pstamler
Well-known member
Hi folks:
I promised, in the thread on Black Market about the power supply board I'm offering as a group puirchase, to do a worked example. The schematic is HERE; please refer to that for part and pin references. Get a cup of coffee before you start, 'cause this took longer than I expected.
Say the project is a small, Champ-class guitar amplifier. (It could as easily be a compressor or EQ or whatever.) Let's define what we want:
B+1: Output tube plate - 280V 38.5mA
B+2: Output tube screen - 262V 4.25mA
B+3: Driver tube - 200V 0.45mA
B+4: Input tube - 172V 0.45mA plus 1mA into bleeder resistor
We'll do cascaded filtering, 1 feeds 2 feeds 3 feeds 4. So we will use J6, not J5. We'll use a standard star ground, so we'll install J2-4, and connect the grounds for the various audio stages to G1-4. For the moment, we will ignore how we get B+1 to the right voltage and simply do the other stages.
Cumulatively, here's what each stage is carrying:
B+!: 280V 44.55mA (includes draw of B+2, B+3, B+4 and bleeder)
B+2: 262V 6.15mA (includes draw of B+3, B+4 and bleeder)
B+3: 200V 1.9mA (includes draw of B+4 and bleeder)
B+4: 172V 1.45mA (includes draw of bleeder)
Fil: 62V 1mA (part of bleeder string - used for elevating filament winding to cut down hum)
Okay, now we can calculate the resistors using the voltage drops between the pins:
(B+1) - (B+2) = 18V; divide by 6.15mA and you get R9 = 2927 ohms. Use 3k; this resistor will dissipate (6.15)^2 x 3000 = 0.113W
(B+2) - (B+3) = 62V; divide by 1.9mA and you get about R10 = 32.6k. Use 33k; it will dissipate 0.119W.
(B+3) - (B+4) = 28V; divide by 1.45mA and you get R11 = 19.3k. Use 20k; this resistor will dissipate 0.042W. (The dissipation is so small that you could use a 1/4W 1% resistor, 19.1k, and be closer to the ideal value.)
(B+4) - (Fil) = 110V; divide by 1mA and you get R12 = 110k; dissipation = 0.11W.
So it looks like you could use half-watt resistors for everything with plenty of safety margin. The board has room for up to 5W resistors.
What about the big electrolytics? Despite what the book says, I've found that an 80uF 450V cap as the first filter capacitor after a 6X4 rectifier tube works just fine; at least, there's been one in my Kalamazoo amp since before I bought it, and I've been chugging away on it for four years without wearing out a 6X4. So let's use that. Well, 82uF if you want to get technical. I like the Panasonic TSHB series, high reliability and a 105 degree rating.
What will the ripple be like? There's a useful ripple calculator HERE. It requires you to enter the capacitance and the equivalent load resistance. The latter is the voltage at B+1 divided by the total current draw, or 280V/0.0445A, or 6285 ohms. (Note that the calculator wants you to enter this in kohms!) We also need the peak input voltage; for the moment, let's just call that 282V, which will give us a final output of about 280V, and a ripple of about 1.3V.
[Does that seem high? The output transformer is about 35:1, so even if the whole ripple voltage appeared across the primary -- which it won't -- it would get stepped down to about 37mV at the speaker terminals, or about 38dB below the 1W level. For a guitar amp that's pretty reasonable.]
Let's make the other three caps 47uF. That size doesn't appear in the TSHB series, but there's a 47uF 400V cap in the TSHA series, also good. How much will they reduce ripple? The formula for ripple reduction in a 60Hz power system is:
Reduction = 1 / (240 x pi x R x C)
where R is in ohms and C is in farads. (For 50Hz power, change 240 to 200.)
For B+2, R = 3k and C = 47uF; ripple will be reduced by a factor of .00941x. So the ripple on the screen is 1.3 x .0094 = 12.2mV.
For B+3, R = 33k and C = 47uF, so ripple is further reduced by .000855x, to about 10.5uV.
And for B+4, R = 20k and C = 47uF, so ripple is further reduced by .00141x, to about 14.8nV. For reference, that's about -154dBu. Cascaded RC filters are powerful gadgets.
That's about it, except that we need to get B+1 up to 282V in the first place. With a 6X4 it's pretty simple; look at the chart in the RCA manual, and you find that for about 45mA draw you want a transformer with something like 260-0-260V on its plate winding, and indeed the Kalamazoo has a tranny of 262-0-262V. Nice when theory and practice come together. In that instance, you'd leave out all the solid-state stuff, and the snubbers, and indeed everything from R8 on leftward. The 6X4 would connect directly to the B+1 terminal.
If you wanted to use solid-state diodes, you could use a tranny with 210-0-210V. That'd give you about 297V, minus 2.5V for the diode drops, or 294.5V. A 270 ohm resistor at R8 will drop it down to about 282V, but that resistor will dissipate about 0.53W. It's a 2W resistor, so it can take it, but you should probably mount R8 half an inch or so above the board for cooling's sake. Or use a smaller resistor (say, 100 ohms) and live with a slightly higher voltage.
With a full-wave center-tapped arrangement, you'd use R1-R2 and C1-C2 for snubbers; you'd ignore C3, which is for a full-wave bridge arrangement. For information on how to calculate snubber values, see Jim Hagerman's article HERE.
The bypass capacitors could be 0.1uF/630V or 0.1uF/400V Panasonic ECQ-P polypropylenes. The snubber caps could be Epcot MKPs or 1kV ceramic discs.
That's a quick tour of how you'd fill in the blanks in the power supply parts list for a typical project. Well, fairly quick.
Peace,
Paul
I promised, in the thread on Black Market about the power supply board I'm offering as a group puirchase, to do a worked example. The schematic is HERE; please refer to that for part and pin references. Get a cup of coffee before you start, 'cause this took longer than I expected.
Say the project is a small, Champ-class guitar amplifier. (It could as easily be a compressor or EQ or whatever.) Let's define what we want:
B+1: Output tube plate - 280V 38.5mA
B+2: Output tube screen - 262V 4.25mA
B+3: Driver tube - 200V 0.45mA
B+4: Input tube - 172V 0.45mA plus 1mA into bleeder resistor
We'll do cascaded filtering, 1 feeds 2 feeds 3 feeds 4. So we will use J6, not J5. We'll use a standard star ground, so we'll install J2-4, and connect the grounds for the various audio stages to G1-4. For the moment, we will ignore how we get B+1 to the right voltage and simply do the other stages.
Cumulatively, here's what each stage is carrying:
B+!: 280V 44.55mA (includes draw of B+2, B+3, B+4 and bleeder)
B+2: 262V 6.15mA (includes draw of B+3, B+4 and bleeder)
B+3: 200V 1.9mA (includes draw of B+4 and bleeder)
B+4: 172V 1.45mA (includes draw of bleeder)
Fil: 62V 1mA (part of bleeder string - used for elevating filament winding to cut down hum)
Okay, now we can calculate the resistors using the voltage drops between the pins:
(B+1) - (B+2) = 18V; divide by 6.15mA and you get R9 = 2927 ohms. Use 3k; this resistor will dissipate (6.15)^2 x 3000 = 0.113W
(B+2) - (B+3) = 62V; divide by 1.9mA and you get about R10 = 32.6k. Use 33k; it will dissipate 0.119W.
(B+3) - (B+4) = 28V; divide by 1.45mA and you get R11 = 19.3k. Use 20k; this resistor will dissipate 0.042W. (The dissipation is so small that you could use a 1/4W 1% resistor, 19.1k, and be closer to the ideal value.)
(B+4) - (Fil) = 110V; divide by 1mA and you get R12 = 110k; dissipation = 0.11W.
So it looks like you could use half-watt resistors for everything with plenty of safety margin. The board has room for up to 5W resistors.
What about the big electrolytics? Despite what the book says, I've found that an 80uF 450V cap as the first filter capacitor after a 6X4 rectifier tube works just fine; at least, there's been one in my Kalamazoo amp since before I bought it, and I've been chugging away on it for four years without wearing out a 6X4. So let's use that. Well, 82uF if you want to get technical. I like the Panasonic TSHB series, high reliability and a 105 degree rating.
What will the ripple be like? There's a useful ripple calculator HERE. It requires you to enter the capacitance and the equivalent load resistance. The latter is the voltage at B+1 divided by the total current draw, or 280V/0.0445A, or 6285 ohms. (Note that the calculator wants you to enter this in kohms!) We also need the peak input voltage; for the moment, let's just call that 282V, which will give us a final output of about 280V, and a ripple of about 1.3V.
[Does that seem high? The output transformer is about 35:1, so even if the whole ripple voltage appeared across the primary -- which it won't -- it would get stepped down to about 37mV at the speaker terminals, or about 38dB below the 1W level. For a guitar amp that's pretty reasonable.]
Let's make the other three caps 47uF. That size doesn't appear in the TSHB series, but there's a 47uF 400V cap in the TSHA series, also good. How much will they reduce ripple? The formula for ripple reduction in a 60Hz power system is:
Reduction = 1 / (240 x pi x R x C)
where R is in ohms and C is in farads. (For 50Hz power, change 240 to 200.)
For B+2, R = 3k and C = 47uF; ripple will be reduced by a factor of .00941x. So the ripple on the screen is 1.3 x .0094 = 12.2mV.
For B+3, R = 33k and C = 47uF, so ripple is further reduced by .000855x, to about 10.5uV.
And for B+4, R = 20k and C = 47uF, so ripple is further reduced by .00141x, to about 14.8nV. For reference, that's about -154dBu. Cascaded RC filters are powerful gadgets.
That's about it, except that we need to get B+1 up to 282V in the first place. With a 6X4 it's pretty simple; look at the chart in the RCA manual, and you find that for about 45mA draw you want a transformer with something like 260-0-260V on its plate winding, and indeed the Kalamazoo has a tranny of 262-0-262V. Nice when theory and practice come together. In that instance, you'd leave out all the solid-state stuff, and the snubbers, and indeed everything from R8 on leftward. The 6X4 would connect directly to the B+1 terminal.
If you wanted to use solid-state diodes, you could use a tranny with 210-0-210V. That'd give you about 297V, minus 2.5V for the diode drops, or 294.5V. A 270 ohm resistor at R8 will drop it down to about 282V, but that resistor will dissipate about 0.53W. It's a 2W resistor, so it can take it, but you should probably mount R8 half an inch or so above the board for cooling's sake. Or use a smaller resistor (say, 100 ohms) and live with a slightly higher voltage.
With a full-wave center-tapped arrangement, you'd use R1-R2 and C1-C2 for snubbers; you'd ignore C3, which is for a full-wave bridge arrangement. For information on how to calculate snubber values, see Jim Hagerman's article HERE.
The bypass capacitors could be 0.1uF/630V or 0.1uF/400V Panasonic ECQ-P polypropylenes. The snubber caps could be Epcot MKPs or 1kV ceramic discs.
That's a quick tour of how you'd fill in the blanks in the power supply parts list for a typical project. Well, fairly quick.
Peace,
Paul
