Need power supply design help

[Midtex]

Hi everyone - I'm new to this forum - stumbled across it looking for power supply design information. Let me introduce you to my project, please. I was given an old Heathkit W4M hifi amp years ago (push-pull 5881) and I have always wanted to scratch build a Bassman-type guitar amp using the transformers. I like the added gain stage of the Pignose G40V so I am basically building it with that modification. I have the mechanical build completed and I applied power to the rectifier and measured the hv for the first time last night. I have not installed any of the audio components at this time. I believe my B+ is too high and I want to explore methods to lower it. The PT secondaries are advertised as 380v-0-380v. I measure 865vac at the secondaries with no load. I realize the PT was designed for 110v and we now supply 125v which accounts for some increase in output voltage (~10%). Heath used a 5V4G rectifier immediately filtered by a pair of 20uf caps in series each bypassed with a 100K 1W resistor which yields a 10uf and 200K filter network. A 7H, 160 ohm, 160ma choke then followed this first filter section. I tried this first and it yielded 595vdc with no load. I then tried moving the RC filter and going immediately into the choke and then into the 10uf 200K RC filter and I still measure 520vdc with no load. My first question is how much voltage drop can I expect when the power supply is under a load? Second, if the voltage is too high, what methods do I have to lower it? I have the original Tung-Sol 5881's from the Heathkit, but don't know if they are good. I hoped to use them or new 6L6's. I guess another option would be to change to a tube that can handle the higher voltages like EL34 or 6550? One more thing - there is no secondary tap for the bias supply so I will have to take it off of one of the hv legs. Any suggestions on how to lower the voltage sufficiently to get in the right range? I have seen some schematics that have the bias supply go immediately into a series capacitor so I'm thinking that's a good start.  Any suggestions would be very much appreciated.

[sluckey]

My first question is how much voltage drop can I expect when the power supply is under a load? Second, if the voltage is too high, what methods do I have to lower it?

There's not a simple answer to that question. Your PT and OT were chosen to work well with 5881s. Just build it and see where your B+ ends up with all tubes plugged in and biased where you want it biased. Then let's see where you are.

This schematic should get you close to the bias voltage range you need using the same values. Juggle the bias range resistor to get the exact range you want.

[Midtex]

I apreciate that - I will go ahead and wire it up and see where my voltage falls under load and report back. Thanks!

[PRR]

> 380v-0-380v. .... 595vdc with no load. ... how much voltage drop can I expect when the power supply is under a load?

WHAT load?

The stock Heath is Class A cathode bias and loaded in (IIRC) 6K-10K. Since current is fairly steady at around 100mA, the most economic design is "saggy" because sag doen't matter.

The Bassman is loaded in 4K which increases full-roar current. It is biased deep AB, low idle current. So the current may run 50mA-200mA.

5V4 is rated 175mA max DC current and would be marginal in a 4K Bassman. Use 5U4.

5U4 datasheet page 4 shows for 865/2= 430VAC each plate just under 600V, as you observe. ASS-ume at 200mA DC it will sag to 400VAC each plate, the curve shows 433V DC which is entirely reasonable for 5881/6L6GC in a bassman. At 50mA idle it may rise to 520V DC, which is "over spec" for 5881/6L6GC, but all decent 6L6GC will stand this fine in guitar-amp duty. You may want to bias to at least 40mA per bottle, say 90mA total idle amp, which will put you near 480V DC which is fine.

It will work.

I'm not sure you can mis-load the W4M OT to meet Bassman loading and watts output. The UL connection is "wrong" for a classic Bassman. (Fender did UL later, but at atrocious power levels, and these are not widely loved.)

I must suggest that the stock W4M might be a VERY good music amp, perhaps with reduced NFB to loosen it up. The holy-grail Standel is essentially this circuit (although every one made was a little different). It is the direct godfather of the Dynaco, and Sunn's early amps were slightly-hacked DynaKits. However the stock W4M has more gain than classic Fender power stages, which influences preamp design. The 5F6A's low power-section gain forces the tone-stack driver to work hard, which adds color.

[sluckey]

Someone mention Sunn? That hifi iron is probably well suited for an old Sunn Spectrum II. It's a loud, clean amp though.

[eleventeen]

What PRR said. First step is to move to a 5U4 rectifier tube which will probably drop about 25-35 B+ volts relative to a 5V4, as a WAG. Cheaper tube, too. Actually, to be picky, a 5R4GY would be preferable to stay with a 2 amp filament (yes, filament, not heater!) vs a 3 amp one, but that ain't gonna matter. Keep your eyes open for a 5R4, though, next time you buy parts.

As for keeping it "stock", it appears to have been changed over to 9-pin sockets and a row of pots installed, so that ship has sailed!

Besides, once any kind of 1/4" jack is in there, it's a guitar amp. End of story!

Looks pretty good!

[Midtex]

Sorry guys if I didn't make it clear that all I am using from the Heath W4M is the iron. I do not intend to run ultralinear or the cathode-bias arrangment. I will be using these transformers in a pretty vanilla Bassman circuit (adding the additional gain stage ala Pignose G40V). The chassis you see in the picture is a new Hammond steel chassis that I bought for this project. I was hoping to use the 5V4G from the Heath as it tests fine, but the PT will handle a 3amp 5v heater, so a 5U4 is definitely an option - especially since I did install a standby switch. I am still wiring this up and I will report back once I get it powered up and get some voltage measurements. Thanks!

[Midtex]

Can anyone tell me how critical the value of the dropping resistor is for the phase splitter power supply? I know that Fender used a 4.7K here as well as most clones do too. Pignose used a 12K and the generic diagram that Sluckey posted above shows a 20K although I can't be sure that is suitable for 5F6. Any guidance would be appreciated.

[RicharD]

>tell me how critical the value of the dropping resistor is

It's totally important and entirely dependent upon the load.  It's typically easiest to reverse engineer a power supply from the lowest voltage backwards to the highest voltage point.  I'll use the 5F6 schematic as an example:
http://www.el34world.com/charts/Schematics/fender/BASSMAN_5F6.pdf

Let's define the B+ supply as having 4 nodes:
A.  432V, comes right off the rectifier & feeds the OPT center tap
B.  430V, comes off the choke and feeds the screens
C.  385V, comes off a 4k7 resistor & feeds the screens
D.  325V, comes off a 10k resistor & feeds the preamp tubes

So let's start at the back end.  Point C = 385V and point D = 325V.  This means they're dropping 60V across that 10k resistor.  Let's plug this into ohms law (volts=current*resistance). We know volts and resistance so we solve for current.  I=E/R = 60/10k = .006 or 6mA  W/O getting into any great detail, we can assume each of the 4 triodes is drawing about 1.5mA (your mileage may vary).

Now lets repeat this for the 4k7 resistor.  45V/4700R= 9.6 mA  Now subtract the previous 6mA and you can finger the PI is drawing 3.6mA or about 1.8mA per triode.

Here's where it gets tricky.  Now we have a choke to contend with so our nice-n-easy resistor equation does us no good.  Time to visit the tube data sheets.  Let's look at the 5881:
http://www.mif.pg.gda.pl/homepages/frank/sheets/127/5/5881.pdf

Oh look... the tube is only listed for 400V max and Leo hitting it with 432V.  Meh.... what 32 volts between friends?  Scroll to the bottom of page 2, and look at the typical operating conditions for 2 tubes in push pull.  Zero signal screen current is 5mA and zero signal plate current is 88mA.  Since this is a guitar amp and the voltage is higher, it's safe to assume these values won't be what you measure + look at what the current do when you're hitting it with max signal.  Counting on thumbs, we'll say the currents double.  Obviously Fender took telemetry with zero signal.  If you look at some of the other typical operating characteristics, you'll note as B+ goes up, current goes down for the same power.  There's a nice little linear equation that's part of ohms law that say power = current times voltage.  So let's say screens draw 5mA and plates draw 80mA at idle.  Now you have enough information to design your new power supply based on your specifications.

So you measured 865VAC HT with no load..... sheesh  That's almost 14% higher than printed on the schematic but 125V is almost 14% higher than 110V so I believe your measurement is accurate.  I'm gonna run against the grain here and suggest you stick with the 5R4 because it has a greater voltage drop than the 5U4.  The 5R4 should get you in the neighborhood of 475VDC whereas the 5U4 is gonna be more like 515VDC.  Personally I think this is bleeding edge for a pair of 6L6GC and probably over the edge for 5881, especially if they're new production tubes.  I see 3 ways to go; live on the edge, buy a different power transformer, or blow off some voltage through a high watt power resistor.  Should you choose option 3, I'd suggest a 25W 500 ohm resistor between the HT center tap & ground.  That should blow off about 50V.  Stick with the 25W rating.  Granted it's only pulling 5W, but it's gonna get rippin hot.  You also want to put a 47uF cap in parallel with this resistor.  Make sure you reverse bias it (+ to chassis) because that's gonna be a negative voltage.  From there you can simply copy all the Fender values for the rest of the power supply.

[Midtex]

Wow - thanks for the detailed response. I read and re-read it several times and I do basically understand what you are saying. I guess my question stems from the fact that Fender fed the phase splitter plate resistors with 385v as you stated and Pignose is feeding the exact same circuit with 328v, so it appears to me that the exact voltage that I provide to my phase splitter circuit (and therefore the remaining drops to the preamp stages via 10K resistors) isn't terribly critical and could probably be anywhere in that range? Also, I can see where the voltage drop across those resistors can be calculated as a simple mathematical function using Ohm's law, but what is trickier to me is understanding how much the secondary voltage at the transformer will drop as a function of total current draw. I am sure that is a function of the transformer itself and the current rating of the winding. I will never know what the operating voltages will be until I know what the starting voltage is. I reasoned that there will be voltage drop across the 7H choke and I thought that if I place it upstream of the first power tap to the OT center tap, that it will help lower my B+. I am also avoiding a large capacitance filter prior to the choke to lower the voltage too. I think if it is just too high, I would rather switch to a different power tube since I am going to have to buy them anyway. Possibly KT66? Anyway - I didn't realize that a 5U4 would drop the voltage more than a 5V4, so that is another good option. Here is what I am planning to do so far as a starting point until I can take some real measurements:

[RicharD]

You really don't want to choke load the rectifier.  That's a whole different can of worms I can't accurately explain.  You want a capacitor load and plates before the choke, everyting else after the choke.  I'd try something like what I've attached.

1.  Raise the center tap off of ground -50ish volts since your PT is putting out excessive HT.
2.  Use a 5R4 for maximum voltage drop.
3.  1st and second capacitor banks can be stacked 47uF.  Caps in series divide like resistors in parallel so this will give you 25ish uF per node.  I'd use 220k balancing resistors.  Less current draw = less heat.  100k are fine too.
4.  Plagiarize the rest of the circuit.

It's not critical that your PI be exactly 385V.  A little higher B+ means your PI can drive the output tubes a little harder.  The 12AX7 is rated for 330V.  You're dropping 150V across the 100k plate resistor so thangs are way cool either way.

I don't know how to calculate voltage drop of a transformer.  It's all based upon load and the power capability of the transformer.  At idle, figure 5% if everything is properly sized.  If you have a 500mA transformer and 50mA of load, you won't see any voltage drop due to loading.  If you have a 25mA transformer and 50mA of load, you'll see a huge voltage drop up until the transformer smokes.  I'll assume your transformer is 200mA and your idle current is.... 80ish mA.  You should expect 5ish % of drop.  Now under full tilt roar, a 6L6 geetar amp can easily hit 200mA.  Now you're gonna get some sag and that's what you want.

Duncan Monroe has written a nice little power supply designing application (that's free).
http://www.duncanamps.com/psud2/
It's not too hard to use once you finger it out and it's pretty much dead on.  You can measure the DC resistance of the primary and the off load voltage of the secondary, input it into PSUD2 and it'll return quite accurate voltages.

-Richard


oops!  attachment fixed

[eleventeen]

You really don't want to choke load the rectifier.

I know you say you can't explain why not, but do you have someplace you can link to that supports this? The choke does not really "load" the rectifier, it hardly loads anything; its purpose is not to load anything, it's to counteract changes in current and to a lesser extent, voltage. 

[ernie_jr]

http://www.vacuumtubes.com/6BG6.html  you could try this tube and use the higher voltages and get the same sound as
6l6's I have used this tube with 500 volts. It has a great sound.
good luck,
Ernie

[RicharD]

Here's a link about "choke input" power supplies shared with me ages ago by PRR.
http://www.tpub.com/neets/book7/27f.htm

Actually a choke input might be the way to go..... it might not.  I'm not going to try to talk somebody through something I don't fully understand.  I do know that your choke will need to be able to withstand 200mA.  I did build an amp (Triple Stout) that used a choke input power supply.  It was built from an old Hammond tone cabinet.  It was not safe and the project got shelved until such time as I can build a safe power supply.  Here's a link to that thread:
http://www.el34world.com/Forum/index.php?topic=5470.msg49013#msg49013

Here's a snippet:

This is a CHOKE-input supply ! ! !

Cap-input formula is VAC*1.41, choke-input formula is VAC*0.9 .

BUT an UN-loaded "choke input" works like a cap-input! It would take an infinite choke to "choke" an infinite load impedance.

There is a minimum value for the choke. Exact analysis is a mess, but there is an approximation which works and is stupidly easy to figure:

Lc=Vout/Iout for 60 Hz operation
Lc=1.2*Vout/Iout for 50 Hz operation
(V in Volts, I in mA, L in H)

Bless Hammond for putting numbers on their plans. You have 415VAC (each side). When working choke-input you expect 373.5V DC. The choke is 2.5H. 373/2.5= 150mA.

If you draw more than this, it is choke-input and about 373VDC (minus 5U4 and choke losses; 310V seems likely).

If you draw less, it rises toward cap-input operation, 415VAC*1.414= 587VDC. Which matches your 85VAC(in) = 425VDC observation, if you assume the plan was 117V wall-juice.

Is 150mA minimum load plausible? The seven power bottles draw 200mA, everything else is small. So in operation, the choke does its thing. Even if one tube goes dead.

But you can NOT run it with multiple power tubes missing. Or with cold fix-bias. Not unless everything tolerates 580V for indefinite time.

I'm somewhat astonished they apparently used an electrolytic first-cap rated only 450V? The 5U4 does warm-up faster than 6BQ7. After 5 seconds the voltage is going past 500V. After 10 seconds it will come down to ~350V, sure. But how many times can you slam a 450V cap past 500V? You can get away with this abuse when you sell $499 amps to guitarists, but churches who paid $4,999+ expect better.

As I understand this, with no load, B+ soars.  This means you have to have a standby switch and you cannot flip it until all the tubes are hot, otherwise everything sees >500V.  All your caps have to be rated at >500V or series stacked.  You have to have a set minimum load.  W/O the load, B+ soars.  Your output tubes make up most of the load.  If one dies, load drops, voltage soars, and the remaining tubes get toasted from too much voltage.  Been there done that simply by firing up my circuit with the standby in play.  IIRC, I lost 3 of 17 tubes and got a little light show to boot.  I thought about a time delay circuit with a current sensor and a 600V contactor relay before I said, "this is stupid, get the right PT."  One shouldn't need a manual to turn on a guitar amp.  On the shelf it went.

[PRR]

> I don't know how to calculate voltage drop of a transformer.

The specs are never complete enough.

Pick it close, then BUILD it. Since the big tubes are 90% of the total load, get 50V adjustable G1 bias, stick in the big bottles, any sorta-close filter cap, screens fed with 1K resistance. Bring it up to ~~40mA per tube, you know your idle B+ and approimate idle G2 current.

If you must know "sag", get a BIG resistor. For a 6K OT, use a 3K resistor. We are hoping for over 400V, so assume 500V and compute a power. 500^2/3000 is 83 Watts, which is twice expected output, which is reasonable. Since we only need this resistor to work for a few seconds, it can be a bit under-rated. OTOH some values/sizes are cheaper than others. If a 3K 100W does not turn up, maybe there's three 1K 25W, or twelve 1K 10W wired 2*6 may be easy to find (and conservative).

Since you are "stuck with what you have", and it isn't way-wrong), don't worry over-much. Guitar-market "6L6" will stand over 500V; if you really land higher, use 6550 or connect for EL34 (it's not a Bassman, but a few good amps do use EL34 in Bassman-like circuits).

Driver voltage: As a ruff guide, for 6V6-6L6-6550, driver supply should be 75%-100% of G2 voltage. That's usually "ample". Sometimes a designer wants a "ruder" amp, or an amp which gets rude without over-doing it... driver supply can be adjusted so driver won't slap Full Power from the big tubes _or_ so that it will smack them senseless. That's really up to you.

The G2/driver B+ _ratio_ found on 5F6A is a good starting point. Since you will probably wind up with higher G2, you may want higher driver supply.

And in that case, you will be very-close using 5F6A dropping resistor here.

I'm trying to avoid the G40V because I think it stinks by design, so that its "sound" will STAND OUT on a showroom floor crowded with bad tranny amps, good tranny amps, tube-emulation tranny amps, hybrid amps, and tube amps of every stripe. That does not affect the big question of building a 5F6A-like amp from a W4M.

[Midtex]

Thanks everyone for all the input. I think am am going to go ahead and use conventional 5F6A filter configuration and series-stack the first couple of capacitor banks and use 600V on the remaining caps to make sure I am safe. I will use 5R4 for maximum voltage drop and just build it and see how things look. Seems like a good pair of 6L6 will handle up to 500V so this might might just be OK. Waiting on more funds, but I will report back when I fire it up.

BTW - Responding to a couple of comments:

"I'm trying to avoid the G40V because I think it stinks by design, so that its "sound" will STAND OUT "    ----  I am basically using the Pignose gain stages, but modding to voice more like 5F6A. They are nearly identical to 5F6A.

"I must suggest that the stock W4M might be a VERY good music amp"   ----  Actually I have a pair of the highly regarded Heathkit W5M Altec Peerless output transformers that will be used in a nice stereo hifi build at some point. I figured I'd get my feet wet scratch-building this guitar amp. Prior to this, my only experience with tubes has been a basic freshening (recap) of my Dynakit Stereo 70 which I must say sounds pretty good.

[PRR]

> a nice stereo hifi

I meant musical instrument amp.

While there are several ways to design and voice a MI amp, a naturally-sweet tube hi-fi amp, overdriven, NFB loosened, is a good (though expensive) way to go.

A McIntosh can't be loosened-up enough. (And they are now far too valuable to hack.)

I'm not a huge fan of the Dynas, as hi-fi amps. They are good, but not rich. Yet of course Sunns work very well as MI amps due to Conrad's insight and robust build.

Someone from Florida has "my" ArKay two-6V6 amp. Not literally mine but the same model. I knew that machine very well, as a hi-fi amp. In retrospect it had the right colors to be a MI amp.

[Midtex]

I'm still waiting on parts, but I decided to dig out my tube stash (mostly used pulls) and inventory it last weekend, and look what I found! I also found a couple of 5U4's so I should be able to experiment with the voltage drop of each of these quite easily. I also found quite a variety of 12AX7's so this should be fun experimenting with those too. Love the shape of that 5R4GY.........

[eleventeen]

Yah, just don't exceed 20,000 feet on the 5R4 or there'll be hell to pay!

[FYL]

Yah, just don't exceed 20,000 feet on the 5R4 or there'll be hell to pay!

Or use a potato masher, spec'ed up to 60,000'.

[sluckey]

This one is rated all the way up to Timothy Leary.   

[Fresh_Start]

Uh, Steve, is there a B52 somewhere missing a high altitude tube?

Chip

[sluckey]

Naaa. Raytheon sold a bunch of those to the FAA. They're part of my old radar system. They were all replaced with ss plugin replacements back in the '70s when it was fashionable to replace tube rectos with ss diodes. So I got a bunch of leftover bench stock.

[Midtex]

Well, I've had some time to think this over and I decided the smartest and easiest move would be to replace my power transformer. The Heathkit PT high voltage winding was rated for 130ma which I am learning is not enough for 5F6A wide open. I found a transformer with universal primary that would fit inside my existing outer case so I can still retain my potted look and not have to hack-up my new chassis. This solves all issues - B+, heater voltage, bias tap, and current rating.  I can also now try the original Tung-Sol 5881's without fear of the plate voltage being too high. Heating the old transformer and scooping out the tar was not fun, but after an hour or so, it was accomplished. Waiting on parts still, but expecting tomorrow and anxious to get building again.