room. I'd hesitate on trying to fit them in a Fender BF/SF chassis though. No way they would fit into a Tweed Fender chassis. Obviously they can be designed into a new amp just fine as Carr uses film caps in all of his amps from what I can tell.

Greg