On my old RKV, there are two big output caps. Electrolytic, 220uf, 350V. Fact is, they need to be changed, they made their time.
The question is : are such big caps useful for 300ohms headphones (I don't plan to use anything below) ?
I found this formula on headwize, giving the frequency corner (3db down) in function of the size of the output cap and the impedance of the cans:
C (uF) = 1,000,000 / [2pi (corner frequency) Rheadphones]
So :
47 = 1,000,000 / [2 x 3.1416 x corner frequency x 300]
=> 47 x 2 x 3.1416 x 300 = 1,000,000 / corner frequency
==> corner frequency = 1,000,000 / 88,593 = 11.30Hz
Which is fine.
The good thing now is that a 47uf, 400V polyprop audyn cap is only 10.25€ a piece. And a high quality electrolytic caps, with this voltage rating would cost arms and legs. And I'm not even speaking of 220uf/385V blackgate (around 130$ a piece).
Is everything here allright or is the output capacitor also dependant of the circuit as a whole ?
The question is : are such big caps useful for 300ohms headphones (I don't plan to use anything below) ?
I found this formula on headwize, giving the frequency corner (3db down) in function of the size of the output cap and the impedance of the cans:
C (uF) = 1,000,000 / [2pi (corner frequency) Rheadphones]
So :
47 = 1,000,000 / [2 x 3.1416 x corner frequency x 300]
=> 47 x 2 x 3.1416 x 300 = 1,000,000 / corner frequency
==> corner frequency = 1,000,000 / 88,593 = 11.30Hz
Which is fine.
The good thing now is that a 47uf, 400V polyprop audyn cap is only 10.25€ a piece. And a high quality electrolytic caps, with this voltage rating would cost arms and legs. And I'm not even speaking of 220uf/385V blackgate (around 130$ a piece).
Is everything here allright or is the output capacitor also dependant of the circuit as a whole ?
hum, the design of the RKV is awfully complicated and i can't associate it with any of those schematics.
Is there a practical way to measure the Rout, i mean with a multimeter or such ?
http://www.audiovalve.de/rkv/rkvpatent.jpg here's a bad pic of the schematic; if you feel you could help, i can provide you the actual schematic but i've been asked not to make it public.
Is there a practical way to measure the Rout, i mean with a multimeter or such ?
http://www.audiovalve.de/rkv/rkvpatent.jpg here's a bad pic of the schematic; if you feel you could help, i can provide you the actual schematic but i've been asked not to make it public.
C (uF) = 1,000,000 / [2pi (corner frequency) (Rheadphones-Rout)]
Uh, no. There is a typo. That last term should be "Rheadphones + Rout"
Try it for the once-common case of a 600Ω source driving a 600Ω load. The way you have it, you would need an infinite capacitor. I don't want to think about the uncommon but possible case where Zout is larger than Zload.
Add the source and load impedances.
> I found this formula... giving the frequency ... the size of the output cap and the impedance....
I hate math.
Just remember: 100µFd in 100Ω gives 20Hz (actually 17Hz, or 16.6Hz).
For other resistances, scale as needed. To reach 20Hz with a 10Ω load you need 1,000µFd.
1000µFd -- 10Ω
100µFd -- 100Ω
10µFd -- 1000Ω
1µFd -- 10KΩ
0.1µFd -- 100KΩ
0.01µFd -- 1MegΩ
If you want a different frequency, scale up or down as needed (bigger cap for more bass).
If there is only one cap in your whole system, a -3dB point of 20Hz will be -1dB at 40Hz. In a real system you tend to have several caps. Four stages each -3dB at 20Hz will be -12dB at 20Hz, -4dB at 40Hz, -2dB at 80Hz. Hi-fi systems generally have their worst cap good for 10Hz, and others around 0.5 to 5Hz. The output cap for a tranformerless tube amp is usually the biggest in the system and thus the "worst" in terms of picking a bass response and price you enjoy, so here 11Hz "is fine" as 00940 says. Other caps in the system should be aimed for lower Hz, especially where it does not impact cost too much.
If you use electrolytics and worry about their distortion, oversize three or more times higher than you "need" to meet your bass requirement, and shunt with a non-electro cap 10 to 100 times smaller.
The schemaic link posted by 00940 is as bad as claimed, and I can't read the tube topology, but it does not matter. It clearly shows voltage feedback from the output node to an opamp. The output cap is outside the feedback loop, no complication. The output impedance is surely "very low". It could be 1Ω or 30Ω but that makes very little difference around a 300Ω load.
From "1000µFd -- 10Ω 100µFd -- 100Ω", we see that 300Ω wants 30µFd for 20Hz response. We want something better, but 30µFd is already a frighteningly huge value (costly) for a film cap. 00940 says 47µFd is affordable. Using head-math, the -3dB point is 13Hz, good. (11.3Hz is more accurate, though 3-digit precision is not necessary in non-filter audio coupling caps, especially around a headphone which may be +/-20% or more off its nominal impedance.)
If 47µFd in 300Ω is 11.3Hz, and the amp output is really 30Ω, the -3dB point shifts to 10.3Hz. If the output were 600Ω, it shifts to 4Hz. So assuming that the output Z is zero is conservative; in practice the response will be a little better.
Uh, no. There is a typo. That last term should be "Rheadphones + Rout"
Try it for the once-common case of a 600Ω source driving a 600Ω load. The way you have it, you would need an infinite capacitor. I don't want to think about the uncommon but possible case where Zout is larger than Zload.
Add the source and load impedances.
> I found this formula... giving the frequency ... the size of the output cap and the impedance....
I hate math.
Just remember: 100µFd in 100Ω gives 20Hz (actually 17Hz, or 16.6Hz).
For other resistances, scale as needed. To reach 20Hz with a 10Ω load you need 1,000µFd.
1000µFd -- 10Ω
100µFd -- 100Ω
10µFd -- 1000Ω
1µFd -- 10KΩ
0.1µFd -- 100KΩ
0.01µFd -- 1MegΩ
If you want a different frequency, scale up or down as needed (bigger cap for more bass).
If there is only one cap in your whole system, a -3dB point of 20Hz will be -1dB at 40Hz. In a real system you tend to have several caps. Four stages each -3dB at 20Hz will be -12dB at 20Hz, -4dB at 40Hz, -2dB at 80Hz. Hi-fi systems generally have their worst cap good for 10Hz, and others around 0.5 to 5Hz. The output cap for a tranformerless tube amp is usually the biggest in the system and thus the "worst" in terms of picking a bass response and price you enjoy, so here 11Hz "is fine" as 00940 says. Other caps in the system should be aimed for lower Hz, especially where it does not impact cost too much.
If you use electrolytics and worry about their distortion, oversize three or more times higher than you "need" to meet your bass requirement, and shunt with a non-electro cap 10 to 100 times smaller.
The schemaic link posted by 00940 is as bad as claimed, and I can't read the tube topology, but it does not matter. It clearly shows voltage feedback from the output node to an opamp. The output cap is outside the feedback loop, no complication. The output impedance is surely "very low". It could be 1Ω or 30Ω but that makes very little difference around a 300Ω load.
From "1000µFd -- 10Ω 100µFd -- 100Ω", we see that 300Ω wants 30µFd for 20Hz response. We want something better, but 30µFd is already a frighteningly huge value (costly) for a film cap. 00940 says 47µFd is affordable. Using head-math, the -3dB point is 13Hz, good. (11.3Hz is more accurate, though 3-digit precision is not necessary in non-filter audio coupling caps, especially around a headphone which may be +/-20% or more off its nominal impedance.)
If 47µFd in 300Ω is 11.3Hz, and the amp output is really 30Ω, the -3dB point shifts to 10.3Hz. If the output were 600Ω, it shifts to 4Hz. So assuming that the output Z is zero is conservative; in practice the response will be a little better.
Thanks a lot for this help.
I got other answers here too btw : http://www4.head-fi.org/forums/showthread.php?s=&threadid=49271
Now I've three ideas in mind, but cost would prohibit to try all of them :
1 - conservative solution : put 470uf of "cheap" electrolytics (panasonic ?) and bypass them by a good 4.7uf polyprop.
2 - "let's go for it" solution : a 82uf polyprop (19€ a piece) and basta.
3 - "in between more costly" solution : a 150uf/350v blackgate (27€ a piece).
I'm a bit lost
I got other answers here too btw : http://www4.head-fi.org/forums/showthread.php?s=&threadid=49271
Now I've three ideas in mind, but cost would prohibit to try all of them :
1 - conservative solution : put 470uf of "cheap" electrolytics (panasonic ?) and bypass them by a good 4.7uf polyprop.
2 - "let's go for it" solution : a 82uf polyprop (19€ a piece) and basta.
3 - "in between more costly" solution : a 150uf/350v blackgate (27€ a piece).
I'm a bit lost
wow I'm tired, but since the banks are closed tomorrow, i can't go rob one early anyway. I'll sleep late. 
I've been looking on blackgate.jp at their current line.
There are not a lot of interesting beasts for me. In fact, I'll have to stick with solution n°3 of the earlier post : 150uf/350V from the VK serie ( "almost non-polarized caps"). There are about no 250v BG. Other interesting one would still be in the vk line, it's a 220uf but only rated at 200v.
I've been looking on blackgate.jp at their current line.
There are not a lot of interesting beasts for me. In fact, I'll have to stick with solution n°3 of the earlier post : 150uf/350V from the VK serie ( "almost non-polarized caps"). There are about no 250v BG. Other interesting one would still be in the vk line, it's a 220uf but only rated at 200v.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.