One for the Tube Guys!

[craigmillard]

I have 2 * 400VA transformer i scrounged with the following secondaries:

0-30
0-30
0-10

Im wanting to power a 50Wish scrapbin build amp with 6l6 or 6v6 output tubes and been reading merlin's power book

Can i use these transformers to accomplish this..

I could use a voltage quadrupler and combine all 4 30v secondaries to get 680V but that is way to high for both 6v6 and 6l6 tubes..
I could use a Voltage Quadrupler on 1 transformer and get 340V but that seems too low for the tubes..

Any ideas.?

[wavley]

I have 2 * 400VA transformer i scrounged with the following secondaries:

0-30
0-30
0-10

Im wanting to power a 50Wish scrapbin build amp with 6l6 or 6v6 output tubes and been reading merlin's power book

Can i use these transformers to accomplish this..

I could use a voltage quadrupler and combine all 4 30v secondaries to get 680V but that is way to high for both 6v6 and 6l6 tubes..
I could use a Voltage Quadrupler on 1 transformer and get 340V but that seems too low for the tubes..

Any ideas.?

I just got a Cornford Carrera, it's single ended and it's not going to hit your 50 watt goal, but it sounds really great and runs 325 v and every octal I've thrown at it sounds wonderful.  So 340 v might not be a no go.

[merlinb]

How about three 30V windings plus one 10V winding, with a quadrupler. That would get you about 560V off load, or perhaps 460V full load.

Not sure I'd try powering a 50W amps with a quadrupler though, as the ripple current is likely to be painful...

[PRR]

400VA??

And you only want a puny 50 Watts out? You could do 100W. 200W if you scrounge a heater transformer.

OK, the lump in the hand beats shipping a lump from Hammond, and 200W OT lumps are heavy price and shipping.

What occurs to me.... may not be your flavor..... 30VAC makes 42VDC, you have two for 84V total or +/-42V. LM3886 (chip-amp) can take that much, almost, kinda. (Use the 10V to buck a 120V wall and you have +/-39V.) LM3886 will deliver 50 watts into 8 ohms at that voltage. And with 400VA on tap you could run four channels E-Z, for 200W total. This could be a dandy Quadraphonic rig (so 1980s!) or National/TI has an app-note about bridge-parallel connection for 200W in one 8 Ohm load.

50 Watts from two 6L6/EL34 really means 450V. That's a long way up from 84V. Roughly a Quintupler, a very odd thing.

Using 30V+30V+10VAC quadded gets you just shy of 400V, which in two 6L6/EL34 is nearer 40 Watts (I think some 40W amps have been popular?). As Merlin hints, the capacitor array for good regulation and ripple may be daunting, and violate a "scrapbin" approach.

30V+30V in a Quadrupler makes 336V. That's actually an excellent supply for 50W from *six* 6V6 and a 3Kct OT (Marshall 3.4K winding is not so far off.)

Would be sweet to find a big cheap _10V_ heater tube and get the heat from the same PT. I can think of some monsters with 10V heat but they tend to need frightening plate voltage. There is a 5V6 and they can be had for $6/each and are surely all NOS golden-age.

[amptramp]

You could get 25 watts from a pair of 6BQ6GTB's or 35 watts from a pair of 6DQ6B's or 75 watts from a pair of 6146's but you need low screen voltages (preferably regulated).  6L6's and 6V6's are premium tubes now that bring a good price due to their favour with the cork-sniffing set whereas sweep tubes (6BQ6, 6DQ6) and transmitting tubes (6146) are still relatively cheap.  Once you get up to these power levels, you have to know what you are doing with the design and build in some precautions.  It was common for transmitters to use a clamp tube (and a clamp transistor would probably be better) which was connected from screen to ground and held in cutoff by the grid bias.  If the bias disappeared, the clamp tube would come on and drag the screen voltage down to the point where output tube dissipation was survivable.

BTW the 25BQ6 and 25DQ6 variants are plentiful and permit lower filament currents.  High plate voltages may require higher voltage than you can get in electrolytic capacitors, so you may need series capacitors with bleeder resistors across them to equalize the voltage drops.  This is especially true if you use conservative derating rules such as 60% for capacitor voltage i.e. the design is done such that an electrolytic sees no more than 60% of its rated voltage.  The power transformer and output transformer will be expensive, so conservative design is recommended.

[PRR]

> need low screen voltages (preferably regulated).
> may require higher voltage than you can get in electrolytic capacitors, so you may need series capacitors with bleeder resistors

Both points moot with a low-volt high-VA PT.

He needs >100V caps for the initial quadrupler-string. The tappings give solid voltages without bleeders.

There can be a bunch of part-volt taps on a quadrupler for G2 supply, and with a VA rating far in excess of need (and assuming good cap values) the regulation will be ample.

There's some over-demand for sweep tubes (it's no secret), though the ones you cite don't seem to have gone up in price.

The circuit values for a good *safe* (non-smoking) sweeptube guitar amp are not as well-known as values for "the usual bottles".

The good (for TV) bad (for amplifiers) thing about sweepers is that they will pass HIGH current. Which means that where a mis-loaded 6L6 amp will just be strained, a sweep tube with poor choice of values may melt itself and its iron.

We can buy an OT for The Usuals by description, a sweeptube version has to be designed to suit available OTs.

I like looking at such oddballs, sure. (I like HK-257 as monster single-ended.) But actually building one may not be everybody's cup-o-tea.

I'm still a bit proud of the 5V6 idea. 6V6 values are everywhere, and 5V6 is just a different heater. 5V6 is cheap, and good-quality (late 1950s-early 1960s). Probably mostly made at Ken-Rad. The 30+30 winding quadruples to a good B+, also <450V so standard e-caps work for further filtering. The 10V winding powers 5V6 heaters in pairs. The 6X 45+W and 12X(!) 90+W conditions match Marshall 50W and 100W OT impedances, which are commodity parts.

[craigmillard]

Wow! Loads of info as per usual guys!! Thanks

Posted by: PRR
400VA??

Yup, lol, its a bit of a monster! Pulled from some broadcasting euro-rack gear Going to use the rack as the chassis for the build too!

I like the idea of the 5v6 tubes PRR but almost impossible to get in the UK at the cheaper rate!:(

I can on the other hand get hold of some 6p6s russian tubes which are similar to the 6v6's for about £15 for 8
I also have a spare 2 * 12 30va torroid which should have enough juice to power the heaters..

30V+30V in a Quadrupler makes 336V. That's actually an excellent supply for 50W from *six* 6V6 and a 3Kct OT (Marshall 3.4K winding is not so far off.)

I thought that was a touch low.. most amps iv seen run the 6v6's at more like 380v? Would the sag from the load not drop the voltage even more?
This is the way i think im going though..

When stripping the eurorack i also came by 10 6800uF 50v caps and also have 12 220uF 400v caps lying about, these should work well in the Quadruppler i think.
trying to get my head round how you work out the capacitance needed in the quadruppler string.. In my quest i found this site quite good:
http://www.ozvalveamps.org/ava100/ava100psu.htm
Is it worth using the 6800uF caps in series (Increase voltage) or not and just stick to the 220uf's?

Posted by: merlinb
How about three 30V windings plus one 10V winding, with a quadrupler. That would get you about 560V off load, or perhaps 460V full load.

Hi Merlin, you book is helping me out no end at the moment! great read:)
So you can combine different voltage secondaries together without issues? i assume the current maximum will be smallest secondary * 4. Unfortunately i dont know the breakdown of current across the windings in this tranny, i know the 30v secondary is the main one but no idea what the 10v ones can handle!

Posted by: PRR
There can be a bunch of part-volt taps on a quadrupler for G2 supply, and with a VA rating far in excess of need (and assuming good cap values) the regulation will be ample.

Do you mean there are points i can tap off the quadrupler for different voltages, for the screens say?

[amptramp]

If you have a 12 volt toroid, all versions of 12BQ6 and 12DQ6 use 12.6 volt 0.6 amp filaments or you could run 6-volt tubes in series so that if one fails, both stop conducting current and you don't have a problem with self-bias dropping if you are using self bias.  A 12BQ6GT pair will give you 25 watts at 300 volts on the plates with a 270 ohm 2 watt bias resistor for resistive bias or 30 watts with fixed bias of -30 volts, so this should be adjustable from -25 to -45 volts.  You use an output transformer set up for 4000 ohms plate-to-plate.  If you want to make it look awesome, use a VR-150 from cathode to screens and have it visible from the front so you can see the nice pink argon glow.  The original -GT suffix tubes can use supply voltages up to 550 and the -GTA and -GTB suffix tubes can run from 600 volts.  Now here's the fun part:  Need more output?  Pull the 12BQ6GT's and put in 12DQ6 or 12DQ6A tubes - a little wider diameter but capable of more power and here is the nice part - it still uses the same grid bias and filament current so you don't have to change anything, although you may want to check the bias setting.  An amplifier using plate caps and a pink VR tube is going to look awesome!  The gain of the 12DQ6 is higher by the same proportion as the current-handling capacity, so the drive to the tube does not need to change.  Run 10 ohm cathode resistors so you can measure plate current if you use fixed bias.

As long as the output transformer can handle the maximum DC current of the tubes (110 mA for each 12BQ6, 120 mA for a 12DQ6 and 140 mA for each 12DQ6A), you do not have to restrict the design that much.  If you are concerned about the possibility of excessive plate current, you could use a fuse, circuit breaker or transistor current limiter in the cathodes.

Transmitting tubes are also popular for power audio.  A pair of 6146's will give you 75 watts running from a 500 volt supply with 4200 ohms plate-to-plate, 95 watts with 600 volts and 5200 ohms plate-to-plate or 120 watts with 750 volts and 6700 ohms plate-to-plate.  Although sweep and transmitting tube audio systems are unusual, amateur radio operators made good use of these tubes for AM plate modulators for many years.  The example values I have used are from the 1956 Radio Amateur's Handbook.

[PRR]

> most amps iv seen run the 6v6's at more like 380v?

If you read the 6V6 data-sheet, the suggested conditions are 250V-285V, and the Absolute Max rating is 350V.

Yes, good 6V6 will stand a lot more than that. Fender pushed them toward 400V and all modern copycats follow. I'm not real sure it is about "tone", but Fender using the cheapest tubes for the most power (short of 6L6/5881 design/cost).

> Would the sag from the load not drop the voltage even more?

A few 6V6 might suck 30W-60W of DC. A 400VA transformer won't even notice that, won't think about sagging.

True, with a mult-upler supply the ripple may be enough to warrant an additional filter stage. Unless your scrapbin runs to good chokes, that's an R-C filter. But at today's cap prices you can lean to small R large C and get low drop.
____________________________________

> how you work out the capacitance needed in the quadruppler string..

It's a brain pain. Plagiarize (knowing that many posted plans are also brain-pain limited), breadboard (with BIG load resistor), or use the idiot assistant. SPICE does math good, but is very likely to tell you exactly what you asked instead of what you thought you were asking.

> In my quest i found this site quite good: http://www.ozvalveamps.org/ava100/ava100psu.htm

Well, isn't the "Phase 5 PSU" about 3/5 down the page pretty much what you need? 30V+30V windings. Dead PC power supply caps. Up to 240mA of DC output.

Here's my version:


The 170V tap is solid, though may need an added filter for very-clean use.

Idiot Assistance gives these peak cap voltages (for 60VAC input):

C1 C2 > 85V
C3 C4 > 170V

Round-up generously, over-volted caps stink the room and leave glop on the ceiling. At least 100V and 200V.

There is a trade-off between C1 C2 size and C3 C4 size. You'd think the lower volt caps should be bigger uFd. However the high-volt caps set the output ripple. I like setting them all the same for simulation simplicity. If you like to balance uFd against V so the physical can-sizes are all the same, that's cool too. So is odd-lot scrapbin sizing.

470uFd all around with 2K load: 320VDC 160mA out, ~~5Vpp ripple (1.5%)

This 1.5% ripple is really "too good". It suggests we over-spent on caps. A commercial design would use about 3X smaller caps, take the 5% ripple, and either deal with it (push-pull pentodes usually can) or add a second stage of filtering with the 2/3 of cash we saved with 1/3 size first caps.

But as that page says, there's two 200-300uFd caps in any good PC supply. Or--

> 12 220uF 400v caps lying about

Gross overkill for C1 C2, but better put them to work before they decay of old age and idleness.

320V 160mA is pretty much four 6V6 at 28 Watts clean out. 470uFd (or 2*220u) would be ample up to the 75W out level.

!! RMS current in D2 D3 is *1.3Amps* at 1 second (7A first cycle)! D1 D4 a bit less. These can not (reliably) be 1N400x parts for >100mA output. Use the 2.5A diodes.

[chobot123]

Maybe a stupid idea, but what about using the two transformers back-to-back, perhaps with combined windings? I mean mains into primary of X1, then one or more secondaries into the 30V sec of X2. I'm not sure I' able to calculate the resulting voltage, but if it works, you can avoid the quadrupler...

[tubegeek]

So I happen to have a schematic for a hi-fi amp (no bandwidth shaping going on) that illustrates using the mid point of a doubler to drive the voltage amp, and the higher voltage to drive the power amp. (It also illustrates the back-to-back transformer idea in the post above.)

I did this so I could direct-couple the stages and it's single-ended.  But there's nothing that says you can't cap-couple and use the whole B+ to get more output power, if you want that, and go push-pull if you want that too.

Also built into this one is the idea of using a dual-dissimilar triode bottle for a fun party trick. I can offer that up along with the observation that the 15FM7 is cheaper than the sockets it uses. (I used 13FM7 in mine but 15FM7 is the same: triode #1 has a mu of 66 and triode #2 is a low mu, VERY low rp power amplifier. It's a tube used for vertical oscillator in TV service, designed to hit a transformer hard at 60 Hz. Makes a wonderful audio tube.)

You'll notice perhaps that the audio circuit has almost no parts in it - what parts there are are bulky (plate chokes, a lot of power supply caps.) Direct coupling does that for you. If you wanted to go push-pull you could do it with a very similar setup and use a transformer coupled phase splitter - that would be The Balls if you ask me, and would certainly give you a different sound in a guitar amp than all the other kids have. A little more expensive but Edcor has all sorts of cheapish transformers we could sift through together, that might cough up a candidate.

A pair of 15FM7's would do push-pull with two bottles (one voltage amp, one concertina phase splitter, two power amps) and give you maybe 3-5W pushed hard, and of course the heaters would go in series across a 30V winding to fit your transformer. More of a bedroom amp than a plasma cutter/rail gun/melt-the-polar-ice-cap sort of power amp, but anyway another way to go.

Here's how I did mine, maybe there are some ideas in it that will be useful in this discussion. Anyway it was irking me that I had not joined a thread titled "One for the Tube Guys!" yet so I had to do SOMETHING.

Obviously you'll be leaving a lot of your 400VA on the table with this sort of idea, but maybe you can do 5 channel audio or something with the rest and have a tube home theater setup.

[PRR]

I can't say I've ever seen quite that topology.

You have >300V difference between the two cathodes on one heater. You better bias the heater winding up about a hundred volts; even then you are over the rating and may have to discard 10% of tubes for H-K arcing. {EDIT-- whoops, I see you know and did that.}

The top side of your doubler runs 60mA while the bottom runs like 6mA. Very unbalanced PT flux, poor economy of iron. (Yeah, the starter post has way-ample iron, so moot in this thread.)

[tubegeek]

I can't say I've ever seen quite that topology.

The amp that inspired me - with the voltage doubler split supply - was by a Japanese audio designer Mr. Komuro.

Check this article out for more info on this style of amplifier with some variations:

http://home.earthlink.net/~jeremyepstein/freelunch.html

[wavley]

My HiFi amp is 8 watts a side (plenty for the super sensitive Janszen electrostatic speakers it drives) and uses 11BM8 tubes which are dissimilar tubes in push-pull, each tube has both a triode and pentode, the whole thing runs on like 250 volts and sounds really great.  You could use them for the power section (or something like them) and a normal dual triode for your first two gain stages.

Just a thought.

[craigmillard]

Hi guys,

I have come back to this as been a bit busy of late!:)

I have the multiplier working as per PRR's diagram below.. but i also need a negative bias voltage! Where can i tap from to get this while using a voltage quadruppler?

Merlins book has an example for a voltage doubler but i get no negative voltage at that point.. that is tapping from the top of C1 before the diodes..

cheers all!

[craigmillard]

I Have been looking into this a bit more..

Is it possible to use a JCM900 style bias supply:

http://www.freeinfosociety.com/electronics/schemview.php?id=488

so tap off the secondary into a cap then resistor to ground then the reverse diode?

What ya think? or am i way of the mark?

[merlinb]

Is it possible to use a JCM900 style bias supply:

No, but you could do this:

[craigmillard]

Cheers Merlin!:)

So the high negative voltage is due to the first doubler stage?

What would be the bast way to drop the excess voltage, its for a 6v6 amp so ideally 20-50v range?

[PRR]

Use cathode bias. It is simpler and much safer.

Fix-bias comes into its own at HIGH plate voltages, where cathode-bias and reasonable load impedance would melt the tubes. A pair of 6V6 can do 14 Watts with 285V cathode-bias, 25+ Watts with 360V and fix-bias. But 6V6 at 300V (and 30V cathode-bias) is a very fine amp, and in-line with your simple 330V supply.

[amptramp]

As I mentioned on Feb 07, you can get 25 watts using cathode bias or 30 watts using fixed bias with a pair of 6BQ6GT's.  There are pros and cons to self bias.  As the tubes age and lose emission, the reduction in self bias tends to make up for it.  With fixed bias, if the tubes lose emission, they lose gain and power because the bias is not dependent on current.  The only time tubes could be endangered by self bias is if the resistance goes down or if it is bypassed by a capacitor that goes leaky and leakage is temperature sensitive.  (I had a Heathkit 14 watt amp redplate due to a cathode bypass electrolytic that was leaking.  I chopped it out and everything went back to normal.)

If a fixed bias source goes down for any reason, you could have tubes running at zero bias which will kill tubes, output transformers and power transformers in short order.  Transmitters used a clamp tube (which could be made more reliable using transistors) which was held off by bias.  If the bias disappeared, the clamp tube would ground the screen supply and save the high-power parts.

Also if the driver stage provides a signal that is above the fixed bias, the grids will conduct and rectify the signal input, shutting off the output for a duration based on the RC time constant at the lowest frequency.  This happens with both the self-bias and fixed bias, but if the self-bias resistor does not have a bypass cap, then the effect goes away immediately.