Hello,
I'm working on an Audio Note P2 SE amplifier which is a parallel single ended class A amp based on 6L6GB tubes (2 per channel).
Each 6L6 is cathode biased to 65mA, so all in the complete PSU current draw including the driver (2x6SL7GT) is approximately 300mA.
The original power supply is CRCRC type (schematics attached). With last RC leg doubled, one per output transformer.
.
Previous owner has upgraded the C1/C2/C3/C4 to Epcos 330uF/450V which is not bad per se, as these are extremely durable and high ripple current caps. As simulated with PSUD2, the C1 sees a ripple of 1.1A. The upgraded capacitance also leads to decreased overall ripple (90mV p-p) over the original 220uF capacitors (180mV p-p).
I'm toying with the idea of changing the C3 and C4 capacitors to film caps, Mundorf Tubecap 100uF (larger wont fit).
As simulated with PSUD, the effective ripple in 330uF+330uF+100uF configuration will be identical to originally specified 220uF+220uF+220uF, so that should be OK.
Can I expect any negative consequences from a reduction of capacitance going to the output transformers (and therefore 6L6 plates)? Is there a reason why these capacitors should be of larger value?
I'm working on an Audio Note P2 SE amplifier which is a parallel single ended class A amp based on 6L6GB tubes (2 per channel).
Each 6L6 is cathode biased to 65mA, so all in the complete PSU current draw including the driver (2x6SL7GT) is approximately 300mA.
The original power supply is CRCRC type (schematics attached). With last RC leg doubled, one per output transformer.
. Previous owner has upgraded the C1/C2/C3/C4 to Epcos 330uF/450V which is not bad per se, as these are extremely durable and high ripple current caps. As simulated with PSUD2, the C1 sees a ripple of 1.1A. The upgraded capacitance also leads to decreased overall ripple (90mV p-p) over the original 220uF capacitors (180mV p-p).
I'm toying with the idea of changing the C3 and C4 capacitors to film caps, Mundorf Tubecap 100uF (larger wont fit).
As simulated with PSUD, the effective ripple in 330uF+330uF+100uF configuration will be identical to originally specified 220uF+220uF+220uF, so that should be OK.
Can I expect any negative consequences from a reduction of capacitance going to the output transformers (and therefore 6L6 plates)? Is there a reason why these capacitors should be of larger value?
Last edited:
Gee, I would use 40-80uFd C1 C2, 10uFd at C3.
Hundreds-uFd caps are a newer thing, with falling prices of "big" caps.
I think the original design was generous, and rounded-up all caps to 220uFd to get the quantity discount. Sure think 100u will be fine at C3.
Hundreds-uFd caps are a newer thing, with falling prices of "big" caps.
I think the original design was generous, and rounded-up all caps to 220uFd to get the quantity discount. Sure think 100u will be fine at C3.
The often recommended smoothing capacitance is 2000uF (2mF) per ampere of current draw.
Where ripple voltage is not a concern this can be reduced to 1mF/A
Where ripple voltage is more of a concern, for instance in amplifiers with a low PSRR, one would go up to 5mF/A
For super duper go as far as 10mF/A
100uF for 300mA is equivalent to .33mF/A
330uF for 300mA is equivalent to 1mF/A
3off 330uF for 300mA is equivalent to 3mF/A
Maybe the guidance stated earlier does not apply to multi-stage RCRC and maybe it does not apply to high voltages since the relative ripple to B+ ratio is quite different.
Where ripple voltage is not a concern this can be reduced to 1mF/A
Where ripple voltage is more of a concern, for instance in amplifiers with a low PSRR, one would go up to 5mF/A
For super duper go as far as 10mF/A
100uF for 300mA is equivalent to .33mF/A
330uF for 300mA is equivalent to 1mF/A
3off 330uF for 300mA is equivalent to 3mF/A
Maybe the guidance stated earlier does not apply to multi-stage RCRC and maybe it does not apply to high voltages since the relative ripple to B+ ratio is quite different.
The final cap essentially has two roles:
1. ripple attenuation (in conjunction with whatever else comes before it in the PSU)
2. provide low impedance supply for the circuit, especially the output stage - to some extent independent of the rest of the PSU
You have already calculated the effect in terms of role 1.
For role 2 you need to ensure that at the lowest audio frequency you are interested in the cap provides an impedance which is small compared to the speaker load reflected at the output anode. My guess is that 100uF is adequate, given that 1950s amps would have managed with rather less.
1. ripple attenuation (in conjunction with whatever else comes before it in the PSU)
2. provide low impedance supply for the circuit, especially the output stage - to some extent independent of the rest of the PSU
You have already calculated the effect in terms of role 1.
For role 2 you need to ensure that at the lowest audio frequency you are interested in the cap provides an impedance which is small compared to the speaker load reflected at the output anode. My guess is that 100uF is adequate, given that 1950s amps would have managed with rather less.
Can you point me to an example of such calculation? Speakers are 4ohm, but I assume the output transformer plays a role as well.
Yes, the OPT transforms the speaker impedance. I can't point you to a ready-made calculation because it is not the sort of thing I need to look up. You need to find what load impedance is presented at the anode (or find an approximate value from a valve datasheet) and calculate the cap impedance using the usual formula.
Thanks for the tip. I looked up the typical load impedance and it seems to be around 4kOhm which results in -3dB point at 0.4Hz with a 100uF capacitor. Should be allright then!
> I assume the output transformer plays a role
If I read that small schematic correctly, the "100uFd" is for the *driver*, not the output stage.
Since I do not know what the driver is, speculation is empty speech (typing).
But assuming it is a tube or two with some large plate resistors, the speaker or OT has almost no impact, you want to look at the driver and its resistances. And as a quick approximation, assume the whole driver is "like" one plate resistor. If it is 50K, then 100uFd is "enormous" at any audio frequency.
You can still get in trouble, motorboating due to subsonic sneakage back from the hard-slammed power stage through the R-C to the driver. But without a clue I won't try to predict if that may be a problem.
If I read that small schematic correctly, the "100uFd" is for the *driver*, not the output stage.
Since I do not know what the driver is, speculation is empty speech (typing).
But assuming it is a tube or two with some large plate resistors, the speaker or OT has almost no impact, you want to look at the driver and its resistances. And as a quick approximation, assume the whole driver is "like" one plate resistor. If it is 50K, then 100uFd is "enormous" at any audio frequency.
You can still get in trouble, motorboating due to subsonic sneakage back from the hard-slammed power stage through the R-C to the driver. But without a clue I won't try to predict if that may be a problem.
That circuit seems to take the driver supply from before the output supply, not after as is more usual. Still I suppose one must expect unusual circuits in PSE, as PSE itself is suboptimal.
Driver circuit has a third RC filter on the main board, which is why this is not visible in the PSU schematic.
- Status
- Not open for further replies.