Yeah, and your amp would be as enormous as some of the monster stuff that crazy German guy Mayer builds! He has a power supply bigger than my refrigerator in his equipment closet. What a nut!
GZ34 is rated at 250mA, so they have to be paralleled for big amps as currently discussed. Two per leg will be required for 300mA.
Another option for high current is transmitting rectifiers, but then you will have a lot of heater current to supply, B+ would appear on exposed top caps, and with directly heated rects you can't use only one heater winding in a FWB. This would lead to a LARGE power supply.
Silbatone has put tube rectifiers in the bridge previously(WE422A in our 845 amp). I believe that the preservation of the rare tubes was at issue in this design, since replacements are $1-2k+ each--more for super rare versions. An over-designed silicon bridge with slow turn on might be the most reliable option.
Keep in mind that this is a commercial amp for rich guy domestic use, not for the maniac DIY equipment shed, and it is already the size of a small table and a two-man lift.
If you're building for yourself, anything is game. ALL TUBE is theoretically best in a few ways, and I am sure jc will agree, although he isn't as allergic to ss diodes as some of us. The amp does not sound hard, glarey, bright, hazy, or edgy in any way, so he might be right.
The hash chokes between the secondary winding and bridge should help keep the diode switching spikes from ringing the power trans, as would a small 5kV ceramic cap shunted across the winding. Check with scope, because some L/C values could form a nice resonant circuit with the trans winding and make things worse.
jc would have more to say but he may or may not re-appear on this thread. He shows up in audio forums less than I do.
I used to drop by goodsoundclub to skin The Cat once or twice a year, but he it appears that he kicked me off the board, presumably because I took issue with his foolish yet headstrong critiques of equipment he never heard and knows nothing about...and because I am one of the few who can outlast him.
Hi
No, in the actual circuit two rectifier tubes in parallel are used, placing them in the top half of the bridge, each of them would only carry half the current, since only one of them would be conducting at any time.
The voltage drop would be a a bit higher, but that could be compensated for (if necessary) in the power transformer
Thomas
No, in the actual circuit two rectifier tubes in parallel are used, placing them in the top half of the bridge, each of them would only carry half the current, since only one of them would be conducting at any time.
The voltage drop would be a a bit higher, but that could be compensated for (if necessary) in the power transformer
Thomas
Hi Joe
Actually putting the two tubes in the upper half of the bridge rectifier to create a hybrid bridge would be exactly as big as the way JC designed it. In fact he could have saved two silicon diodes.
JC might have had good reasons to do it this way, maybe because the peak voltage drop would have been a bit higher in a hybrid bridge. Still I think the hybrid bridge would have been the better and more elegant solution.
Of course I would have done a fully tubed bridge
Thomas
Actually putting the two tubes in the upper half of the bridge rectifier to create a hybrid bridge would be exactly as big as the way JC designed it. In fact he could have saved two silicon diodes.
JC might have had good reasons to do it this way, maybe because the peak voltage drop would have been a bit higher in a hybrid bridge. Still I think the hybrid bridge would have been the better and more elegant solution.
Of course I would have done a fully tubed bridge
Thomas
Awesome. Thanks for the additional clarification Joe. The Silbatone amp is a real beauty, but far beyond my budget so I have no plans of building a clone, but the PS topology really peaked my interest.
The current SS shunt PS design for my amp only delivers 100mA, but I want to change the tubes and raise the bias levels so I need more power. I could dial it back to a more reasonable bias levels and limit the current to around 200mA per rail in order to allow it to get away with one tube per rail to then go all tube, but with the hybrid supply I could include a provision to add a second tube per rail if needed.
Would the hybrid PS recover faster than an all tube design or is the recovery time dependent upon the chosen rectifier tube in the supply?
Your comment about protecting rare tubes struck a cord. I'm planning to build the Octode amp that was published in AX, which is really a modded Yundt design with rare NOS matched MWT 6V6GT tubes, so I think I will use this PS design for that build too.
Like I said...a fancy toy for rich guys. Must be nice, right? We don't expect anybody on this board to buy or even build one. I know if I had those tubes, I'd sell them on ebay in a flash and get myself a "new" used Volvo plus good beer for a year and still have enough remaining for an amp build.
OK, I admit we find it fun and engaging to act like overstuffed hedonists and fire up $2000 museum-grade exotic triodes. Who wouldn't? However, I suspect jc himself would secretly rather use sweep tubes from old Russian B&W TVs. I swear the dude has latent Bolshevik tendencies...
Morrison and I are earthy and humble DIY people from the urban streets, but the goal at hand is to expose the luxury "high end" audience to a higher level of execution, so to speak, while entertaining our peeps with some interesting work. Gives us all something to talk about during a slow news cycle.
I'd go with two...maybe two damper diodes? No absolute need to use Mullard GZ34s, unless you are a tube pimp.
Recovery isn't the issue as much as the turn-on spike. Tubes have higher internal impedance and hence yield a much lower Q system. The desired outcome is DC which is as "slow" as it gets...the slower the better!
If the circuit is Class A, the voltage drop across internal resistance of the rectifiers is more or less constant, so it can be factored in from the start...but I think mosfet regulators were discussed and that would further distance us from the problem of equivalent resistance.
The combined resistance of the B+ winding and the tube rectifiers is easier on the filter caps, limiting charging currents, but then you have to think about maximum filter input capacitance rating of your tubes. That is one advantage of solid state.
There are many angles and lots of "studies" and opinions on rectifiers to consider--much of it more sophisticated than my preferred simple-minded approach. I think there is a wide range of reasonable practice here. The thing to avoid or moderate is sharp switching impulses that can twang the transformer and radiate HF trash in all directions. My basic instinct is the same as T. Mayer's...use as many tubes as possible, but many options will "work" just fine.
Now if I could only get my hands on some NOS selenium rectifiers...
No, no Mullard GZ34s for me. I'm pretty good at arc-y spark-y and releasing the magic fairies so I'll stick to JJ 5AR4s.
But I have learned to triple check every component and make sure any part deviations I choose pass the approval of the designer or those far more knowledgeable here on diyA.I'm much better at forming and manipulating the sheet metal.
In my differential P-P there will be MOSFET regulation after the filtering. I'll probably implement the same regulation topology for the big 6V6 P-P speaker amp.
I think Papa has a big stash of those.
hey, i REALLY shouldn't be doing this... but i am already invested in putting this stuff out there. don't want it to be totally misrepresented, yet.
i have done the power supply all different ways and i will say something for symmetry. and something about the old school.
i really like ss rectifiers. they have very little insertion loss and low impedance. this is good for AC to DC power supplies. they have some problems... the energy storage (switching spikes) and change of state characteristics that cause trouble for audio, almost all of which can be solved with money. hexfreds and synchronous mosfet rectifiers are, for me, probably the best "sounding" (wince, argh, choke, retch..., like i said, i shouldn't be doing this) way to rectify an alternating current into a pulsating dc current.
i don't like the hybrid bridge with the tube on top and the SS on the bottom. it creates offset pulses and harmonics. the ss still makes spikes. admittedly, the magnitude of the harmonics is small... but the ripple is complex, not simple. the spikes still face the chokes so it shouldn't be a big deal. but i like symmetry. better with 4 tube rectifiers, and yes they take up a lot of space and heater current. works and sounds fab.
the tube rectifiers on one leg fix everything, you can parallel them for a low drop and a slow turn on. i have no favorite audiophile rectifier. what ever you don't want, i will use.
the chokes on the AC side are sneaky. they do, of course, work as "hash" chokes, as joe says... but at the values i am using, "hash" includes 50-60 Hz. putting them in front of the bridge (and caps) allows them to provide negative regulation. look at it again. making it symmetrical, chokes in all 4 legs, is the be all end all in my book, and very 1928. not cheap or compact either. but both reduces inrush current and restores peak draw. these are AC chokes and require no airgap. you don't need as much inductance in front as you will after, and it isn't critical, although you can calculate it or spice it for best efficiency (lowest cost). the one downside to this approach, just as it would be with a regulator, is the insertion loss. low dcr is even more of an issue with this many chokes, than it is using one. for 1350 volts, i needed 1250 AC. compare this with one choke in the normal place, and 920 AC. my filter outperforms the simple pi filter by combining choke input qualities with better filtering. this is far better than a CLCLC type and can also be used for class AB2 amps... which makes the other transform your fancy amp into a not so fancy compressor.
this is very old school thinking, but using modern parts. sometimes, by the book is, well, by the book. its just that the book got chucked in the garbage where i retrieved it. now i'm using it to rip up the man (NOT!). that's bolshevik engineering. or fight club? both. neither... i did leave the USA.
i have done the power supply all different ways and i will say something for symmetry. and something about the old school.
i really like ss rectifiers. they have very little insertion loss and low impedance. this is good for AC to DC power supplies. they have some problems... the energy storage (switching spikes) and change of state characteristics that cause trouble for audio, almost all of which can be solved with money. hexfreds and synchronous mosfet rectifiers are, for me, probably the best "sounding" (wince, argh, choke, retch..., like i said, i shouldn't be doing this) way to rectify an alternating current into a pulsating dc current.
i don't like the hybrid bridge with the tube on top and the SS on the bottom. it creates offset pulses and harmonics. the ss still makes spikes. admittedly, the magnitude of the harmonics is small... but the ripple is complex, not simple. the spikes still face the chokes so it shouldn't be a big deal. but i like symmetry. better with 4 tube rectifiers, and yes they take up a lot of space and heater current. works and sounds fab.
the tube rectifiers on one leg fix everything, you can parallel them for a low drop and a slow turn on. i have no favorite audiophile rectifier. what ever you don't want, i will use.
the chokes on the AC side are sneaky. they do, of course, work as "hash" chokes, as joe says... but at the values i am using, "hash" includes 50-60 Hz. putting them in front of the bridge (and caps) allows them to provide negative regulation. look at it again. making it symmetrical, chokes in all 4 legs, is the be all end all in my book, and very 1928. not cheap or compact either. but both reduces inrush current and restores peak draw. these are AC chokes and require no airgap. you don't need as much inductance in front as you will after, and it isn't critical, although you can calculate it or spice it for best efficiency (lowest cost). the one downside to this approach, just as it would be with a regulator, is the insertion loss. low dcr is even more of an issue with this many chokes, than it is using one. for 1350 volts, i needed 1250 AC. compare this with one choke in the normal place, and 920 AC. my filter outperforms the simple pi filter by combining choke input qualities with better filtering. this is far better than a CLCLC type and can also be used for class AB2 amps... which makes the other transform your fancy amp into a not so fancy compressor.
this is very old school thinking, but using modern parts. sometimes, by the book is, well, by the book. its just that the book got chucked in the garbage where i retrieved it. now i'm using it to rip up the man (NOT!). that's bolshevik engineering. or fight club? both. neither... i did leave the USA.
Thank you so much for your insight. I have to admit I've become even more smitten with the PS design. While I'm going to go a different route for the 400V split rails on the 'stat headphone amp, I'm going to use your design for my modified Yundt "octode" amp with 6V6 and 7591 output tubes. I was planning on 6922s for the signal tubes, but I may go more esoteric if I can find a good, low cost alternate. I like the D3As, but they've become quite popular lately and are harder to get.
I'd really like to read some of your old books! Care to share a reading list in a PM?
I LOVE it! I'm a 'merican born and raised, but my wife and I desperately want to ex-Pat our family to New Zealand and I'll be going back to college to get a new degree in the near future to help make that happen as soon as she receives a PhD. Very interested in radiology at the moment and we're seriously looking into a stint in Vancouver, Canada if we can make it happen sooner. We'll definitely holiday in a few locations around the world before we finalize our plans.EDIT: And along the lines of old school with modern parts or technique mods, I'm going to fabricate the chassis for my Octode in Art Deco style with strong Modern Streamline influences. I'll be using an old GE turntable as the basis of the design inspiration, a large tear drop shape when viewed from above with incredible softened waterfall sides. The transformer covers will be highly polished skyscraper style and I'll mill or turn custom knobs, have them electropolished, then anodized.
Last edited:
the 2 H chokes are separate because this is a TRANSMITTING tube power supply. 1750 ppV ac! you can screw around with your life, but mine is protected with 1/4" of solid linen phenolic, some teflon, and then a massive billet aluminum chassis. i like amps that can kill you, but i'm not that into death.
for a preamp, you can use an isolation transformer (1:1) "turned" sideways, for the AC side chokes. works great and the coupling further diminishes ripple and restores drops. cheap iron buzzes a little more but trannies buzz. no biggie.
you can do this for heaters, with ferrite core dual bobbins used for switching supplies between the regulator and the tubes... cleans up things mo...
i reuse trannies all the time. stuff found in the garbage has mojo and doesn't cost much! it does whittle away at your partner's lifestyle and sense of right and wrong, tho. nothing beats down a relationship like old transformers, except maybe old test equipment.
yes the batteries are the silba way to do it! but there is a limit to the insanity, of course. 110 gel cells need to be lifted. and also charged... not by me or joe.
although, i do not any longer put it past jrob or chung to come up with the most kicked out projects. those boys is crazy.
for a preamp, you can use an isolation transformer (1:1) "turned" sideways, for the AC side chokes. works great and the coupling further diminishes ripple and restores drops. cheap iron buzzes a little more but trannies buzz. no biggie.
you can do this for heaters, with ferrite core dual bobbins used for switching supplies between the regulator and the tubes... cleans up things mo...
i reuse trannies all the time. stuff found in the garbage has mojo and doesn't cost much! it does whittle away at your partner's lifestyle and sense of right and wrong, tho. nothing beats down a relationship like old transformers, except maybe old test equipment.
yes the batteries are the silba way to do it! but there is a limit to the insanity, of course. 110 gel cells need to be lifted. and also charged... not by me or joe.
although, i do not any longer put it past jrob or chung to come up with the most kicked out projects. those boys is crazy.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.