A Rant About DC Heater Supplies

[Ronsonic]

In high gain amps it's nice to supply the preamp with DC for the heaters. Helps keep down the noise.

The easiest way to get such a supply is to rectify the 6.3VAC supply and use one or more huge caps to jack the voltage back up. Back when I was doing the ol' ultra-fi we used to consider this a bit of a bodge, since that supply also went to the power tubes and the least hiccup would disrupt or destroy the supply. It also meant that the diodes would see serious instantaneous currents and the whole thing was a bit inefficient, unstable and unlovely.

Well, we have been seeing this same approach in mass market guitar amps for a few years. And we've been seeing problems from it. Gee, what a surprise.

I have now seen the absolute extension of this bad idea. Modern channel switching amp that takes a 6.3VAC supply feeds it to an 8A bridge then uses a 68,000uF cap to jack that up to 7VDC and then adds resistors and a trimmer to create a center point to produce a +/- 3.5VDC supply. This supply feeds the preamp tubes, Digital switching and relays. And surprise, surprise it is failing. The DC voltage was way low ~5V rail to rail, and got lower as the amp and bridge heat up. The preamp full of cold tubes is noisy, relays start to chatter, the amp fails to pass signal eventually.

Pull the preamp tubes and the voltage pulls up to spec. The bridge measures normal in circuit and it measures good out of circuit (passes DMM test and curve trace).

(How dodgy is this design? It uses paralleled wires and conx to get from the xformer to the PS PCB. Yeah, instead of using adequate wire and adequate connectors, they used more of them. Where the hell did these people come from that they consider that acceptable practice. It's one thing to double up for security entirely another to do it when one isn't enough.)

Call the manufacturer (not the name in big letters on the amp) and you get some hand waving about power tube heater to cathode leakage. Which may well be true in some instances, this circuit lives on a high wire so maybe that would be enough to kill it. Pulling power tubes had no effect on the unit on my bench.

What did fix the one on my bench was replacing the seemingly good rectifier bridge with one heavily derated. That was after replacing the fuse and securing/cleaning conx and reflowing everything.

What I don't get is that for a clean sheet design, why anyone would use this circuit. Back with the hifi amps we did it because we were modding and bodging and building for fun with existing components. Why the hell, when you've got to spec and order a transformer and design a power supply anyway wouldn't you start from the ground up.

If you need a DC supply, design a damn DC supply and be done and have a reliable, cool-running product. Bodging an AC supply when it isn't necessary doesn't even make sense from a cost POV. Not if warranty costs and reputation have any value.

If any manufacturers read this stuff consider it free advice from the field. Yeah, I know, I'm a tech and tech's always overbuild compared to engineers. There's a reason for that.

</Rant>

[R.G.]

If any manufacturers read this stuff consider it free advice from the field. Yeah, I know, I'm a tech and tech's always overbuild compared to engineers. There's a reason for that.

There is, but it's not the reason you think, and there are at least three of them.

(1) Competent engineers who speak English almost never design power supplies these days. If you're a good EE, you can make a lot more bucks and respect doing something else in the EE spectrum of stuff, English speaking or not. You don't win recognition for your power supplies, no matter how good and reliable they are. This has been true and accelerating since at least 1980, and probably before. Designing power supplies is one of those jobs that you have to be smart enough to do but dumb enough to take. That's why there are so bloody few people who have a clue about it.
(2) MBAs design power supplies, in that even if a competent engineer is allowed to touch the design, the MBA will say "No, you can't change the power transformer, the PCBs, or the parts list. It costs too much. You can make it work with some diodes and a capacitor."
(3) Techs generally don't have a whole stratum of MBAs sitting above them reviewing every penny they "waste" on things that won't do anything except cut down on off-warranty repair parts profit. They also don't usually live in a corporate design pressure cooker in general.

I get tired of "the engineers are the problem" getting tossed around. I've spent time in the pressure cooker, and I know what happens in there. Putting a label other than "the engineers" on the problem would be a whole lot more accurate. It makes as much sense to blame customers who'll buy something else if a $1000 amp's price increases by $20.

Sorry. I feel a little better now and I'll try to play nicer in the future. I'm writing "Play nice." on the black board 100 times as we speak.

[diagrammatiks]

this seems as good a place as any to ask a question...

all things being equal in terms of iron...

from a technical standpoint is it more efficient/better regulated to run the multiple necessary secondary taps for a dc supply from 1 transformer or an entirely separate transformer.

[jrfrond]

I have to say that I am not all jazzed on DC supplies either. Perhaps only for preamp tubes, and that's it. I am ESPECIALLY not fond of multi-pronged schemes like this to save $$$, at the expense of reliability. For those of you who worked on VCR's, it smacks of the cheap move by Panasonic when they went to their hellish G-Chassis design which used a single motor to drive the cassette housing, load mechanism and reel idler, by using a system of levers and cams to transfer motion. If something was the least bit askew, the mechanism would drop out of alignment, even by so much as one gear tooth, and not work at all. Real piece o'sh*t.

Hey Mike, how about a clue with this manufacturer you are speaking of? Or PM me.

[R.G.]

from a technical standpoint is it more efficient/better regulated to run the multiple necessary secondary taps for a dc supply from 1 transformer or an entirely separate transformer.

A single transformer. Sheer size of the transformer usually has the side effect of less resistive loss and sagging, which is what transformer guys call "regulation". It is also often more cost efficient to make one transformer of X volt-amps output than two which have outputs which add up to X.

[Gtr_tech]

Now *Thats* why I never got into the engineering side of things. From be'n a tech and see'n things from that side of the fence, there's a lot of stuff you come across that just smacks of bean counter interference. The guy doing the design may not have anything to say about it as long as it "works". Save a buck or 2 per unit....sure that may look like a significant savings to the bean counters.....until the warranty claims start stacking up. Pay now, or pay more later.....

[hithisishal]

So what do you think is the ideal way to build it? I imagine the cleanest would be to string heaters together in series, assuming you have enough tubes of the same current. For example, 4x12ax7 and 2 or 4 power tubes could each get their own lower current, higher voltage tap on the PT. Would you have problems floating the heater? Or could you reference the center of the heater chain to ground? I guess the problem with this is that each model would require a different PT, if it has a different tube configuration.

Or would you just opt for a bit more throw-away voltage such that less capacitance is needed for filtering?

[R.G.]

So what do you think is the ideal way to build it?

Me? I'd add up all the preamp tubes I wanted on DC, the build a DC supply for those heaters only. A 12A*7 heater is 150ma on 12V, 300ma on 6V. If you had eight of those, you could do a nice 2.4A 6Vdc supply with much less trouble than doing 8-10A of DC for all the heaters. The output tubes don't need DC heaters. The first couple of input tubes can profit from them, as could a reverb recovery tube. The bigger the signal already is, the less you need DC. Divide and conquer.

If you were careful, you could get it to three tubes on DC heaters and use one 7806 regulator because it's under 1A. Beyond that, I'd use two 7806 regulators. These are remarkably handy to apply if you watch the minimum input voltage and provide enough heat dissipation.

IF I was doing this for a production amp, I'd include a separate winding designed for such a DC supply. In doing this, I'd make it a bit higher than the nominal DC needed and regulate the DC. This can save you as much in filter caps as the regulator costs and make the DC heater supply very quiet indeed, although just going to DC is probably massively quieter already. If I was doing it as an amp mod, I'd put in a small toroid to run just the DC heaters.

One problem with most DC heater supplies is that they try to do 100% of the heaters on DC, and that's a huge current. It's made worse by the fact that rectifying AC to DC causes the current to be drawn in huge peaks, much higher than the average DC out, and more than the AC the winding would have provided to heaters on AC. A FWR produces an RMS/heating current in the transformer winding of 1.6 to 1.8 times the average DC, which is what the AC would have been on AC heaters. In addition, 6.3Vac is NOT a high enough AC voltage to do a good job of supplying 6VDC. The current peaks make it much worse even. So the heater windings get much hotter providing DC equal to the old AC RMS current.

All this is pretty simple, but as I said, competent engineers will never admit they know anything about power supplies for fear of getting sucked in, and the MBAs won't let the ones who are left do it right. The right technical answer is deemed to be the wrong economic answer.

Would you have problems floating the heater?

Heaters are insulated from the cathode by about 300V in 12A*7 tubes and many others. One problem is that mad genius guitar amp modders would immediately start subbing in tubes and get the equal-heater-currents messed up.

Or could you reference the center of the heater chain to ground?

It just has to be tied to ground *somewhere*. If this is to be DC, I'd ground the negative end.

I guess the problem with this is that each model would require a different PT, if it has a different tube configuration.

Yep. That's another thing that a regulator takes care of.

Or would you just opt for a bit more throw-away voltage such that less capacitance is needed for filtering?

Yep.

[R.G.]

...there's a lot of stuff you come across that just smacks of bean counter interference.

Smacks, nothing. They use baseball bats and blackjacks.

The guy doing the design may not have anything to say about it as long as it "works". Save a buck or 2 per unit....sure that may look like a significant savings to the bean counters.....until the warranty claims start stacking up. Pay now, or pay more later.....

I'd always ask ugly questions of the guys I managed - like "what's the combined Intrinsic Failure Rate (IFR) for that set of components, and what's the MTTF for the assembly, including projected operating temperature effects? Well, why HAVEN'T you estimated the operating temps? And while you're at it, what is the cost of the field stock for the replacement boards? Do we save money by making two smaller, easier to replace units or one big one with no connectors? Ask the field cost accounting department to estimate it - at least that's one thing they ought to be doing."

People always hated it when I did that.

[Prairie Dawg]

Kinda reminds me of the X frame Chevrolet or maybe the Pinto....sheesh. That flopping sound you hear is Leo Fender rolling over in his grave at about 700 RPM.

[Ronsonic]

I get tired of "the engineers are the problem" getting tossed around. I've spent time in the pressure cooker, and I know what happens in there. Putting a label other than "the engineers" on the problem would be a whole lot more accurate. It makes as much sense to blame customers who'll buy something else if a $1000 amp's price increases by $20.

Oh, I don't "blame" engineers. Not in the way you seem to think. My dad was one, and I hang with lots of 'em.

In general they exist because they CAN get things done cheaper, smaller, lighter than the next guy. Nothing to blame there. That's a valuable skill. The profession earns its keep with a slide rule in one hand and an adding machine under the other. Hell, during the depression some of those guys stayed employed working on a commission basis of labor, costs and materials saved. Nothing blameworthy there.

And you're right there's no bragging rights in power supply design, it's considered a grudgingly necessary expense and relatively unloved even at that. The gripe in this case is a pretty specific unit with a fancy "boutique" designer's name mass produced in China by a big American brand name.

Techs in general, are inefficient and overbuild every damn thing. It's the way we are.

And yes. We should blame the bean counters and marketing and the buyers whom marketing presume to understand. If there's a general criticism I'd make of EE's these days is the more recent generation's almost touching faith in published specs. OTOH, this is also the generation that is making those specs more meaningful and are accomplishing things that were pure voodoo only a few years ago.

[diagrammatiks]

dude stop copying me.

It seems like small linear dc supplies for 6, 12/-12, whatever is something that a engineer could whip up in their sleep...and something necessary to have a couple of at all times.

it's not that scary though...the little pcb can go next to the turret board.

the cooling needs and the size of capacitance is a little annoying though.

also...

op spill the builder!

[Ronsonic]

So what do you think is the ideal way to build it?

I would add a separate winding for the DC heaters and whatever else is needed from that supply. That solves what I see as the two big problems with this design.

The inevitable power tube hiccup won't take out your DC supply.

You can use a more sensible design without the huge caps to jack the voltage up.

Even if you just design for 6.3VDC without regulation, you can then add a low drop regulator of some sort to get a 5V supply for your nowadays-essential switching & digital stuff. Or design for more and regulate both the heaters and your digital stuff. Say do a 12.6DC and regulate for 10 and 5 for USB, MIDI and other stuff.

Anyway there are options.

[NorCalTuna]

I think this is where guys like dumble got, and where I feel myself. A few years back I really wanted to see my nameplate on a national stage, but after a few years and a few NAMMs, I can't figure out how to put the amp(s) I want to build- and do build- out at a price that can be profitable above a one-guy-at-home (or maybe two) production level without seriously cutting down some things. FWIW, I have always had as much luck floating the center tap with, say, 35v, as I've had with DC heaters, which 6.3 isn't really enough for anyway without silly regulation. To me, the heater tap is for heaters and thats that, I install tiny 5v psus externally for 5v etc.

HAD likely didn't give a damn about johnny bedroom, didn't have to try to swing endorsers to convince people to buy amps with that name on them (regardless of whats being played or what guts are inside the thing) and I'm not sure he was wrong. Yah, commercial amps today aren't going to last 40+ years (or 70ish, like a rick I repaired once) much less sound as sweet 59 as the bman I played today (thanks RB!) but here's the rub. Yah, MBAs can be jerks, like anybody, and many recently are very interested in blasting out stuff from the mainland thats almost unspeakable, not even worth the materials used for construction, much less the labor, energy and shipping. And RG is partially right that customers shouldn't be so f'in cheap, as they have no clue whats in the box, but 1000 seems to be a breaking point, which also is about the point of min. prac. profit for a HB amp, as in, this is my job, it pays my rent and my kids go to the dentist.

This is where my cali yankee falls on the floor. A deluxe reverb was 250 bucks in 1967. An amp which with proper care was made so well it could have been used frequently until this day, with only minor work. That 250 is 1614 in 2010 dollars, and a p-ton R was 1100 ish, so, yah, a good amp has always been about 1000-1500 bucks. Since the dawn of f'in time. I chose fender because it was prod line but is essentially the parent of this space and it's reliable, and it was also mid priced, iirc, compared to the magnatone and standel (IIRC!)

SO, why is that that today that 1000 gets you a pretty poopy amp compared to that fender? I mean, a DRRI is like list 1000 and are very prone to failure. The HRD series is kind of a disaster, longevity wise... The MBAs are again to blame. Not only are they skrimping on amp parts and sweetness, but they're skrimping-or should i say grabbing- everywhere. Take a look at income charts, corrected for inflation. Median household income is WAYYY down from where it should be, and the top is just rolling in it- no wonder PRS started making amps- so occasionally a hook drops down and an amp goes up to some dude with some loot, but the common man 1 doesn't have time to play 2 has a lot of other lamer things to do these days than play or listen to music and 3 sure as hell doesn't have 1500 for an amp. thanks Jeff Immelt!

.... not only are dc heaters a poor solution to a problem which shouldn't exist- the 5150 has ac heaters and hisses but does not hum, so theres no excuse-, the same people forcing them into the products are likely the same people forcing customers to settle with them and simultaneously making sure fine, craft building is relegated to a dusty, barely or unprofitable corner.

[diagrammatiks]

I'm not sure what you are saying in your post...

Dumble isn't exactly the go to guy for up to date power supply control...

When I first looked up his schematics I was looking left and right for his precision power supply before I realized it was just what he called his series caps with voltage dividers.


Of course, there are many applications where you don't need DC heaters. Classic Fender designs with a couple of gain stages inside a huge chassis, no problem.

For a modern high gain switcher, with multiple eqs, relays, channels, loops, etc...it would be silly to take the option off the table.

A well made dc regulator costs about 10 dollars in parts if you're perfing it or etching the board yourself.

You also need to remember that it's not just the power supply that's being penny-pinched.

I've got a blackface bassman 100 output transformer in my ods clone, with cloth covered leads. It's responsible for 90 percent of the weight of the amp.

Not a single one of the parts in a DRRI is as good as what they had off the shelf years ago...

except maybe the protection diodes they decided to sprinkle all over the place.