I am working on a two-stage DC-coupled PSE amp and am having trouble choosing a cathode bypass cap for the output stage. The cathode resistor is 2200-ohms. Calculating using this value and a cutoff frequency of about 4Hz gives an absurdly small value of 18uF. I get similarly absurd small values calculating with anode resistance (~900-ohm) and transformer impedance (3K:8, 20H) taken into account.
I have modeled the circuit in LTSpice using 220uF and 470uF on the cathode. Typically, the AC analysis reports voltages in the mV range on the cathode across the audio band, yet I see 2.1V at 20Hz, 0.7V at 40Hz, 0.3V at 100Hz, etc. Does this mean there will be substantial crosstalk between channels at these frequencies?
Also, what kind of capacitor should I use here? The cathode sits at +140V and the amp reaches full power with 30V on the grid. So, I suppose I need at least 170V caps. This limits the playing field. Normally, I choose Elna Cerafine or Silmic II. Are there any caps offering similar sonic attributes at higher voltages? Or, should I consider running two higher quality caps in series?
Thank you!
I have modeled the circuit in LTSpice using 220uF and 470uF on the cathode. Typically, the AC analysis reports voltages in the mV range on the cathode across the audio band, yet I see 2.1V at 20Hz, 0.7V at 40Hz, 0.3V at 100Hz, etc. Does this mean there will be substantial crosstalk between channels at these frequencies?
Also, what kind of capacitor should I use here? The cathode sits at +140V and the amp reaches full power with 30V on the grid. So, I suppose I need at least 170V caps. This limits the playing field. Normally, I choose Elna Cerafine or Silmic II. Are there any caps offering similar sonic attributes at higher voltages? Or, should I consider running two higher quality caps in series?
Thank you!
The resistance value to use is the parallel combination of the cathode resistor and the cathode impedance. The cathode impedance will be (from memory) (1/gm) + (RL/mu), where gm is transconductance and RL is the resistance seen by the anode.needtubes said:
I think the rp term is just an alternative way of writing 1/gm. I have a vague memory that the actual formula is rp/mu + RL/(mu+1) - but I could be wrong.
As far as I remember it is (Ri+Ra)/(µ+1) in parallel with the cathode resistor.
(1/gm)+(Ra/µ) is almost the same.Since gm= µ/Ri , you get (Ri+Ra)/µ without the +1.
Mona
(1/gm)+(Ra/µ) is almost the same.Since gm= µ/Ri , you get (Ri+Ra)/µ without the +1.
Mona
Got it. Looks like I need 500uF or so. Any downsides to using two high quality 1000uF, 100V in series?
Also, any word on the potential crosstalk between channels?
Thanks!
Also, any word on the potential crosstalk between channels?
Thanks!
Last edited:
Two in series would work, but perhaps with increased ESR. Voltage balancing resistors might help.
I am unclear why you ask about crosstalk - surely the two channels will not be sharing a cathode resistor and bypass?
I am unclear why you ask about crosstalk - surely the two channels will not be sharing a cathode resistor and bypass?
No, they will not share cathode resistors/caps. Guess I am misinterpreting the LTSpice reports...
Having a hard time finding suitable 'audio' capacitors in the 500uF-1000uF range at 100V and over.
Maybe a fixed bias scheme is a better alternative?
Maybe a fixed bias scheme is a better alternative?
You seem to have swapped a minor problem (finding a good high voltage non-polar coupling cap) for a bigger problem (finding a good "audio" electrolytic cathode decoupler). People who insist on DC coupling where it is inappropriate often hit this dilemma. And we won't mention the bias shift created by output valve second-order distortion and the distinction between quiescent current and average DC current, shall we?
Indeed... And, I am starting to lose the nerve to continue sourcing parts for this build.
Looks like 1000uF, 100V Nichicon KX caps are available. These are considered by Nichicon to be 'audio grade' and may be acceptable.
Looks like 1000uF, 100V Nichicon KX caps are available. These are considered by Nichicon to be 'audio grade' and may be acceptable.
So, which is the lesser of the two evils? Place a cap directly in the signal path or put a large somewhat audio quality cap on the cathode?
There's no such thing as an "audio" cap; it's a myth. "Audio" caps are no better or worse at anything in particular than standard caps made from the same materials (actually, standard caps benefit from the homogenising effect of mass production). If you really want an "audio" electrolytic, find a standard non-polar electrolytic. These have distortion an order of magnitude better than equivalent polars.
If you want to pay a little extra, have a printing company make you a really cool label to stick on the capacitor. (You could even add a little hologram of authenticity or something, like you see on baseball cards etc...) This will make the capacitor exactly as good as an "audio" cap, because that it what the makers of "audio" caps do.
Last edited:
Both are equally in the signal path. As film caps and other non-polar caps tends to be nearer to an ideal capacitance at audio frequencies than any electrolytic I would always use the coupling cap.needtubes said:
The cathode decoupler used in a conventional circuit is less in the signal path than that in a DC-coupled circuit, because the cathode resistor is much smaller. You can use a low voltage high value electrolytic, which is likely to have smaller ESR and inductance than the high voltage electrolytic needed for the DC-coupled circuit. So the conventional circuit has better components in less critical circumstances than the DC-coupled version. Given that, why opt for DC coupling? The quick answer is that it superficially looks better; in fact it is worse.
audio grade vs audiophile capacitors
Industry leading companies like Panasonic and Elna have audio grade capacitors. I am not informed what would be their specific quality but I find it hard to believe these brands risk their reputation. Maybe we must differentiate between dubious audiophile caps and better than 'run of the mill' like we do with 'generic' and 'low ESR' type.
The 'audio type' might come down to noting special indeed but just lacking offending values w.r.t. inductance, leakage, resistance, resonance and other measurable properties. It could be a hoax but the story goes Philips engineers in the 1970s craved for Japanese caps in their amplifiers but were not allowed to employ because of the cost.
High End Audio - Electrolytic capacitors
Industry leading companies like Panasonic and Elna have audio grade capacitors. I am not informed what would be their specific quality but I find it hard to believe these brands risk their reputation. Maybe we must differentiate between dubious audiophile caps and better than 'run of the mill' like we do with 'generic' and 'low ESR' type.
The 'audio type' might come down to noting special indeed but just lacking offending values w.r.t. inductance, leakage, resistance, resonance and other measurable properties. It could be a hoax but the story goes Philips engineers in the 1970s craved for Japanese caps in their amplifiers but were not allowed to employ because of the cost.
High End Audio - Electrolytic capacitors
You can look for JJ's 550 uF/ 385V electrolytic cap too. Axial electrolytic caps might be a slight improvement otherwise I really can't hear any difference between "audio grade" and normal good quality electrolytic. If you want better caps you have to forget electrolytic capacitors and in your situation it becomes expensive and bulky, I guess.
- Status
- Not open for further replies.