Power transformer info and order details are attached. If anyone wants to duplicate the order just see below.
I'll upload full circuit and PCB details soon for review. The plan is a separate PS board, two filament boards, and a driver board. They all fit together like a nice puzzle. Designs are 99% complete. I've already posted some drafts and things have changed very little since then.
Power Supply details follow: The input tubes draw 1.8A total so that 6.3VAC circuit is sized to 3.6A. The two filament circuits draw 1.2A each so they are sized at 5VAC and 2.4A each. My low volt circuits use a super low dropout rectifier and schottky diodes. 5VAC ends up at 5VDC and 6.3AC ends up at 6.3VDC. No need to start higher. For the HV I went with 400-0-100-400 @ 250ma. I already designed a doubler for the bias circuit and may or may not keep it. It's easy to wire it up with the doubler disabled. My thought of keeping it are that I'll end up with some extra boards and someone may want the doubler if they're using a stock PS with a 50V tap. Anyway, more on the final circuits in a few days when I post them.
Also, I just received some Audio Note 0.47uF 630Vdc Copper Foil Series Mylar Oil Capacitors. I know everyone has favorite caps, but I'm locked into these at this point.
I'm looking hard at the Khozmo 51k Stepped Attenuator on the input instead of the recommended 40k resistors from the Electra-Print schematic. Room in the chassis could be a consideration. If someone sees logic for a value other than 51k please let me know your thoughts.
I'll upload full circuit and PCB details soon for review. The plan is a separate PS board, two filament boards, and a driver board. They all fit together like a nice puzzle. Designs are 99% complete. I've already posted some drafts and things have changed very little since then.
Power Supply details follow: The input tubes draw 1.8A total so that 6.3VAC circuit is sized to 3.6A. The two filament circuits draw 1.2A each so they are sized at 5VAC and 2.4A each. My low volt circuits use a super low dropout rectifier and schottky diodes. 5VAC ends up at 5VDC and 6.3AC ends up at 6.3VDC. No need to start higher. For the HV I went with 400-0-100-400 @ 250ma. I already designed a doubler for the bias circuit and may or may not keep it. It's easy to wire it up with the doubler disabled. My thought of keeping it are that I'll end up with some extra boards and someone may want the doubler if they're using a stock PS with a 50V tap. Anyway, more on the final circuits in a few days when I post them.
Also, I just received some Audio Note 0.47uF 630Vdc Copper Foil Series Mylar Oil Capacitors. I know everyone has favorite caps, but I'm locked into these at this point.
I'm looking hard at the Khozmo 51k Stepped Attenuator on the input instead of the recommended 40k resistors from the Electra-Print schematic. Room in the chassis could be a consideration. If someone sees logic for a value other than 51k please let me know your thoughts.
Attachments
PLEASE TAKE A LOOK:
I'm very open to feedback. Ideally, I'm not being sent back to the drawing board with an idea from left field. So please frame things in a way that gets me to the finish line. I did receive some feedback early on about inrush from the filaments. I don't see the vast majority of designs trying to fix that. However, I am open to options that could fit on the filament board. That one isn't very densly populated. There's also some unused real estate on the driver board if needed.
The driver board is 11"X3", power supply board is 7"X4", and the filament boards are 2"X4". Arranged, everything is 11"X7". All sockets are on the boards so this is a tidy setup and will it be a really clean build inside the chassis. Everything comes to the boards for wiring. I went with separate boards in case something needs to be tweaked. I'd rather not pull out the whole thing and start over. Also, I might do another version of the driver board to work with a traditional 6SN7 design and cathode bias. For now I'm going with something slightly unconventional, at least when it comes to the driver tubes (6SL7 & 6BL7), cathode follower, and fixed bias. There are no shortages of 6BL7 NOS tubes on Ebay for cheap so no worries there.
The board designs are pretty clean. No vias. Full ground plane for the filament board. There's a split ground plane for the power supply board. The driver board was harder than I thought it would be. If each channel took over a single 6SN7, then it would have been much easier. Using half of each driver tube for each channel made routing and layout a little bit of work even with all the unused space.
Please, someone look at my bias method. This is something you'd see on a push-pull. I'll have three test points and two pots that are all available from the top. Multimeter leads will go into the left and right positive to zero them out with the pot on the driver board. Then one of the leads will go into the ground to bring the average bias to the desired state with the pot on the power supply board. I have this on some KT88 monoblocks and it works nicely. I can bias them in about 60 seconds. I've never seen this for a SET stereo, but why not. Or maybe it's a horrible idea.
I'm very open to feedback. Ideally, I'm not being sent back to the drawing board with an idea from left field. So please frame things in a way that gets me to the finish line. I did receive some feedback early on about inrush from the filaments. I don't see the vast majority of designs trying to fix that. However, I am open to options that could fit on the filament board. That one isn't very densly populated. There's also some unused real estate on the driver board if needed.
The driver board is 11"X3", power supply board is 7"X4", and the filament boards are 2"X4". Arranged, everything is 11"X7". All sockets are on the boards so this is a tidy setup and will it be a really clean build inside the chassis. Everything comes to the boards for wiring. I went with separate boards in case something needs to be tweaked. I'd rather not pull out the whole thing and start over. Also, I might do another version of the driver board to work with a traditional 6SN7 design and cathode bias. For now I'm going with something slightly unconventional, at least when it comes to the driver tubes (6SL7 & 6BL7), cathode follower, and fixed bias. There are no shortages of 6BL7 NOS tubes on Ebay for cheap so no worries there.
The board designs are pretty clean. No vias. Full ground plane for the filament board. There's a split ground plane for the power supply board. The driver board was harder than I thought it would be. If each channel took over a single 6SN7, then it would have been much easier. Using half of each driver tube for each channel made routing and layout a little bit of work even with all the unused space.
Please, someone look at my bias method. This is something you'd see on a push-pull. I'll have three test points and two pots that are all available from the top. Multimeter leads will go into the left and right positive to zero them out with the pot on the driver board. Then one of the leads will go into the ground to bring the average bias to the desired state with the pot on the power supply board. I have this on some KT88 monoblocks and it works nicely. I can bias them in about 60 seconds. I've never seen this for a SET stereo, but why not. Or maybe it's a horrible idea.
Attachments
Big mistake in first stage anode load!
220k is not the part of PSU, but 6SL7 anode load resistor, so if you use C19 capacitor, the anode AC shunted to gnd!
The anode load resistor must be close to the socket pin 2 to prevent oscillations.
BTW what's the role of TR1 and R6?
It generates crosstalk between channels, but does nothing with fixed bias DC voltages of channels.
p.s.
While not use individual bias for each channels?
Even "paired" new tubes requiring different fixed bias voltage, the used tube are mainly.
220k is not the part of PSU, but 6SL7 anode load resistor, so if you use C19 capacitor, the anode AC shunted to gnd!
The anode load resistor must be close to the socket pin 2 to prevent oscillations.
BTW what's the role of TR1 and R6?
It generates crosstalk between channels, but does nothing with fixed bias DC voltages of channels.
p.s.
While not use individual bias for each channels?
Even "paired" new tubes requiring different fixed bias voltage, the used tube are mainly.
Thanks for the feedback. I made the 220k and C19 change after seeing the SPICE simulation. I'll undo those changes. Same thoughts with R11 and C15 that I added to the power supply?
TR1 and R6 is the bias approach I borrowed from the KT88 Poseidon I have. I'll upload the docs for your review. I thought it was a risky approach. I'll look at individual bias for each channel.
TR1 and R6 is the bias approach I borrowed from the KT88 Poseidon I have. I'll upload the docs for your review. I thought it was a risky approach. I'll look at individual bias for each channel.
Last edited: