Sorensen Power Supply

[PRR]

> Sure would help to have a schematic.

Couple places (oddly both in Canada) who will sell or rent(!) manuals for similar models (like a "B" instead of a "BB").

> This is starting to scare me...

Very good idea. These beasts ARE dangerous.

But 1,400V, even CT, is odd for a 325V output. Taking 100V across the 6L6, 100V across the rectifier, the raw B+ might be 525V. For cap-input filtering (choke-input is almost impossible for a supply that is often run with no load) that says maybe 400V each winding (800VCT).

Generally these things worked much like the Heathkit. Start with +500V, drop voltage in big tubes(s). To a first approximation, the output voltage is the same as the big-tube(s) grid voltage. In fact the grid has to be held negative of the cathode and output. So if the output is asked to go to +325V, the big-tube grids have to be pulled to ~+300V; to go to Zero volts, the big-tube grids have to be pulled to -25V or so. This pulling around is done by the voltage amplifier, a small tube. Its output terminal is always positive of its grid or cathode, so to get its plate down to -25V, its cathode has to be tied to a large negative voltage. -150V in the Heath. So we need a negative supply in addition to the big positive supply for the main output.

The symptoms suggest that somehow the "center", where positive and negative supplies come together, has been "lost". So it looks like one extra-big positive supply. It is not clear to me where the fault would be.

The burnt part should be a clue (and yes, if it burnt, you need a replacement at least as large!). But a resistor won't burn-up unless something else is wrong. Maybe a prolonged overload, now long gone, but these boxes typically stood great abuse without failing. Replacing the resistor and watching it burn-up again might lead you to the fault. But I assume the markings are burned-up, so you don't know what to replace it with. With a mostly-complete schematic, we could probably guess the function and value.

 

[mwkeene]

Okay, so I really spent some time cooling down and just trying to get some small reverse engineering schematics figured out on this thing in order to learn how each tube in the supply does its job. I now understand, to some degree, why you guys have been referring to the 6L6s as pass tubes. I see that we have a raw ac going into the 5Y4 and coming out, rougly around 740 volts. I can see that after the rectified and somewhat smoothed current is passed through some L and C, it is then dropped across the 3 6L6s (I have not yet figured out all the control for this), and I noticed that I'm only dropping about 10 volts on all these pass tubes. You guys mentioned I should be getting more like 100 volts dropped off them. From looking at the similarities between this and the other heathkit schematic, in addition to the comment made by PRR, I can see now that I need to focus on whats going on in the area of the 6SH7 tube...this part has the most effect on the operation of the 6L6s, no? Oh, I also replaced that electrolytic and I noticed the output dropped around 80 volts (Now im getting around 710 on the B+, C- goes negative now from 0 to -8). I guess that "DC" is getting a little closer to smooth dc now... I also noticed there are a few inductors in this design, and one of them is huge. Would this mean that we have the elusive choke input that PRR metions?
Anyway, sorry for the bad explanation, I hope to have a schematic together for this thing in a few days from my notes. I'll post it up for you guys as soon as I can get it checked, and if anything you can at least get something interesting to look at.

One last observation, im only getting about 20 volts DC (last i remember, ill have to check again) off the 5Y3, which we surmise is the half wave rectifier for the C- supply. That makes me feel like this has something to do with the "unweighted" positive voltage that PRR mentioned. I've neglected to check the "center" if you will, and I will focus more intently on that in the days ahead.

PRR: what places in particular can I aquire one of these manuals from? I emailed elgin a few days ago about the possibility of buying or seeing a schematic but they havent mailed back...

Thanks.
-Mike

 

[mcs]

[quote author="mwkeene"]I also noticed there are a few inductors in this design, and one of them is huge. Would this mean that we have the elusive choke input that PRR metions?[/quote]
That's easy to check. Does the output from the main rectifier go to a cap first or through a choke first?

If the DC voltage passes through a choke before "seeing" a cap, you have a choke-input supply. A choke input supply will have an output voltage of sqrt(2) * the AC RMS voltage with no load. When the load then increases to above a certain point, the voltage will drop to about 0.9 * the RMS voltage.

Best regards,

Mikkel C. Simonsen

 

[mwkeene]

Hi again,
I had to take a little break from this project because I had my finals, but I have since traced out the circuit, and have posted the schematic here:

EDIT: There should be a 180k pot wired with one end and the wiper shorted (0 to 180k variable) in series with and right above the pot that is paralleled with the 270k resistor... I will fix this in the schematic after work tomorrow...

http://bingweb.binghamton.edu/~mkeene1/Sorensen 325 BB (resize1).JPG
And here's a big one:
http://bingweb.binghamton.edu/~mkeene1/Sorensen 325BB (basic cleaned).jpg

The unit does look pretty similar to the Heathkit Model, with the exception of the 6SL7 and the lack of the seperate transformer windings for the 6L6's. I did not include the Voltmeter and the lamps in my schematic (I didnt think you guys would really care).
In my unit, the 5Y3 tube was no good, as I replaced it with two diodes and the C- supply started working (I finally got to see the OD3s glow purple). As soon as I started getting a negative rectified voltage, the postive output dropped dramatically to around 100 volts, and the trim knob now seems to be having *some* effect on the B+ output (70 - 150 volts now, but should be 0 - 325 volts). There seems to be some lagging between a change in the trim knob value and a change in the acutal ouput by a second or two (the 1uf and the .5uf cap to the grid / trim pot on the 6SL7 still need to be replaced, could these be causing this?).

Also, something kinda funny. The Red post on the C- output is wired to common, and the Black post is the negative voltage tapped off the voltage divider on the OD3. Why would they do this, wouldn't they want to keep common black on all of the posts, or is this just sort of a preference matter? It just seems confusing to think of it this way.

Question: The B+ supply is choke-input, didn't someone say that unloaded outputs on a choke input supply caused strange behavior? Perhaps I should try loading the output down again, now that this thing is semi-working.

-Mike

 

[PRR]

> the trim knob now seems to be having *some* effect on the B+ output

Horay!

> 70 - 150 volts now, but should be 0 - 325 volts.

Yeah, those output divider resistors do not make sense. With the values on your excellent drawing (I read them as 180K, 180K, 270K||??pot, 100K), I get 310V at one end, but hits zero mid-turn. If I assume "270K" is 27K, I get a smooth 140-47, mighty similar to what you see.

So.... how sure are you of these resistor values? Do any look replaced? Might any red stripes have turned brown from heat? If any look for-sure original, say so. If all look dubious, we'll have to do the hard math and invent the proper values.

> The B+ supply is choke-input, didn't someone say that unloaded outputs on a choke input supply caused strange behavior?

Forget about it. If your notation is correct (700VAC transformer, 710VDC on the cap) then it is almost choked. Perfect choke-input gives 0.9VAC or 630VDC, minus rectifier loss. Say 530VDC under heavy load. That puts 200VDC across the 6L6s under heavy load, a plausable condition. No-load, voltage rises toward the unchoked 1.4VAC or 1,000V, but bleeders keep it down to ~710VDC. You will have 710V-325V= 385VDC to 710V-0V= 710V across the 6L6. The later is beyond 6L6 ratings, but at no-load it will be fine for benchwork.

Another imporbable resistor(s): those "2K5 10W" bleeders can't work as marked. 25K 10W seems likely, 2K5 100W is possible but awful darn hot. Wonder what they really are. But the idle voltages look right on the mark, allowing for a possible burnt/bodged divider string.

> The Red post on the C- output is wired to common, and the Black post is the negative voltage

Of course! Red is positive, Black is negative! Don't bet the board on it (long story about a negative-on-red board), but that WAS/IS the common convention. And since the "C-" is a Negative (relative to Common) voltage, it "must" be Black and its return is Red (and Common).

You don't see "C-" supplies in recording audio much, but it is that "bias" supply in high-power guitar amps, the one that some guitarists obsess over. 6V6 may want -12V bias, over-volted 6L6GC may need -50V bias, 300B wants -60 to -90V bias, and a few tubes want more. -150V is rather overkill for a +325V supply, but they have the -150V already. And sometimes you are in a pickle where you use the gear's own +1,200V B+ supply with a bench-supply for heater and C- bias.

 

[PRR]

> There seems to be some lagging between a change in the trim knob value and a change in the acutal ouput by a second or two (the 1uf and the .5uf cap to the grid / trim pot on the 6SL7 still need to be replaced, could these be causing this?).

That, to me, is a good sign they are working. You don't want it jumping around; it is a steady power supply. And the monster 0.5uFd sure would/should make it sluggish. You might clip one end: the DC voltage should not change, but ripple and utility-crap will go up.

This is about as close as I can get using mostly the values you see:

 

[mwkeene]

I've updated the schematic. I made an error copying my previous schematic, and left out a 270k pot (right above the positive trim pot). The 270k resistor in parallel with the positive trim pot measure 130k in parallel...thus the pot comes out to 250k... Apart from that, I'm fairly confident in my resistor values as none appear to have been burned (I know I said I thought some were earlier but these just don't appear to be damaged) but stranger things have happened, so I'll double check them.

I'll check those wirewound bleeder resistors...the writing on them is a little confusing although they measured 2.5k last I checked.

http://bingweb.binghamton.edu/~mkeene1/Sorensen 325 BB (resize1).JPG
And Large:
http://bingweb.binghamton.edu/~mkeene1/Sorensen 325BB (basic cleaned).jpg

Sorry about that error there.
I'll try that simulation again with the updated resistor values and see what happens... looks like R5 would be variable 360k to 630k which should allow for the resistors and pot values I found?

-Mike

 

[mwkeene]

oops, didnt mean to post anything.

 

[PRR]

> didnt mean to post anything.

You can Delete your own posts. Click the

icon above the errant post.

> left out a 270k pot

AHHH! I kept thinking the 2*180K should be more like 500K. But 2*180K is such an odd pick, that I assumed it was correct and tried to fudge the rest.

Also the zero-set is very fussy. Small shifts of value mean it won't go to zero or goes to zero long before you reach the end of the pot-rotation. So it makes a lot of sense to have an internal trimmer to set zero.

With the top-string at 537K (2*180K + 177K pot setting) it just kisses zero at one end. The other end simulates as ~400V, with 325V coming at about 90% rotation. That makes sense to me. There isn't(?) any high-end trimmer, so they have to aim about 20% high to be sure worst-case tolerances still get you to 325V. On a not-worst-case unit, you can turn it higher than 325V. It will even work up there, but only for light loads. If you load it up to the full current rating and "only" get 325V, you can't complain to Sorensen about it. But for mike preamps and stuff that eats 10mA-30mA, you might be able to get a solid 350V, maybe 400V.

> 70 - 150 volts now

Is this still the situation?

Measure all the (non-heater) pins on the 6SL7 and 6SH7.

Both grids of the 6SL7 should be near-equal (a few volts) and about -213V. Its cathode should be a few volts higher. One plate should be exactly zero (re: Common).

6SH7 cathode is tied to -152V. Its grid (and the other 6SL7 plate) should be a few volts negative of that, say -154V. G2 is tied to zero V.

6SH7 plate is up there in the 6L6 outputs, and if things go wrong could be around 700V. Be Careful! But if you are getting +100V output, +100V on 6L6 cathodes, then 6SH7 Plate should be a dozen or so volts less than that. No-load, maybe 30V lower than 6L6 cathodes and B+ output, or around +70V.

If things don't make sense, get the leather gloves and pull 2 of 3 6L6s. You may have a bad one, and it will work at light loads (say 0mA-40mA) with just one 6L6. If it works the same with any of your 6L6s, either they are all good or all equally bad (this is unlikely).

With no load, it may take a while for the output to drop after turning the knob. Find or make about 100K 1 watt to load the output.

> I'll check those wirewound bleeder resistors...the writing on them is a little confusing although they measured 2.5k last I checked.

The bleeders are from the +700V raw DC rail, from raw B+ choke to ground. I see there are other 2K5 10W resistors in the raw negative rail; those look reasonable. But on the positive rail: 700V/2 is 350V, and 350V across 2K5 is 49 Watts, which does not match the "10W" note.

I've never seen a 6SH7. Unless it has become faddish, it is probably $5 plus $20 shipping, something silly like that. It would be possible to redesign the 6SL7/6SH7 stuff with sand. The 700V is a challenge, but a 1000V MOSFET might do the whole thing (again: $5 plus $20 shipping, but you might find HV MOSFETs laying around).

 

[mwkeene]

Well, it seems to be working now! I changed out the .5uF cap tied to the grid/pot on the 6SL7 and started getting an output range that seemed a lot better. The cap must have been leaky, as I noticed about halfway through the rotation of the + output knob I got to zero (maybe 150 volts full up). Now all the caps (except the .1uf going from chassis to Com) have been changed.

[quote author="PRR"]
With the top-string at 537K (2*180K + 177K pot setting) it just kisses zero at one end. The other end simulates as ~400V, with 325V coming at about 90% rotation. That makes sense to me. There isn't(?) any high-end trimmer, so they have to aim about 20% high to be sure worst-case tolerances still get you to 325V. On a not-worst-case unit, you can turn it higher than 325V. It will even work up there, but only for light loads. If you load it up to the full current rating and "only" get 325V, you can't complain to Sorensen about it. But for mike preamps and stuff that eats 10mA-30mA, you might be able to get a solid 350V, maybe 400V.
[/quote]

Yes, you seem to be pretty close. By my calculations I get 191k, you get 177k, and zero seemed to come on the unit around 200k, so we're all in the ballpark. You're also pretty close on the high end of the voltage output: with a 100k load, max output is roughly 380V, and min is 0V (give or take 200mV).


[quote author="PRR"]
6SH7 cathode is tied to -152V. Its grid (and the other 6SL7 plate) should be a few volts negative of that, say -154V. G2 is tied to zero V.
[/quote]

Yep!

[quote author="PRR"]
The bleeders are from the +700V raw DC rail, from raw B+ choke to ground. I see there are other 2K5 10W resistors in the raw negative rail; those look reasonable. But on the positive rail: 700V/2 is 350V, and 350V across 2K5 is 49 Watts, which does not match the "10W" note.
[/quote]

Yeah, I think the 2.5k's on the negative rail were dissipating about 5 watts at most, but that will probably drop when I put a working 5Y3 back in.
I'm looking at them now...and I realize I made a mistake. They ARE 25k, according to the markings, I missed the extra 0. I could have sworn that I measured them, but I can only say for certian I measured the 2.5ks. This makes much more sense now, 2.5k 10W on the negative side, 25ks on the positive, and they both dissipate about 5W.

[quote author="PRR"]
I've never seen a 6SH7. Unless it has become faddish, it is probably $5 plus $20 shipping, something silly like that. It would be possible to redesign the 6SL7/6SH7 stuff with sand. The 700V is a challenge, but a 1000V MOSFET might do the whole thing (again: $5 plus $20 shipping, but you might find HV MOSFETs laying around).[/quote]

I paid $6.00 plus $5.00 shipping for a lot of 5 on ebay...anyone need one? I have a feeling I won't be using all of these up any time soon.

So, basically this thing is just a big differential amplifier? It's kinda hard for me to make out an op amp with the 3 6L6's, 6SH7, and 6SL7, although I do see how 6SL7 could be the differential input stage of the op amp.

PRR, bcarso, mcs, thanks so much for all the help, I'm really excited I finally got this thing working. Its been kinda tough for me cause this is the first real project I've taken on since my father passed in December, but it really feels good to know that there's people out there willing to help!
-Mike

 

[PRR]

> basically this thing is just a big differential amplifier?

Sure. One side is the steady voltage from a gas-tube. The other side is an (adjustable) fraction of the difference between the output voltage and a steady voltage. These go to the twin-triode diff-amp. Its output goes to a pentode amp. Its output drives a (big!) cathode follower.

Compared to a general-purpose op-amp, it has this "flaw": the output must be (much) more positive than the input. Tubes only come one way: output more positive than input. If we had to, we could bring it back down with resistor or gas-tube dividers. But in this case, we want a very positive voltage. Though because we stack four tubes, we have to start far below the lowest output swing (we start from 300V below the lowest output).

Because things are a little tight, the 6SL7 runs at a gain of maybe 20. There is a design up-thread using just that much gain: it is not wide-variable voltage and probably not great regulation. Sorensen wanted more range and more regulation, so they added the pentode. It is working under very good conditions: grounded (to -150V) cathode, solid G2 voltage, and its plate resistor is bootstrapped off the cathode follower. With reasonable assumptions, its gain is over 100 (though varies a lot with load and voltage). The cathode follower has gain generally over 0.9. So overall gain is over 1,000. This full gain is available to reject AC above ~10Hz, giving low output ripple and output impedance well under 1 ohm. For DC, the volt-set divider has gain of 0.5 to 0.25; still excess gain of 250+ which will keep line/load-variation down to 1%. i.e. if you set B+ to 250V, and then change something in the unit under test, you mostly do not have to go back and see if the +250V changed, it has enough feedback to be "rock solid" for any practical benchwork.

If you disconnect the top of that 0.5uFd cap, and stick audio in it, your output will be a DC voltage plus an audio voltage. You would not drive a speaker this way: the big DC would upset it (anyway it won't put much DC into 8Ω). But you could drive the power rail of a 1MHz oscillator and have an AM transmitter: you make AM by taking a 100V-powered oscillator and swinging it up to 200V and down to 0V with the audio signal. (Good AM is a little more than that...)

Yup, 25K makes sense for bleeders. 14mA bleeder current is about 10% of the maximum current the choke has to supply (about 130mA, 125mA output rating plus a few for the circuit). It is possible to design a choke-input for a 1:10 range of current, or rather a 2:10 range of fully-choked operation and a slight rise in the bottom of a 1:10 range. The rise is no big problem: at light load, ripple is small, and the cost of higher voltage caps is less than the cost of a bigger choke or more power transformer to carry bigger bleeder load. (Someone probably did a lot of sums to find a commercially optimum ratio of choke, caps, and bleeder.)

> put a working 5Y3 back in.

I wouldn't. IMHO, hollow rectifiers are a bad idea; we just had no alternative. At the very least, you save 10 or 15 watts of heat by not having that filament glowing. And if anything is going to really kill this monster, it is time*heat age effects on the transformer. Personally, I'd sand-state the 5Y4 (though 1000V diodes are not really good enough on the positive rail).

Check that the gas-tube current is not over the rating; if it is, fudge the 2K5 resistors in the negative rail. You can determine the negative current by (carefully!) measuring the voltage drop on one of the 2K5 resistors. Subtract around 4mA for other loads, the rest is in the gas-tubes and IIRC should be less than 35mA for this tube. So you do not want to see more than about (35+4)*2K5= 100V across the 2K5. Gas-tube current should not be less than 5mA or the glow breaks-up, so you wanna see at least (5+4)*2K5= 25V across a 2K5 resistor.

Hmmmm... without the 5Y3's losses, your gas-tubes may be running too rich. I think you need >7K1 between perfect rectifier and gas-string, you have 5K plus the choke. I'd throw a couple of K extra in there (probably just before the gas-tube). That also makes sense considering the typical drop of a 5Y3 at these currents.

> this is the first real project I've taken on since my father passed

It's Memorial weekend. Have a hot-dog in Dad's memory. Stick two clean nails in a hot-dog. Wire it to the power supply. Be careful! Crank the voltage up: at around 100V the dog will cook pretty good.

 

[mwkeene]

Thanks for the op amp explanation.

I measured the voltage, and I get about 47mA through the 2.5k resistors on the negative side. Data sheet for the OD3 says 40mA max, so I guess I need to throw another resistor in like you said. Adding one for a total of 7.1k makes the current about 30 mA, which looks good.

Wow, okay these rectifer heaters draw a lot of current: 2A for the 5Y3 alone. I see why you say that they tax the transformer like that, and I don't want to shorten the life of these as they are the only parts I cant easily replace.

Tube data for the 5Y3 says it drops about 50 volts, so can't I just put a 5W 47volt reverse biased zener in series with the forward biased 1N4007 on each side of the rectifier instead of adding the resistor down the line? I mean this is a more expensive option, but it seems like it would work too.

So 1000V Diodes won't be enough for the positive rail? We're gonna have almost 1600V max on the diodes (when reversed biased) right? Couldn't I just put two 1000V diodes in series to get proper ratings? Again maybe with a 47 volt zener?

Okay, I'll cook a hot dog or two. I'm hungry.
-Mike

 

[bcarso]

If you carefully match the capacitances and leakage currents of diodes you can make series strings of them. It is best to put some high voltage resistors in parallel to ensure equal voltages as well. And those resistors have to be rated for that voltage withstand, so the whole thing gets to be complex.

Or, you can pay somebody for a part that is made this way internally.

WAR STORY ALERT (you may elect to exit now---use your seat cushion as a flotation device)

TRW had a contract with the Navy for a high-powered chemical laser. After the initial prototype got close to working the whole shooting match was moved to a facility near San Juan Capistrano.

The electrical folks had to provide a substation to bring in 23kV three-phase at many amps. Directly rectified this makes a moderate-ripple d.c. that was to be the power source for the capacitor banks and triggered spark gaps. This beast was a monster, lasing in a hypersonic stream of deuterium and fluorine which added the energy of the chemicals making the DF to the electrical energy coursing through it.

Of course only electricians are licensed to put stuff in the substation together. After a master tech at TRW had spent many many hours selecting and matching big HV power diodes and assembling them in potted modules, and after the man with the stencil and orange paint had come round and put the TRW logo on the modules and another person had given them a part number, the diodes were shipped to Capistrano.

No offense to the electricians in the audience, Toobie et al., but it must be said that when presented with unfamiliar situations some electricians tend to...errr.... royally screw up.

The one assigned by PG&E did.

When they switched on the power the diodes blew immediately and spectacularly.

Eventually they got it right. The problems of the energy source were fixed. There were many other problems, especially fluorine leaks and cows downwind, but that's for another installment ;-)