DCAudioDIY.com

Interesting discussion.

As a follow up to my earlier post, my recommendation for a large bypass cap extends beyond LF extension. By bypassing the cathode resistor, you are actually creating a fixed-bias power supply that happens to offer the advantage of dc stabilization. It is my experience that at least half the sound quality of the amplifier is reflected by the quality of the power supply, and you want stiff, high-energy-storage power sources to feed your audio stages. In my experience, voltage regulation is a very distant second here, even though I use it but mainly for stability because I use so much dc coupling.

If the cathode of the output stage is bypassed with an undersized capacitor, the impedance at the cathode begins to rise as the frequency of the audio drops. This begins to give ac degeneration (negative feedback) and the impedance at the plate increases. This is reflected through the output transformer and causes an increase in amplifier output impedance for the speaker. Thus speaker damping is reduced, perhaps where you need it most. Now sonically, a rise in output impedance may be a good thing with some speakers as you may get a nice bass bump at the speaker resonance. So a listening comparison done with various bypass capacitors in an amp with a given speaker may indicate an advantage with a certain capacitor of limited value, but you should be aware of what may be going on. The danger of restricting a certain performance parameter in an amplifier is that you start to blame attributes of associated equipment for things, good or bad, that have nothing to do with that equipment. Ultimately you may build up an audio system that is full of peculiarities that become unpredictable as you try to make changes.

Interesting. I am in the position of being able to easily add another 80 to 100uF of PP film capacitance to the cathode bypass capacitor (currently a 50uF composite film). That should get my cutoff point at 1.44 Hz. The fact of the matter is that with interstage tranformer coupling, I will not have much signal that low anyway, but at least it will reduce the phase shift at 20 Hz. My speakers extend down to 40 Hz resonance, but feed my subwoofer plate amps off the amp output terminals.

David

_________________
Walt

Just got in a pair of copper plate GM-70s, so will try to give these a try over the weekend. I do have to say that the new amplifiers (with the graphite plates) have really exceeded my expectations. The amps are so much better sounding than my SV811-10/SV572-10 amp (with the same output transformers), which had a good amount of slam, but not much with respect to refinement. I am inclined to say they are better than my 300B DRD amps, since they seem just as refined and expressive in the midrange, but have the dynamics (at over twice the power). They work better for all types of music, from large orchestral to small jazz ensembles. Very happy!

I am thinking is may be time to start clearing out some of my older amps which are or will be gathering alot of dust.

David

Here is another way of looking at the cathode bypass network. Let's say we have a 100 uF cap and an 800-ohm resistor. The capacitive reactance of the 100 uF is 80 ohms at 20 Hz. So this is about 10 % of the value of the resistor, so certainly you get most of the way to effectively bypassing the resistor at 20 Hz. I don't have a GM70 tube to test, but I made a test on a 6B4G tube whereby I curve traced the tube with no cathode impedance and then with 78 ohms cathode impedance. At the same operating point, the plate resistance went from 888 ohms to 1333 ohms, or about a 50% increase. This increase in plate resistance will, however, cause a reduction in the frequency extension at 20 Hz due to the inductive reactance presented by the primary inductance value of the primary winding combined with the reflected speaker impedance. The inductive reactance hurts and the reflected speaker impedance helps, but not as much. Sounds complicated, but all these things come into play. The 1000 uF cap I suggested gets the capacitive reactance down to 8 ohms at 20 Hz, which is insignificant compared to the 800-ohm resistor and would cause much less of a plate resistance increase compared to grounding the cathode and using a negative bias on the grid. When you oversize the capacitor it is not important to use such a good quality capacitor and an electrolytic is fine. When the capacitor is undersized the quality of the capacitor becomes more of an issue because the voltage is varying across it making dielectric absorption possibly important. This plate resistance test is an easy one for me to do and David, if you would like to to bring the GM70 tube by sometime I would be happy to run the plate impedance test at your operating point. If you wanted to lug the amp over too, we could do a measurement of the primary inductance of your transformer. This would help with a prediction of how the frequency response would be affected. Then a measurement of the output stage of your amp could be made to see if the prediction was good or not. This would be a learning experience for me as well as I have not done this. Perhaps some others interested in this topic could come too.

But don't power tubes like to "see" a high load impedance. It seems to me that most production tube amplifiers with output transformers have a compromise transformer to help deal with different speaker loads -- hence the different taps at 4, 8, and 16 ohms. The nominal impedance of my Altecs is 8-ohms, but the sound really opened up when I started feeding them from the 16-ohm taps. My 845s seem to like a higher plate load.

Paul Elliott's Magnepans have a nominal impedance of 4-ohms and yet when we ran them off of the 8-ohm taps of his them Monsoon amps. they opened up in a similar fashion. Paul's KT-88s liked the higher plate load.

That said -- in my cathode bias world -- when replacing an aluminum electrolytic I hope for enough physical space to replace it with nice film cap(s). (typically polycarbonates because of their extremely small size for a given value) I go for low end at no higher than 5-Hz. If I don't have enough room I down size the electrolytic with a signal path worthy electrolytic and then fill in with film to get the best possible audio quality. I always get the intended result -- better sound reproduction.

Besides the usual suspects like DA and DF, aluminum electrolytics get "hokey" past 10-kHz. that has the effect of rolling off past 10-kHz. That can be a nice "band-aid" for systems that have issues with a hard edge else-where in the signal path, but with a system that "steps out of the way" further up the signal path, my experience has been that the cathode bypass caps have the same effect as mediocre signal path components.

_________________
Walt

yes, if there is a gathering for this, I would like to come and get educated!

do you use a film/foil or metallized film? The metallized film caps can be had in smaller sizes per value than film/foil, although I was told by a studio engineer that film/foil was superior in sound and what he used on his tape machines as power supply bypass...

_________________
Walt