DRRI Power Supply Capacitors

[printer2]

Anybody know why the DRRI has two caps in series for the power supply, a 220uF 100V and a 47uF 500V?

[Chuck H]

That's super interesting. The net capacitance remains low for the rectifier but the uf's in play are larger than a different circuit would offer. I wonder if there's an advantage to this? I hope one of the higher minds sees this and reports. Sorry for posting without an answer, but I wanted to speculate a little as well as secure a subscription so I can follow.

[Pedro Vecino]

I think it's a practical way to approach the 35 microfarads starting from one of 47 using standard components.

[g1]

This same question came up awhile back, can't find the post or remember the exact answer, but I think Pedro has it right.
There may be some issue with tube supplier about max. capacitance for the rectifier tube/warranty etc. So how do you get max. allowed capacitance with components currently on hand or readily available? This arrangement was their answer.

[printer2]

This same question came up awhile back, can't find the post or remember the exact answer, but I think Pedro has it right.
There may be some issue with tube supplier about max. capacitance for the rectifier tube/warranty etc. So how do you get max. allowed capacitance with components currently on hand or readily available? This arrangement was their answer.

Yeah that makes sense. Really bugs me when I look at a circuit and can't figure out why a part is there.

[Chuck H]

I dunno... 68uf is a common value (x2 series = 34uf). So is 33uf. Why use a 47uf and a 220uf with staggered balance resistors. I understand that 47uf and 220uf are slightly more common values, but staggered balance resistors? I'm not convinced.

[Enzo]

This is not my area, so just me thinking here...


But I just can;t get my head around them getting all concerned their filter caps were exactly 32uf or whatever. The amp this is supposed to be like certainly had tolerances so loose you could drive a truck through them. The old schematic may have had 32uf on it, but they expected that to range up to as much as 60uf. Or down the other direction too. I can't think they are concerned that a 33uf cap wouldn;t be close enough.


My hypothesis: REctifier tube doesn;t want to see a ton of filter cap. This whole stack can't appear like more than the 47uf bottom guy, and likely less. With B+ at 400v, the split puts about 330 on the lower cap. About 70v on the upper cap. Now here is my crazy idea. MAYBE the basic stack charges up to the B+, but the ripple is just a few volts up top. Maybe the 220uf sits on the shoulders of the 47uf, and smooths ripple like a 220uf, but the rectifier tube sees it like a 35uf. In other words, the 220uf thinks it is smoothing a 70v supply.

Somehow I am reminded of supplies like a 20vDC that feeds into a 7815 for 15v, then feeds THAT into a 7805 for 5v.


I am prepared if I made some glaring error there.

[R.G.]

I've pored over it and can't figure out why a designer would do that deliberately in that circuit. My speculations about what it MIGHT be for follow:

- It wasn't for that circuit; could be a leftover, carefully (or inadvertently, or lazily) preserved from some previous prototype, where it served an unknown function, now removed; I've seen this happen in production computers and heard stories about it in spacecraft design and design of stills.
- It is to tune in a specific amount of ripple. Someone may have had a bee in their bonnet about a magic amount of ripple sounding better, and tuned it in.
- Appended to above, there may have been a shortage of 33uF/450V caps; or a shortage of 68uF/350; these are unlikely as the maker is buying in big enough lots that the manufacturers will go make a batch of caps just for them.
- Maybe it was a scheduling thing; faster to make a marketing deadline with the 220/47 than to wait for 33s or 68s.
- Maybe there was a big stock of leftover caps from a previous manufacturing run, although I can't ponder why they'd do this just to use them up, as a 47 by itself has better filtering than the series string, and is cheaper. Perhaps they designer thought that 450 wasn't enough voltage rating and wanted a little more for insurance, and could not talk the boss/accountants into two 68/350 or 100/450 caps for some reason.
- Maybe there was a PCB fit issue on the cap sizes in combination with one of the above.

All of these come under the general heading of "some reason unrelated to its electronics function", a side effect of something else.

Here's what I think it's not for:
- It's not for the rectifier capacitor rating, unless they got stern warnings from the makers of their present suppliers. The 5AR4 is rated for 60uF as a first filter cap in my old tube handbook.
- It's unlikely to be for a specific cap value, as 33uF is a standard value, as is 68uF.
- It's probably not for a quirk of the 220 filtering part of it better. I would sign on for that if there was any connection to the junction of the two caps, but there's not. Things in series are not sensitive to order unless you "look at" one of the nodes in the middle. If, for instance, the screens were connected to the mid point, that would be a different story. Even then, this is an odd way to do it, as it would funnel more ripple into the screens than a resistor dropper and a separate screen cap to ground.

Bottom line is I can't think of any good electronic design reason for doing that, even a perverse one, which I'm usually good at.

[Chuck H]

I think I'll side with your speculation that it was a prototype anomaly that was incidentally designed into the final circuit. Perhaps when the prototype was being built the designer wanted the extra voltage rating, DID have a final amount of uf's in mind and those just happened to be the caps he had on hand.

I've seen this sort of thing before. I did some design prototype changes to an amp that was going to be made overseas. One thing I did was change the treble cap in the tone stack to 330pf. I didn't have a 330pf cap so I paralleled a 250pf and a 100pf. Once the changes were approved the amp was sent to the manufacturer along with a schematic that clearly showed a single 330pf cap. They then made a "here's what we'll ship" prototype and sent it back for evaluation. Sure enough the amp had a 250pf and a 100pf cap paralleled for the treble cap even though it was contrary to the schematic.

[Enzo]

Just for the forensics then, the circuit is in all the revision levels A through E. The board layout for the earlier revisions are marked 1993 design, while the E rev are on a 2002 board. All with same caps.

I note the 65 Twin Reverb RI has a pair of 220uf stacked.


In general, I tend to put low credence on the "parts they had on hand" stories. The designer might have knocked out a prototype, though I suspect it was mostly cut and paste. But a company like Fender doesn't build amps at the kitchen table out of bags of parts from Mouser. The bill of materials is designed no less than the amp, and parts are ordered from the cap and resistor makers in lots for production. If they need 50,000 22uf 500v axial caps for a production run, they are not out in the back looking to see if they have a few hundred of something they might use up first Looking at the board layout, C32 is three to four times the size of C31 physically, and the cap board is designed for it. I compare to the Twin Reverb board and that one is designed for two caps both large. I have to think that even if it started life as some sort of hangover, when they redesigned the board 9 years later, it would get filtered out. (pun intended) SO whatever it might be, I gotta think this was done purposely at some level.

I am looking through other amps of the era to see if they might have used the same board in something else. No hits so far.

[R.G.]

In general, I tend to put low credence on the "parts they had on hand" stories. The designer might have knocked out a prototype, though I suspect it was mostly cut and paste. But a company like Fender doesn't build amps at the kitchen table out of bags of parts from Mouser. The bill of materials is designed no less than the amp, and parts are ordered from the cap and resistor makers in lots for production. If they need 50,000 22uf 500v axial caps for a production run, they are not out in the back looking to see if they have a few hundred of something they might use up first

Well, yes and no. I spent over three decades in the belly of a computer company most often named only by its three initials. While you'd expect things to always be done in discliplined, good order - well, sometimes things are done by expediency, and sometimes "expediency" is determined by a mid-level accounting manager who will get a poor yearly job rating if his inventory levels of parts in hand are too high compared to the levels set by a bright young MBA at another corporate office who's trying to get noticed for promotion by his manager who's busy ... You get the idea.

For instance, the US Army is big enough and well funded enough to get its way most often, but it only takes a little while talking to a veteran to get stories about semi-irrational things happening.

Yeah, parts-on-hand isn't how it would work mostly, but it could happen. The problem with going back and determining how something like this happened is complicated by the fact that anyone involved will deny it all.

[Leo_Gnardo]

I've pored over it and can't figure out why a designer would do that deliberately in that circuit. - - - snip - - - Bottom line is I can't think of any good electronic design reason for doing that, even a perverse one, which I'm usually good at.

FWIW the Engl "Straight" 100W amp head early 90's also had dissimilar filter cap stack much like the one being discussed.

[R.G.]

That's bizarre. I can understand one side-effects process happening once, but not twice, so it's probably there for a reason - unless Engl was copying the random choice that led to putting it in the Fender. Stranger things than that have happened in the music world.

[Leo_Gnardo]

That's bizarre. I can understand one side-effects process happening once, but not twice, so it's probably there for a reason - unless Engl was copying the random choice that led to putting it in the Fender. Stranger things than that have happened in the music world.

Or Fender was copying Engl . . . or it's just a coincidence. Bizarre indeed the world of amp manufacturing.

[printer2]

Damn it, just when I thought it might make some sense.