Regulated Voltage Supplies

[bob p]

well, I guess there's a good reason that high voltage, high current bench supplies aren't a dime a dozen. link to Lambda.

For single channel use I had originally looked at the Hammond 278CX, which is 400-0-400@465 mA with a 6.3V CT@6 A. Its the only Hammond plate and filament transformer that even comes close, but the secondary voltages are a bit on the anemic side. Even with a big Pi filter and choke, in a fixed bias 100W mono application the voltage regulation is just awful.

The poor regulation with cap input filters led me to look at using choke input filters with the Hammond 700 series plate transformers. They in turn, required supplementary 6.3V CT filament transformers, like the Hammond 167Q6 and 167S6. I had been modeling the 720 and 722 with 193 series chokes, but in a fixed-bias application the regulation was still pretty bad. That essentially led to this thread on regulated supplies.

Perhaps R.G. was just following my lead on the 700 series Hammonds because I had mentioned them. The only reason that I have been looking at Hammond FW CT transformers instead of purpose built transformers for bridge circuits is because the Hammonds are all that I have been able to find. I agree that more efficient transformer utilization is better (I had considered using the 700 series Hammonds with bridge rectification), but iron availability seems to be a significant limiting factor. If there are better iron options out there, I'm just not aware of them. I'd appreciate any helpful ideas that anyone might have.

[Steve Conner]

There is a guy on fleabay who sells toroidal PTs for tube amp use, amongst other things:

http://stores.ebay.com/Antek-inc

They are listed as, for example, "700V CT" but I think they might actually have two separate 350V windings that you could series or parallel as you wanted.

Another possibility might be a PT out of some old ham radio set or other. People like Yaesu and Swan used to make some pretty hefty transmitters back in the 60s and 70s, that put out up to 300W with paralleled TV sweep tubes. I bought a PT out of a Yaesu FT-560 off a friend for my 300W bass amp project.

[Amp Kat]

How about an old SVT tranny should be more than adequate for that type of application.

[R.G.]

I don't know what they are like in the states, but over here, you can get one that transforms between 240V and 480V with tappings at 380, 400, 415, in various sizes up to 1kVA. I'd be very surprised if that wouldn't power your proposed amp, even if it might look and sound like a small substation

I believe those are autotransformers, no line isolation.

The cheapest thing will be an isolating mains conversion transformer with a couple of 120Vac windings and an isolated 480 winding, perhaps as two isolated 240V windings. That's fairly common, and cheap.

However, those transformers are not built with being inside audio equipment in mind, so they are run at high flux densities and radiate a lot of magnetic field, and have no shielding or banding to stop that.

[bob p]

iron availability is, and always has been, the limiting factor in building high power tube amps. its pretty hard to find off the shelf iron that is suitable for a job like this one. we've discussed lots of options that don't quite seem to fit the bill. so even though the Hammonds are designed for both FW and Bridge use, and a bridge optimized PT would be better, it seems that the Hammonds are still among the best off the shelf options, because:

a) there are models that are capable of handling the job,
b) they're readily available, and
c) they're reasonably priced for what they are.

How about an old SVT tranny should be more than adequate for that type of application.

i don't think that an old SVT transformer would fit the bill, primarily because they're not commonly available at a reasonable cost. if you've got a line on a pair of old SVT trannies that nobody wants, please do tell!

the only other option, I guess, would be a repro SVT tranny. looking at MM iron, there's a 60s SVT PT available, but they ask $400 retail for it. the good news is that I've bought from them at end column pricing in the past, so it would probably cost about $220 -- a bit more than the big .75A Hammonds. I have no idea what sort of specifications the SVT transformer would have, so don't know if that's a good option for us or not.

R.G., if the design criteria for my experimenter's dream DC supply that we discussed earlier complicate the project in terms of design, heat, cost, etc., then let's revise the design criteria to a more realistic objective. Although it would be great to have a benchtop supply that provides 400-600 VDC across a wide range of loads, I can live without it. Let's simplify the design objectives, and go back to basics. I'd be more than happy with two isolated supplies (for twin mono amps) that will provide 560 VDC of regulated voltage within the 100W amp's operating range of 100mA to 375mA.

Now the idea of building a pair of amps around an SVT trannie have me thinking about a pair of BF monoblocks ... "Too much is never enough!"

[R.G.]

I think this is what Steve was referring to; it's what I meant.
SINGLE-PHASE-1KVA transformer on ebay.
It looks OK for the power. You hook up the two 120's in parallel for a primary and take 480 off the secondaries in series. That's 678V nominal, probably 720 peak, might sag a bit, but then also maybe not; it's a pretty healthy chunk of iron.

So for perhaps under $100 you get a ... kinda... shielded power transformer that will make all the B+ you could need up to over 600V. You'll still have to go get filament supplies, but monster 6.3V transformers are easier to find. It's 25 pounds as it sits, and you'll need about another 8-10 pounds of filament transformer.

Frankly, the size of the iron and work needed on the regulators make me think that this is better built as a rollable stand with the power supply on bottom, and a power cable plugging the heater and DC into the amps which sit on top. With output transformers and such, the amps might be liftable without the power supply. It certainly makes it easier.

The rectification and filtering are not all that tough, as 1KV/6A diodes ought to do it. Four of those INXS 1200V/12A diodes would do it. You'll have to heat sink a bank of power MOSFETs to get rid of the spare heat.

I do always tell people that power amps are really power supplies with a thin skin of amplifier draped over them.

[bob p]

I think this is what Steve was referring to; it's what I meant.
SINGLE-PHASE-1KVA transformer on ebay.
It looks OK for the power. You hook up the two 120's in parallel for a primary and take 480 off the secondaries in series. That's 678V nominal, probably 720 peak, might sag a bit, but then also maybe not; it's a pretty healthy chunk of iron.

So for perhaps under $100 you get a ... kinda... shielded power transformer that will make all the B+ you could need up to over 600V. You'll still have to go get filament supplies, but monster 6.3V transformers are easier to find. It's 25 pounds as it sits, and you'll need about another 8-10 pounds of filament transformer.

Here's a link to the page in the Grainger catalog that lists the general purpose transformers from Square D: > link <
The list of available single phase transformers covers a pretty wide range of KVA ratings, from 0.05 up to 50. (!)

I'm about the farthest from a transformer engineer of anyone who's been participating in this thread, so please forgive me when I ask questions that make me look stupid. ... Am i correct in assuming that this sort of regulated supply application would require in the ballpark of about 1 KVA for each pair of 100-W mono amps? (or for a single 100W stereo amp?)

Regarding filament transformers, i had been thinking that the 100W monoblock application would require a 6.3V @ 6A transformer, and the stereo application would require 10A. These are commonly available Hammond items, like the 167Q6 and 167S6. I have no idea what they'd weigh, but the Hammond options go all of the way up to 6.3V CT @ 12A, which should be big enough even for the most obscene SVT-like application.

Frankly, the size of the iron and work needed on the regulators make me think that this is better built as a rollable stand with the power supply on bottom, and a power cable plugging the heater and DC into the amps which sit on top. With output transformers and such, the amps might be liftable without the power supply. It certainly makes it easier.

The rectification and filtering are not all that tough, as 1KV/6A diodes ought to do it. Four of those INXS 1200V/12A diodes would do it. You'll have to heat sink a bank of power MOSFETs to get rid of the spare heat.

I had been thinking that this type of system was going to require some sort of modular setup. When I was first thinking along the lines of the mono amp, it looked like I could shoehorn a 450 mA Hammond 720, the choke and filament transformer, and all of the tubes into a single Hammond 10x17 chassis. If it didn't fit, I was thinking that I'd have to go with a separate power supply chassis, so that the "monoblocks" would each end up being 2 piece units, and the stereo setup would be comprised of either 3 or 4 chassis. Rackmounting the whole setup was the next logical step. A cart or a rack on wheels is an even better idea. Because I don't like the idea of having wires that carry 600VDC dangling in the air, I'm thinking that the system needs to be built into a rack/cabinet enclosure, and/or the power supply wires are going to have to travel from the PSU to the amps through conduit.

When I looked at some of the commercial stuff, like the monstrous Audio Research 200 W monoblocks, I noticed that they're essentially two separate chassis bolted onto opposite sides of a 19-inch rack enclosure. One side of the amp is a vertical power supply chassis, and the other side of the amp is a vertical audio chassis -- with with vertical circuit boards and horizontal tubes. The top and bottom are screen covers, and there's a great big fan in the midddle that sucks air in from the bottom and blows it out through the top.

Another layout that I had considered was a PC Tower type of setup. The transformers would be on the bottom level, the power supply on the next level, and the audio amp would be on a level on top of that. At least if I went that way, I'd be able to do something useful with some old PC boxes that I have lying around... but they'd probably be so heavy that I'd need a hand truck or a dolly to move them.

I do always tell people that power amps are really power supplies with a thin skin of amplifier draped over them.

You're absolutely right. I really didn't have an appreciation for this truth until I started thinking about this project.

[bob p]

if we limit this project to something halfway rational, like a pair of 100-W mono amps, or a 100W stereo amp, is it better to use 700 series Hammonds or something like the 1KVA general purpose transformer?

looking at a 100W stereo amp, I'm guessing that it would require the 1 KVA general purpose transformer, or maybe the 750mA Hammond. If that assumption is correct, the price of going either way looks to be about the same -- just under $200 for new iron. It also looks like it would cost about the same or less to buy a pair of 450mA Hammonds (under $100 each) and build two separate mono amps.

Insofar as the prices aren't significantly different, do any of these approaches have a clear design advantage over the other?

[R.G.]

Am i correct in assuming that this sort of regulated supply application would require in the ballpark of about 1 KVA for each pair of 100-W mono amps? (or for a single 100W stereo amp?)

I was just roughing it out from your numbers. 600V at 750ma is 450W of DC per channel, so you're pushing 1KW if both channels are pumping full power all the time.

That's not strictly true - which is good, because the RMS current from a full wave rectifier circuit is about 1.6 to 1.8 times the DC average out of the filter caps, so you could make a case for a 1kVA tranny being needed for each channel. But music being music, the amps don't live at 100% power all the time, so you can put in about half the theoretical transformer power and get by with it.

In fact, the correct size of a transformer for an AB audio amp is pretty indeterminate. A case can be made for transformer ratings from 1/10 to 2x the rated output power.

So I think you need about 500Va per channel for your 100W channels, 1kVA for a stereo rig. The filament power is whatever it is. Having lumps of iron that can be arranged symmetrically in a power supply bucket is a good idea for balance if you ever have to lift them.

[Steve Conner]

Yes R.G., those transformers are the type I meant. They tend to have more taps over here, and I've not seen them potted in those steel boxes either, but it's the same idea.

How much power do you really need for a hi-fi amp? I guess with tube equipment the main factor is how hot you decide to bias it. If you wanted to deliver 100W in pure Class-A, you have to bias to something like 200W dissipation at idle. By the time you've been through your regulator with its dissipation, you will need 300-400W of continuous DC power per channel, so the 1kVA transformer starts to look reasonable. You also have a nice space heater for your living room in the winter...

This also implies using six KT88/6550 type tubes per channel and about another four per channel in the regulator if you used tubes for that.

If you are happy with a hottish biased Class-AB, I'm sure you could get by with four tubes per channel and a 500 or 750VA plate transformer. Large capacitors on the DC bus before the regulator would probably give all the peak power you need to deal with musical transients. If you use a silicon rectifier, you can pretty much stack on as many inverter grade caps as you like, although these big capacitor banks can explode stuff violently if anything goes wrong. I've seen them throw pieces of transistor over 20ft.

[R.G.]

Like I was saying, it's really hard to define power for AB transfomers. You pretty much have to know what music will be playing. For our government-required testing for commercial equipment, the stuff must reproduce a sine wave output of the rated power continuously for some amount of time without overheating shutting down, and some other conditions. The quirks are all in the interpretation of that "continously" and in what music you play.

For hifi, the crest factor is about 10:1 for peak to average power, so you could make a case that a transformer which would provide more than 1/10 of the rated power continously was adequate. Many hifi makers do that. The other end of the spectrum is the folks who do build power substations to run their hifi gear. These guys want the power supplies to be able to run the equipment with zero sag.

Class A is much easier with constantish power use.

The added quirk I mentioned about rectifier circuits causing RMS currents in the transformer of more than the average DC is real. Depending on the degree of capacitive loading, a full wave rectifier circuit causes an RMS current in the transformer secondary of 1.6 to 1.8 times the DC current out of the filter caps. That eats up the transformer capacity that much faster than a linear load like a light bulb or motor. You're only getting (in round numbers) about half the power out that the transformer sees in internal dissipation.

So the sizing is murky.

Yes, a 100W per channel hifi setup playing typical music for repro will get by fine with a 500W tranformer. It will sag on prolonged peaks, which there may not be, depending on the size of the caps.

For monster, I-made-it-myself bulletproof applications, 2kva is about right and will will power both channels at full power indefinitely if the rest of the power supply is done right.

Back when I was designing power supplies for a living, we were making minicomputer power supplies in the 1-2 kW class. My tech was assembling my latest creation and got the diodes in backwards. This was on a 1KW tranformer and about three 3" x 9" filter cap cans. It threw pieces of diode across the lab and vented the caps vigorously enough to douse the ceiling. He refused to do first power ups for me any more...

[Steve Conner]

So he blew your power supply up by assembling it wrong, and then got scared and refused to blow up anything more? Sounds like a great tech

The sizing thing always worried me, too. I built a solid-state stereo amp ages ago that used a pair of 160VA toroidal transformers, one per channel. When it was done, the maximum unclipped output turned out to be 130W per channel into 8 ohms and 180WPC into 4 ohms.

So once you take efficiency (~60%) and that RMS-to-average thing into account, the transformers are pretty seriously undersized and would burn up if the amp was operated at full power for a long time, even into 8 ohms. It also idles hot at about 25W dissipation per channel.

However, it has run fine for many years now and I still use it in my stereo system. I guess the 10:1 thing that RG mentions really is true. I wouldn't feel happy using it to drive a PA though.

[bob p]

I was just roughing it out from your numbers. 600V at 750ma is 450W of DC per channel, so you're pushing 1KW if both channels are pumping full power all the time.

I guess I should have just expected a straightforward Ohm's Law calculation for the roughed-in numbers. The big question in my mind is how to accurately account for variable power demands, and I see that you've addressed that in a later post. Thanks.

That's not strictly true - which is good, because the RMS current from a full wave rectifier circuit is about 1.6 to 1.8 times the DC average out of the filter caps, so you could make a case for a 1kVA tranny being needed for each channel. But music being music, the amps don't live at 100% power all the time, so you can put in about half the theoretical transformer power and get by with it.

In fact, the correct size of a transformer for an AB audio amp is pretty indeterminate. A case can be made for transformer ratings from 1/10 to 2x the rated output power.

Yes, i guess it would be much easier to just say "Class A" and work around a quasi-constant load. In fact, that turns out to be how a lot of the big names in HiFi built their less expensive amps -- like the Dynaco ST70 for example. They had a pretty anemic PT that sagged a lot, so they used cathode bias and took it close to Class A operation to minimize the effects of sag. Its kind of funny that audiophiles make a big deal about pure Class A operation, when in reality it turns out to be the easy way out in power supply design.

So I think you need about 500Va per channel for your 100W channels, 1kVA for a stereo rig. The filament power is whatever it is. Having lumps of iron that can be arranged symmetrically in a power supply bucket is a good idea for balance if you ever have to lift them.

Yes, that's a great argument in support for symmetrical modular design, big handles, wheels, etc.

[bob p]

Yes R.G., those transformers are the type I meant. They tend to have more taps over here, and I've not seen them potted in those steel boxes either, but it's the same idea.

are those transformers actually potted? or are they just bolted into a metal box?

How much power do you really need for a hi-fi amp? I guess with tube equipment the main factor is how hot you decide to bias it. If you wanted to deliver 100W in pure Class-A, you have to bias to something like 200W dissipation at idle. By the time you've been through your regulator with its dissipation, you will need 300-400W of continuous DC power per channel, so the 1kVA transformer starts to look reasonable. You also have a nice space heater for your living room in the winter...

the space heater thing is something real. i used to develop an open source operating system that was built entirely from source code, and every time that a toolkit change was implemented, the entire OS had to be recompiled. needless to say, running a compiling operation that lasted forever would get your CPU pretty hot -- hot enough that you could feel that the room was hotter.

to cut the compile times down i developed a distributed processing system, where i had the compiling job distributed across an entire gang of computers that i had in my basement. when doing a complete OS build, I had 12 CPUs running full tilt for the better part of a day. the basement got HOT, and after doing this day after day, I noticed a significant bump in my electric bill. it actually doubled. we made lots of jokes about heating the house in the winter with CPU cycles. that project is over for me, and all things considered, i'd rather be heating my house with a big tube amp.

This also implies using six KT88/6550 type tubes per channel and about another four per channel in the regulator if you used tubes for that.

Steve, you've piqued my interest -- did you crunch the numbers for the Class A application yourself? I've been looking at the KT88 and 6550 data sheets, and the ones I'm familiar with don't have tube characteristic graphs, they only have application tables. The application tables are all for Class AB applications, either pentode/tetrode or UL, fixed or cathode bias. I haven't seen any data sheets that list P-P Class A operation data. So I was wondering how you got the numbers. I'm also wondering how you came to the tube count for the regulation circuit. Was that a ballpark guess or did you run through some calculations?

[bob p]

Like I was saying, it's really hard to define power for AB transfomers. You pretty much have to know what music will be playing. For our government-required testing for commercial equipment, the stuff must reproduce a sine wave output of the rated power continuously for some amount of time without overheating shutting down, and some other conditions. The quirks are all in the interpretation of that "continously" and in what music you play.

Regarding the 1:10 crest factor rule, I'm not one of those guys who will try to pass off a 35W RMS amplifier as having 350 watts of "musical" power. To me, that kind of rating enters the realm of the HiFi snake oil salesmen. To me, 100W means 100W RMS continuous sine wave, and load testing a 100W amplifier would involve sweeping clean 100W RMS sine wave signals into a resistive load on the bench for a couple of days before it moves upstairs into my living room. So I guess my bias is to rate power with a continuous sine wave output, like the government would rate something for a military application.