Step 6: The Power Supply
Traditional tube amp power supplies are old school--relatively high voltage, with big "iron," and generally not regulated. Typically, they supply a range of voltages for different purposes--a current source for the output transformer, voltages for the preamp tube plates, and sometimes (in this case) a separate voltage for the pentode screens.
Unlike regulated supplies, the different supply voltages are created with current-limiting resistors. These are often called "voltage-dropping resistors," but their operation depends on the current draw of each stage.
Designing a power supply
The first step is choosing the right power transformer (see the "How did this project get started?" section.) To pick the right transformer, look at the data sheets for the power tubes.
The 6DG6GT tubes have a max plate voltage of 200V. Theoretically, an AC RMS voltage is ~0.7 of the peak voltage, and the peak is approx 1.414 * the RMS. In practice it's lower--the transformer is under load, there are losses in the caps, etc. So something less than 1.4 is more realistic. (Gotta dig that crazy square root of 2...that 1.414 number pops up in so many places!)
I'm not certain about the availability of PTs with secondaries in the 125-150V range. But maybe the 6DG6GT can handle somewhat more than 200V. Another alternative is to use a "choke input" power supply--that's connecting the choke FIRST, before any filtering cap. A choke input should drop the secondary voltage to 0.9 of the RMS (vs the 1.414 for a standard filter), so a 225V RMS AC secondary yields 202.5 VDC, also excellent.
My "recycled" transformer was ~140V (142) RMS AC, which, when rectified, (in an ideal world) becomes 200.788 peak (VDC)--perfect! (in practice--rectified, filtered and loaded, it's about 190V, still excellent.)
The solid-state rectifying bridge was chosen over a tube rectifier to retain as much of that voltage as possible. That's OK--the much vaunted "sag" effect of tube amps doesn't apply to single-ended, Class A amps. They draw the same amount of current whether there's an input signal or not... Also, the PT doesn't have a centertap, so unless I used two tube rectifiers (or went with a half-wave design), solid-state was the best solution.
These voltages were needed by the circuitry:
B.1 : 190V -- Max voltage for the power tube plates/output transformers
B.2 : 180V -- A tap for the preamp tubes (Added during build)
B.3 : 120V -- Screen voltage for the 6DG6GT power tubes (between 115-125V, depending on the data sheet)
I did the initial design using an excellent (free) design tool: Duncan Amps PSUD2 Designer
The final result varied quite a bit from the simulation in PSU designer, however. That could be related to the unknown current-suppling potential of the TV transformer--but I'm beginning to suspect that the 6DG6GT screens draw much less current than noted noted on the data sheets...
A Redesign Partway Through the Project...
The design evolved. Initially the first filter stage was an RC (Resistance-Capacitance) filter, but that changed quickly. To get a clean signal, I'd need to insert something like a 50 ohm, 20 watt resistor. But when I saw the amount of current wasted, I balked, and changed to an LC (Inductance-Capacitance) filter design.
Also a significant change--there was no B.2 supply at all, originally. I had planned that the preamp would run from the lower screen voltage (120V.) For the 12AX7, that's a pretty low operating voltage. So the preamp supply was added.
The Inductor for the LC Filter
It helped that the gutted TV also included a (mighty big) inductor. It's an unknown value (inductors are measured in Henries), but it was matched with the TV power trannie, so I was sure it would work--and it did. And honestly, an LC filter does a much more efficient job of smoothing out the supply ripples in a single stage than an RC filter does.
Incidentally, it was the addition of the the LC filter (pi filter) that prompted me to add the standby switch--the initial inductance spike exceeds the 200V max of the 6DG6GT's, by a fair amount. But during the testing phase the switch wasn't wired. There have been no negative consequences and I'm not sure the standby will be wired in. It's kind of silly, really--NOS tubes were often run at 150% of their rated voltage, so a short spike at startup wouldn't amount to much...
Also changed--originally, the preamp plate supply was slated to run on the same voltage as the screens. But it made sense to run the preamp at a higher voltage. So an additional RC stage (B.2) was added:
Preamp Supply
Preamp supply (B.2): As noted, this section was inserted AFTER the first version was built. I started with a 220 ohm resistor for the RC filter, but settled on a 1K value for a smoother supply. 1K didn't drop the voltage much at all (which had become obvious before when building the screen supply.) It would be nice to run the preamp tube directly off the B.1 supply, but preamps need something less noisy...
Screen Supply
Screen supply (B.3): Originally the second section of a two section PSU; in real operation it didn't match the Duncan PSUD2 software very closely. The simulator estimated the resistor for the last RC filter at 2.7K - 3.3K. But during the build the screen voltage was much too high--over 170V. with substitution, the eventual 15k value was chosen, which placed the screens at a nice 120V. A 20K resistor would probably work just as well... Surprisingly, the amp still functioned (poorly) with the initial high screen voltages, and the tubes weren't damaged. Vacuum tubes are amazingly forgiving of abuse...
Misc
The PS voltage-dropping resistors are all 5 watt, although a 3 watt type would have been fine for the B.3 section (15k.)
Regarding capacitance values, perhaps four 100uf caps are overkill, but they do the job. 100uF would be too high for a tube rectifier, but isn't a problem with the SS bridge.
No "bleed resistor" has been installed. One quirk of this amp--the PS caps seem to drain through (cathodes to screens) the 6DG6GT tubes, possibly due to the very hot filaments. They keep the tube internals warm enough after power-down that the tube keeps functioning for a second or two. I don't know this for sure, but when I was experimenting with "triode mode" for the power tubes (screens not connected to main B.3), the caps were NOT draining.
Regardless, ALWAYS check the filter caps before touching the internals.
Like the whole build, the power supply's appearance is a bit inelegant, but it was modified several times during the project... Eventually it should be disassembled and reassembled in a sensible fashion.
I've included a PDF on toroid transformer construction, for the adventurous...
Unlike regulated supplies, the different supply voltages are created with current-limiting resistors. These are often called "voltage-dropping resistors," but their operation depends on the current draw of each stage.
Designing a power supply
The first step is choosing the right power transformer (see the "How did this project get started?" section.) To pick the right transformer, look at the data sheets for the power tubes.
The 6DG6GT tubes have a max plate voltage of 200V. Theoretically, an AC RMS voltage is ~0.7 of the peak voltage, and the peak is approx 1.414 * the RMS. In practice it's lower--the transformer is under load, there are losses in the caps, etc. So something less than 1.4 is more realistic. (Gotta dig that crazy square root of 2...that 1.414 number pops up in so many places!)
I'm not certain about the availability of PTs with secondaries in the 125-150V range. But maybe the 6DG6GT can handle somewhat more than 200V. Another alternative is to use a "choke input" power supply--that's connecting the choke FIRST, before any filtering cap. A choke input should drop the secondary voltage to 0.9 of the RMS (vs the 1.414 for a standard filter), so a 225V RMS AC secondary yields 202.5 VDC, also excellent.
My "recycled" transformer was ~140V (142) RMS AC, which, when rectified, (in an ideal world) becomes 200.788 peak (VDC)--perfect! (in practice--rectified, filtered and loaded, it's about 190V, still excellent.)
The solid-state rectifying bridge was chosen over a tube rectifier to retain as much of that voltage as possible. That's OK--the much vaunted "sag" effect of tube amps doesn't apply to single-ended, Class A amps. They draw the same amount of current whether there's an input signal or not... Also, the PT doesn't have a centertap, so unless I used two tube rectifiers (or went with a half-wave design), solid-state was the best solution.
These voltages were needed by the circuitry:
B.1 : 190V -- Max voltage for the power tube plates/output transformers
B.2 : 180V -- A tap for the preamp tubes (Added during build)
B.3 : 120V -- Screen voltage for the 6DG6GT power tubes (between 115-125V, depending on the data sheet)
I did the initial design using an excellent (free) design tool: Duncan Amps PSUD2 Designer
The final result varied quite a bit from the simulation in PSU designer, however. That could be related to the unknown current-suppling potential of the TV transformer--but I'm beginning to suspect that the 6DG6GT screens draw much less current than noted noted on the data sheets...
A Redesign Partway Through the Project...
The design evolved. Initially the first filter stage was an RC (Resistance-Capacitance) filter, but that changed quickly. To get a clean signal, I'd need to insert something like a 50 ohm, 20 watt resistor. But when I saw the amount of current wasted, I balked, and changed to an LC (Inductance-Capacitance) filter design.
Also a significant change--there was no B.2 supply at all, originally. I had planned that the preamp would run from the lower screen voltage (120V.) For the 12AX7, that's a pretty low operating voltage. So the preamp supply was added.
The Inductor for the LC Filter
It helped that the gutted TV also included a (mighty big) inductor. It's an unknown value (inductors are measured in Henries), but it was matched with the TV power trannie, so I was sure it would work--and it did. And honestly, an LC filter does a much more efficient job of smoothing out the supply ripples in a single stage than an RC filter does.
Incidentally, it was the addition of the the LC filter (pi filter) that prompted me to add the standby switch--the initial inductance spike exceeds the 200V max of the 6DG6GT's, by a fair amount. But during the testing phase the switch wasn't wired. There have been no negative consequences and I'm not sure the standby will be wired in. It's kind of silly, really--NOS tubes were often run at 150% of their rated voltage, so a short spike at startup wouldn't amount to much...
Also changed--originally, the preamp plate supply was slated to run on the same voltage as the screens. But it made sense to run the preamp at a higher voltage. So an additional RC stage (B.2) was added:
Preamp Supply
Preamp supply (B.2): As noted, this section was inserted AFTER the first version was built. I started with a 220 ohm resistor for the RC filter, but settled on a 1K value for a smoother supply. 1K didn't drop the voltage much at all (which had become obvious before when building the screen supply.) It would be nice to run the preamp tube directly off the B.1 supply, but preamps need something less noisy...
Screen Supply
Screen supply (B.3): Originally the second section of a two section PSU; in real operation it didn't match the Duncan PSUD2 software very closely. The simulator estimated the resistor for the last RC filter at 2.7K - 3.3K. But during the build the screen voltage was much too high--over 170V. with substitution, the eventual 15k value was chosen, which placed the screens at a nice 120V. A 20K resistor would probably work just as well... Surprisingly, the amp still functioned (poorly) with the initial high screen voltages, and the tubes weren't damaged. Vacuum tubes are amazingly forgiving of abuse...
Misc
The PS voltage-dropping resistors are all 5 watt, although a 3 watt type would have been fine for the B.3 section (15k.)
Regarding capacitance values, perhaps four 100uf caps are overkill, but they do the job. 100uF would be too high for a tube rectifier, but isn't a problem with the SS bridge.
No "bleed resistor" has been installed. One quirk of this amp--the PS caps seem to drain through (cathodes to screens) the 6DG6GT tubes, possibly due to the very hot filaments. They keep the tube internals warm enough after power-down that the tube keeps functioning for a second or two. I don't know this for sure, but when I was experimenting with "triode mode" for the power tubes (screens not connected to main B.3), the caps were NOT draining.
Regardless, ALWAYS check the filter caps before touching the internals.
Like the whole build, the power supply's appearance is a bit inelegant, but it was modified several times during the project... Eventually it should be disassembled and reassembled in a sensible fashion.
I've included a PDF on toroid transformer construction, for the adventurous...
Winding Toroids.pdf(576x782) 305 KB
Remove these ads by
Signing Up
Visit Our Store »
Go Pro Today »
The white boxes with numbers on the side, what are those and what purpose do they serve?
Thanks.
The white things are power resistors, and are part of the power supply. It's necessary to drop voltage for various stages in a tube amp, and it's almost always done via "brute force" (resistors).
Thanks.
JB
Swap out the tubes for known good ones? I've learned that lesson from bitter experience (sometimes my backup tubes weren't any good either, and I wasted weeks replacing stuff that was fine).
I really appreciate you help.
JB
Oh, and can describe the test I can run for this and the other resistors with a DMM?
You can easily test those resistors that can be isolated from the circuit by pulling the tubes--the cathode bias resistors. Most of the other will need to have one leg disconnected.
Old resistors can still test good when cold, but may drift more when they get hot....
If it were me, I'd check the voltages in the PS, and the tube plates. This is DANGEROUS! Deadly high voltage! Don't do it unless you're confident working with HV.
Don't forget to clean and re-tension all the tube sockets.