Does the ripple at C2 need to be anywhere near this low with the canceling of power supply hum via the output transformer in push pull? I've modeled some existing designs in PSUD and have gotten 1 to few volts at the B+ to the final tubes.
At this time, I might chip in and point out a safety measure that one of our staunch members suggested, viz put the choke in the earthy side of the power secondaries instead of the 'hot' side! Filter caps then have their +'s connected together and their minus sides going separate ways - one to earth.
This came from SY some time ago, and I still think it is darn good practice.
One further small point: Some posts ago I saw a number of high-voltage zeners used in series to get to several 100V. Just remember that the higher the zener voltage (above 6,8V), the higher the drift with temperature. I once used 3 x 100V types in series, and the drift under-chassis came to 12V total!
[As an aside, a year+ ago I finished a dual 100W amplifier, much the same as the object of this thread, and I am waiting with baited breath to find how it will compare! My topology was different: 4 x 6L6GC per channel and using partial cathode feedback (two 6L6GCs cost about the same here as one KT88). Also I did not have transformer/choke woes, because down here one can have your own wound - or even DIY - exactly to specs, and cheaper than import. But no comment from me on this thread, because I have nothing to add!]
This came from SY some time ago, and I still think it is darn good practice.
One further small point: Some posts ago I saw a number of high-voltage zeners used in series to get to several 100V. Just remember that the higher the zener voltage (above 6,8V), the higher the drift with temperature. I once used 3 x 100V types in series, and the drift under-chassis came to 12V total!
[As an aside, a year+ ago I finished a dual 100W amplifier, much the same as the object of this thread, and I am waiting with baited breath to find how it will compare! My topology was different: 4 x 6L6GC per channel and using partial cathode feedback (two 6L6GCs cost about the same here as one KT88). Also I did not have transformer/choke woes, because down here one can have your own wound - or even DIY - exactly to specs, and cheaper than import. But no comment from me on this thread, because I have nothing to add!]
hey-Hey!!!,
When those Cree SiC diodes came out I spoke to their Tech folks and got instruction *NOT* to series them to increase PIV. They told me parallel to increase current capacity was OK, but not to series them. A full bridge on 550 VAC would call for PIV minimum of root2*550 and since those diods come in 600v or 1k2V, go with the 1k2V rated ones. I've used these in L-C filters feeding them 450VAC with no issue so far...
cheers,
Douglas
I thought this was common knowledge. It seems to be ok to series connect them with normal slow diodes and retain the reverse spike advantage.
That first option with 15mv ripple looks like my best bet. What does everyone think about those cap sizes? I'll wait for any addtional comments before locking down that part of the schematic.
It would appear they fail like a row of end-up dominos instead both at once...though I now wish I had spent more time with that helpful man from Cree to understand exactly why he told me these things.
cheers,
Douglas
That is to be expected - I would hardly think that PIV could be so well-controlled as to be exactly equal between components. In experiments we found it is not.
But does all this include using resistive dividers across them to equalise PIV? I very often see circuits without that (with ordinary diodes, whatever). What would be the problem; very narrow spikes?
Brander,
For much the same reason as one uses dividers across electrolytics in series - to ensure that more or less equal voltages appear across the components. With capacitors, otherwise, voltage will divide according to leakage current. As said above, that is not guaranteed to be similar, only not to exceed a specified maximum value.
With diodes in cut-off the 'reverse' leakage is even less than with capacitors. One might have a larger (reverse) voltage drop across one than another. Since there is ac across them, unequal circuit capacities might also play a role. The values may be quite high - 1 megohm. [Only, in high voltage uses, one needs to remember that resistors also have a maximum voltage rating. This is specified, usually about 400V for 1/2 W types. That means that it is better to make the resistance per diode up from 3 or 4 discreet units in series, taking care in the mounting that other leads do not come too close, as in rubbing against them.]
For much the same reason as one uses dividers across electrolytics in series - to ensure that more or less equal voltages appear across the components. With capacitors, otherwise, voltage will divide according to leakage current. As said above, that is not guaranteed to be similar, only not to exceed a specified maximum value.
With diodes in cut-off the 'reverse' leakage is even less than with capacitors. One might have a larger (reverse) voltage drop across one than another. Since there is ac across them, unequal circuit capacities might also play a role. The values may be quite high - 1 megohm. [Only, in high voltage uses, one needs to remember that resistors also have a maximum voltage rating. This is specified, usually about 400V for 1/2 W types. That means that it is better to make the resistance per diode up from 3 or 4 discreet units in series, taking care in the mounting that other leads do not come too close, as in rubbing against them.]
That's precisely why I suggested snubbed UF5408s on the bridge's ground side and Schottkys on the "hot" side. The 1000 PIV UF5408 is capable of "holding the fort down" by itself. While the Schottkys keep switching noise out of the B+ rail, snubbing the PN diodes prevents switching noise from sneaking into the power trafo and, ultimately, the signal path via the heater supply.
Final schematic for 4-20 KT88 PSU (Excuse the mess, this is my first schematic rendering)
I summit this to the forum for final approval / corrections before I head to Digikey.
This is the final version by Wonderboy with small choke, but bigger cap values. Schoktty diodes are being used.
Recommended caps are Panasonic 500V (snap in). Any problem using that for P2P?
Any "higher end" Cap recommendations? 330uf 500V are hard to find other than the Panasonic.
Any final tweeks on the PS for this "Ultimate Mullard"?
Eli?
I summit this to the forum for final approval / corrections before I head to Digikey.
This is the final version by Wonderboy with small choke, but bigger cap values. Schoktty diodes are being used.
Recommended caps are Panasonic 500V (snap in). Any problem using that for P2P?
Any "higher end" Cap recommendations? 330uf 500V are hard to find other than the Panasonic.
Any final tweeks on the PS for this "Ultimate Mullard"?
Eli?
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
Other than "typos" around CRF, the schematic seems reasonable. You'll have to build and load it down to find out what the voltages really are.
CRF is 1000 pF. and either NPO ceramic or mica. Either material yields a non-polarized part.
Panasonic (Matsushita) makes good 'lytics.
Snap in style parts are fine. Just "sister" some hook up wire to the pins. Mounting the big 'lytics in holes and securing them with clamps is 1 good building method.
Thanks Guy's
I'll consider all mounting options. I've even played with the idea of custom PCB's from PCBExpress, but that may be too adventurous, not to mention the $$$$$. Still it would afford another learning opportunity, and learning is what this build will be about.
I guess that about wraps the PS up until built and further tweeked while under test. I'll hold off on more questions for a bit while digging into my study material, and reading up on old OP transformer threads.
Eli, I may shoot you a PM a little latter on to get your thoughts on the 6GK5 vs the EF86.....
I'll consider all mounting options. I've even played with the idea of custom PCB's from PCBExpress, but that may be too adventurous, not to mention the $$$$$. Still it would afford another learning opportunity, and learning is what this build will be about.
I guess that about wraps the PS up until built and further tweeked while under test. I'll hold off on more questions for a bit while digging into my study material, and reading up on old OP transformer threads.
Eli, I may shoot you a PM a little latter on to get your thoughts on the 6GK5 vs the EF86.....
Last edited:
Geek over in Vendor's Bazaar sells power supply PCBs.
You can do SS or tube rectification, CLC, CRC, etc. If you decide to use this pcb, order the TS-UP caps in the taller/skinnier form factor when given a choice, as the caps spec'd are large and the PCB may get crowded.
http://www.classicvalve.ca/images/CK-25_r1_front.jpg
http://www.classicvalve.ca/docs/CK-25_docs_R1.pdf
If the 6GK5 can be used in place of the EF86, please post some details/circuit mods......
You can do SS or tube rectification, CLC, CRC, etc. If you decide to use this pcb, order the TS-UP caps in the taller/skinnier form factor when given a choice, as the caps spec'd are large and the PCB may get crowded.
http://www.classicvalve.ca/images/CK-25_r1_front.jpg
http://www.classicvalve.ca/docs/CK-25_docs_R1.pdf
If the 6GK5 can be used in place of the EF86, please post some details/circuit mods......