Would someone have a look at my new tube psu?

"What about a 12v one? 12v rectified would be 16.8. Wouldn´t this be enought for the regulator? A 12.6 volt one is hard to find for me. "

It used to be the other way around, when 12.6VAC was the readily available trafo, but I guess times have finally changed.

Don't forget diode voltage drops in your estimation.

Yes 12.0VAC would be fine. To see what the reg will see put a load equal to the total filament current across the supply and look at the minimum points of the ripple riding on the d.c. when rectifying 17.0V peak. It might be good to use "real" diodes of the type you intend to use rather than Circuitmaker's default bridge, and you could put a little R in series with the trafo secondary for winding resistance (this is taken into account for a.c. loads by the trafo vendor---in fact a better model takes the real open-circuit voltage of the trafo and then adds resistance).

The regulator is happy with about 2V or more across it.

Note as well that the tolerance of the output voltage for some versions of the 7812 is rather poor, so you can have an output V for some ranging from 11.4 to 12.6 taking all the variables into account.

(BTW also somewhat related: see the latest AudioXpress for a controversial article on tube heater operation)

The "floating" heater reference should NOT have any 230uFd on it! 0.1 or even 0.01uFd to swamp the pFd of heater leakage, but not something that will ZAP the heater insulation when it starts to fail.

> I was thinking about using a 110:48 wired in reverse as 110:252v. 252 rectified will lead me to 352.

NO!!!!

Never put more voltage on a power transformer than it was made for. A 110/120V winding and core will take 120, 125, 130V, but will be running extra warm because the iron saturates and the reactive (useless) current in the winding increases.

Remember that, unlike audio where a little iron saturation is distortion, power transformers are wound to the edge of gross saturation, about 3 times more flux (at rated voltage) than any audio iron would be worked at.

I bet 110V across a 48V winding will NOT give the calculated voltage, but will start to stink in minutes. If it is big enough, fuses will pop.

Cathode followers should be fed AC via properly twisted and routed wires. DC is a waste.

> sucks too much current in the B+ part.

OK, so how much is "too much"? How long is a rope? How high is a B+? Don't make us beg for answers you have at your fingertips. You might get "lucky": your answers may be wrong and Jon might spot your misunderstanding.

If stray clues can be believed: 350-400V at 80 mA is a BIG power supply. A very deluxe home radio with push-pull 6L6, 10 Watt output. If the circuit also has negligible PSRR, you are looking at a lot of filtering. Expect to put 10% to 40% of your raw voltage into the resistors in a C-R-C-R-C filter. Say 20% trial estimate. You want to end up at 350V? Then you start with 120%*350= 420V, with 70V drop at 80mA, about 1K resistance. You seem to be fond of 230uFd caps so start there. First cap is about 500mV ripple. Take 500Ω and 230uFd R-C stage: you get maybe 100:1 attenuation at 120Hz, so you have 5mV. If the audio stages have 10dB PSRR, that's 1mV on the audio output, maybe 60dB signal/hum ratio, poor. Add another stage of 500Ω and 230uFd, hum may be low. (But taking 40dB ripple reduction in one stage requires careful layout.) So that works but at 20% overhead and excess heat. (And capacitors which would have been absurdly large in 1958.) For lower loss, find a choke with less than 500Ω DCR and much more than 500Ω impedance at 120Hz. 1H will do, though 5H or 10H would be better.

I love super-size caps but they are not magic bullets. An economic design will generally take 20dB to 30dB of ripple reduction in each stage. Economic, and also you want multiple stages to REALLY smack-down the 240Hz, 360Hz, 480Hz trash. A single stage reduces high-harmonics of buzz about 6dB per octave. A 3-stage filter, 18dB per octave. Also to really get more than 30dB per stage you have to keep parasitic common resistance very low. Careless wiring will bypass buzz around the fat cap and into your circuit.

> controversial article on tube heater operation

What's to argue? He's shifted the tubes from space charge limited to temperature limited. The equations are in any basic tube theory book. Gain is more linear but falls off bad. In particular, the rise of Gm with rise of current can't happen: the tube is choked for current. Yes, there may be a place where the 2nd harmonic cancels (for some specific operating point). Power output will be reduced. And oxide cathodes can't run this way for long: the oxide is mostly a matrix for the thorium, which lives above its melting point and thus evaporates off the surface. You need to keep the thorium hot so more of it will diffuse up from the bulk oxide and cover the surface. The exact rate seems to have always been a proprietary secret, but going so cold that the cathode becomes temp-limited instead of charge-limited kinda has to be bad.

OTOH, I don't believe a 12AX7 needs a reg-diode to bring it up to 12.6V. 12V is within specs. Several good boxes deliberately ran low-level tubes a volt below nominal to reduce noise. And the 12AX7 cathode is clearly plenty big for the mA or so that can be squeezed past the high-Mu grid without going into positive grid operation.

PRR you should send Ed Dell an emailed letter to the editor w/r to that article, particularly with regard to the thorium stuff.

IIRC the temp-limited operation should also be fairly foul from a noise standpoint, as one gets little or no space-charge smoothing.

When I was at UCLA and fooling about with silicon-target vidicons I tried to enhance performance by running at lower heater currents---with the idea that a lower cathode temp would correspond to lower readout noise. The tubes had so many other problems that the whole exercise was fairly pointless. Then we migrated to self-scanned photodiode arrays and I was able to abandon the vidicons.

sorry to get back to the non-interesting stuff, but I mus thank Mr. PRR very very much. It all looks fine now.

The transformer I will be using has a 313vRMS secondary. So, I calculated 406v peak to peak

I want B+ at 320v, so that it will fit ok for the MB1 and the other circuit also, that´s very close to the MB1.

Current draw will be 40ma.

I have a big choke of 7H at 70mA, 210R. I guess it will work as a 10H at 40mA in this circuit...

406v - 320 = 86... 86v at 40mA = 2k15... 2k15 - 210R = 1k94 ... 1k94 / 4 = 485... So, by some aproximation, I got 470R. Four 470R + 270uF filters and one 10H at 210R + 270uF filter. Should be enought...

The problem is that My 270uF caps are only 400v. They are not very big. I hope they will stand it... Or I´ll have some smoke :?

So, I simulated this:

http://hps.infolink.com.br/rafafredd/b+_low_ripple.gif

Now the ripple looks very low and the current draw looks OK even at SPICE. I´ll build the thing.

Thanks!

BTW, why did you set about re-inventing the wheel when you could have been soldering? Tube pwr supplies have been around since Granny. Just look at any schematic and steal. I have been watching you collect every schematic under the Brazilian sun for 3 years now, so I know your pocket is full!

cj

Yeah, you are right CJ. I just wanted to avoid having to buy a new power transformer, but it seems that it´s the only way to get a well filtered unregulated PSU... I´ll have to go high voltage transformer. I just asked for some quotes locally, so I may have an answer tomorrow about prices. The good thing is that I´ll have three separate secondaries for B+, heater and phantom and probably I´ll go toroid.

Just a little more money and I´m done with this one.