This is exhausting and silly. Maybe? Or maybe not? Valuable discussion.
Manifesto SE ... a single ended tube amplifier project
happybear
New Member
Epcos DC Link polypropylene film vs. Sanyo Oscon SP-series solid aluminum electrolytic.Aren't you doing a non-polar versus electrolytic bypass capacitor test?
If not, it's my mistake. Sorry.
Any subjective review should be taken with a grain of salt.
The only way to know if something works for your application is to try it out.
HB
happybear
New Member
HB has scope and PC FFT for bandwidth measurements.
Even if -3dB roll off can be determined, the issue with measurements in the electrical domain is the signal source and resolution.
The FFT sweeps 1:1 or 1:3 octave resolution with pink noise and the scope uses a sine wave. It is a basic test but does not tell us everything.
That is not what we are listening to with typical music tracks. Most of the music we hear is pulsed transient energy - the cathode bypass cap can emphasize certain frequencies. Depending on the transducers and the listener's hearing ability, the Rk bypass can require multiple capacitors of various uF values to get a perceived linear response. This is where the "art" of Rk bypassing comes into play.
For the average hobby-level tube amp builder, the multi-cap Rk bypass would not be worth the time or expense to do it correctly by ear.
HB
Last edited:
That's only part of the story. A capacitor with high internal resistance can also produce a flat response. If that resistance is characterized by hysteresis or other non-linearities, signal quality will be compromised. The trick is to ensure there's an absolute minimum of signal AC at the positive end of the cap throughout the audible spectrum. That means it's bypassing the cathode effectively.
Jack
happybear
New Member
Manifesto SE bandwidth sweep in ARTA.
HB added a multi-capacitor bank (.01uf + .033uF + .068uf + .68uF) in parallel with the main 50uF (measured 47.3uf) cathode bypass cap.
There was no measured benefit to adding the multi-capacitor bank of smaller uF values if the Rk is fully bypassed, which it seems to be.
The multi-capacitor bank was then removed, leaving a single Rk bypass capacitor, the Epcos DC Link.
There was some high frequency attenuation, probably due to the lack of FBK in triode mode or the OPT's themselves being budget units.
However, the upper bandwidth was extended beyond HB hearing limits, so the result is acceptable.
Note: it has been suggested not to use more than a 15uF cap as the main Rk bypass capacitor when adding a multi-capacitor bank to the main Rk bypass capacitor.
This suggestion was for DHT 2A3 finals, not sure if that "rule" applies to the 6V6 family of tubes.
HB
HB added a multi-capacitor bank (.01uf + .033uF + .068uf + .68uF) in parallel with the main 50uF (measured 47.3uf) cathode bypass cap.
There was no measured benefit to adding the multi-capacitor bank of smaller uF values if the Rk is fully bypassed, which it seems to be.
The multi-capacitor bank was then removed, leaving a single Rk bypass capacitor, the Epcos DC Link.
There was some high frequency attenuation, probably due to the lack of FBK in triode mode or the OPT's themselves being budget units.
However, the upper bandwidth was extended beyond HB hearing limits, so the result is acceptable.
Note: it has been suggested not to use more than a 15uF cap as the main Rk bypass capacitor when adding a multi-capacitor bank to the main Rk bypass capacitor.
This suggestion was for DHT 2A3 finals, not sure if that "rule" applies to the 6V6 family of tubes.
HB
Last edited:
Any signal voltage not bypassed by the cathode capacitor acts as a signal between grid and cathode. If it's distorted by hysteresis or other non-linearities characteristic to the cap, those anomalies will be transferred to the signal path. I don't know what you mean by "1% residual K signal," but I would always want the residual to be as small as is practicable for this reason.
Jack
OK, I see what you mean. I'm not sure how well that relates to bypassing as it might influence the G/K differential. For instance, if the grid is driven by 1V, what voltage will appear across the cathode RC using this theoretical cap? That's what is most important, and I suspect it really needs to be measured to be properly considered.
Jack
And, it also helps if the cap is biased correctly. For a polar electrolytic, you usually want to have at least half the rated DC voltage on it, at idle. That makes sure that the dialectric never gets close to the zero transition- which is where hysteresis is the worst.
This is a reason to not use extreme high voltage electrolytic caps in places that don't need them. Always try to keep the electrolytic cap to where it is rated at no more than twice the DC idle voltage of the circuit, if possible.
However, this shouldn't usually apply to film caps. As long as the film cap is rated at greater than the voltage, you should be OK.
Regards,
Gordon.
Last edited:
There are two issues with all this in my mind. First, people often hear what they want to hear. I don't recall anyone claiming to have heard this effect who had performed valid A-B testing and/or made even minimal measurements such as distortion and frequency response. Second, to ensure this isn't merely the result of more effective bypassing, one would need to compare film VS electrolytic when the amplifier is already bypassed with a value selected on the basis of low frequency extension. I suspect that in cases where some effect can be verified, simply adding another electrolytic might provide the same result. I'll add that I've tested a number of electrolytics for response, and many are flat into the MHz region. On that basis, plus the inability to demonstrate this effect in my own gear by ear or with measurements, I'm pretty sure it's not an issue of frequency response.
Jack
Last edited:
happybear
New Member
After listening to the Epcos 50uf 630V 10% as a cathode bypass capacitor for SE 7C5 finals for several evenings afterwork, HB deems it to be acceptable.It's all cool!
No, I understand about YMMV. Just curious in your opinion.
The higher voltage rating would make the Epcos more useful than the Sanyo OsCon SP-series electrolytic capacitors in directly coupled SE circuits.
Subjectively, the Epcos presents slightly better low-level detail than the OsCon SP-series but there is not a huge difference in overall subjective performance.
Keep in mind that these are "economy-grade" capacitors in a single Rk bypass capacitor application.
By no means does this represent "state of the art" Rk bypassing concepts.
HB
happybear
New Member
Manifesto SE: finals cathode bias adjustment switch added.
Rk consists of 250 ohm + 40 ohm resistors in series for 290 ohms.
The switch shorts out the 40 ohm resistor setting Rk to 250 ohms.
This will run the SE 7C5 triode mode finals a bit hotter at 13.5 watts on the anode.
If a return to tetrode mode finals is preferable, the switch is opened setting Rk back to 290 ohms.
Kimber TCSS clear jacket is used to wire up the switch.
HB
Rk consists of 250 ohm + 40 ohm resistors in series for 290 ohms.
The switch shorts out the 40 ohm resistor setting Rk to 250 ohms.
This will run the SE 7C5 triode mode finals a bit hotter at 13.5 watts on the anode.
If a return to tetrode mode finals is preferable, the switch is opened setting Rk back to 290 ohms.
Kimber TCSS clear jacket is used to wire up the switch.
HB
Last edited: