cathode by-pass in push pull six tube hifi amp

[bud]

This is for an advanced tube tech. Someone who really understands the how and why of cathode by-pass caps on push pull output tubes fuctioning in class AB, with common adjustable bias. Feedback is adjustable from zero to about 10.

This is the schematic of a Mesa Baron dual mono amp using 12 6l6GC. Wonderful amp generally speaking: In All triode, it is absolutely amazing; in 2/3 triode & 1/3 pentode still amazing; in 1/3 triode and 2/3 pentode, is good; and in all pentode it will get on your nerves unles you play heavy metal. I have made many modifications which have helped substantially, along with old stock tubes, and it has been modified to run any tube in the 6L6/KT family as well as el34's.

That said, even with the mods, the all pentode or mostly pentode state is a bit edgey, strident, as if there is third order harmonics dominating (like a common transistor amp).
The cathode by-pass caps are about the last thing to replace/up grade (Sprague...who says they don't leak). The cap is general purpose, 10K uf @16 volt electrolytic. I am wondering if these caps are the reason for the pentode harsh edge. A tech at Parts Connection says 'take them out'. The tech at Mesa says 'leave them in'. I can't connect to the designer to ask advice...just techs who may or may not know the real answer I am seeking.

So far, it appears that the caps do, in fact, need to be there for many researched reasons. The first question is, what kind of caps should they be replaced with...low impedence, low ESR, low leakage????? Since signal DOES go thru these caps, the quality should matter, but the characterists....this is the question (my guess is low impedence). And...the second question is, since the cap value is so high would several smaller values in parallel be better than one at stated value?




As a last thought/question, many add a 10 ohm to the outputs of 7591 tubes in vintage amps. Should these resistors be by-passed since there is no longer the cancellation benefits of a common ground/resistor?

thanks everyone !

[Chuck H]

I wouldn't trouble about the cathode bypass caps on the powe tubes. The bias arrangement looks to be a combination of negative grid voltage and positive cathode voltage. With the greater factor being the negative grid voltage. So the actual affect of those bypass caps is pretty small. Certainly small enough that you won't hear a difference between different brand capacitors. But on that note... If the amp is sounding too shrill in pentode the nod would be toward a cap with a HIGHER impedance. Let me qualify that. The HF performance of electrolytics is poor compared to other cap construction. And that performance decreases when frequency increases. So, the cap with the least performance in this circuit will be the one that bypasses the least HF. Since there is a gain increase on the bypassed frequencies, and the amp sounds too shrill, poorer performance by these caps should mean better sound quality for the whole amplifier. That is, as the amps performance relates to those capacitors. Now...

If you want to try removing them, go ahead. it's safe and it may well soften the tone a tad. But as I noted, the difference will probably be too small to matter.

One reason triode mode is used in many hi fi amps is that people like the way it sounds and performs. The best way to get triode sound and performance at higher power is to get a bigger triode amplifier. I'm not surprised that pentode mode sounds harsher than triode mode. A few things are in play. One biggie being the OT impedance. Which should be idealized for triode or pentode operation. Only then can the feedback loop be adjusted to provide maximum performance at the lowest distortion. So keep in mind that all the features of the amp can't give maximum performance without changing other things that aren't offered for adjustment. Just set the amp up how it sounds best to you and enjoy it.

As to the 10R resistors used on 7591 tubes. I don't know about it. I would say it's either the same idea as pulling the bypass caps on your 6L6's, or it's just a means of providing a convenient node from which to measure bias.

If you really MUST tweak something on the amp to be satisfied, change the V2 cathode bypass cap to 47uf. The 15uf value over a 470R resistor isn't really a full bypass. And it should be. Again, it's not going to make much, if any audible difference because the actual gain increase for the bypassed frequencies in that circuit is only about 2dB.

Sorry I didn't have more positive advice. But it's actually a pretty simple circuit. Nothing wrong and no point in changing much about it. If you find a sweet spot in the operation at some setting between triode and pentode that is probably because the amp was optimized for that setting.

[bud]

Thanks Chuck for taking the time to give a comprehensive answer. Had the design been just triode or pentode, I would have seen the answer but this blend had my brain seeing complications that maybe are not there. Yes, enjoy the amp in the setting I like best is the real answer, triode or 1/3 pentode for bite and no feedback. The amp can pull out the best in a recording by making the changes. You are right, it can't be a slave to so many masters, so full pentode is the looser...but I am past hard music.

My personal confusion comes from reading to many 'opinions'. I read the 1955 article in Audio about the cap on PP in class A, and the last line left me in the corn field..." AB push/pull should always have the cap if they are to be modern hifi design" or words to that effect. I see most triode power tubes run the cap. So all these 'inputs' plus tech advice, creates great confusion. I am trying not to let my 'ear' be the judge at the cost of possible amp damage in the flash of the eye (say the pair of resistors failing). I assumed the cap was there more for the triode setting than the pentode setting, your comments indicate maybe not. Since the Mesa tech was pretty insistant that it needed to be there, I will try your higher impedance cap idea ( cheaper too). Returning to that article idea, the experiments indicated that both the tube selected and the the type of distortion were both affected by the cap in or out, but IM and THD were lower when the cap was there. This is where I got the idea that the 10Kuf caps on my amp may be a part of the problem....I think I am hearing higher order harmonic distortion in full pentode.

I figured, as you pointed out, the cap keeps the cathode more positive or stablizes it during demanding power requirements. The Mesa tech was none to happy with the Sprague caps in that spot, seems the all fail and leak. This was one reason why I wondered if that cap could have added to the harsh pentode mode. The harsh edge apperars to be in the upper mid range, perhaps 6khz to 12khz area. Your suggestion to pick a cap with a higher impedence for less effect seems to make sense.

On the 7591 question, when the tubes' cathodes no longer share a common resistor to ground or were directly connected to ground, and then are independently isolated by that 10 ohm resistor....is there some IM or THD increase or phase cancellation that is being lost? Something is try to connect in my brain that the methode is missing something.

again,many thank for your help
bud

[Steve Conner]

A pentode output stage is supposed to generate lots of odd harmonics and sound edgy and strident. What style of music are you playing?

Edit: Oh god, I just realised the Baron is sold as a hi-fi amp, not a guitar amp. My advice is to forget about pentode mode, use triode mode, ideally with KT88s. What you're hearing is the IMD and poor HF damping factor of the tubes themselves, when the extra negative feedback of triode mode is taken away. Not the shortcomings of non-audiophile cathode bypass caps.

[bud]

thanks...yes, the tube choice does have an effect. Russian ( new Sensor) and Chinese the worst, JJ el34 are not so bad, USA the better. Type of tube also has an effect...5881, 6L6gc, el34, ect. Yep, pentode is only good for metal music which I have grown out of. Now it is jazz, lounge, symphonies, medium rock...about everything but 'tune-um-up-jazz 'what the hell are they playing', and metal. Have not tried KT66 0r 88 yet.

[Chuck H]

On the electrolytic cap type, Sprague makes good caps. Or at least they use to. I've only had minor issues with them. Still enough that I don't use them anymore. But there was a lot of copies on the market at one time. Blatant copies being sold as ATOM caps. They are reported to be pretty awful. These caps even made their way into some small manufacturers products from what I've read. This could be why rhe performance has been poor for thise caps in those amps. They may also have been ld caps when they were installed and already failing. Just pull them and see if you like the change. If not, replace them with something better.

[bud]

On the electrolytic cap type, Sprague makes good caps. Or at least they use to. I've only had minor issues with them. Still enough that I don't use them anymore. But there was a lot of copies on the market at one time. Blatant copies being sold as ATOM caps. They are reported to be pretty awful. These caps even made their way into some small manufacturers products from what I've read. This could be why rhe performance has been poor for thise caps in those amps. They may also have been ld caps when they were installed and already failing. Just pull them and see if you like the change. If not, replace them with something better.

The amp does have ATOM in it in the power supply, along with Illinois low current leakage type, and CDE hgih ripple for primary B+...not junk for sure. The ones that leaked are general purpose, and yes, they could have been counterfiet but the amp is 1997 vintage...not sure how much of that was going around then (but is rampent on eBay now from overseas sellers).

I likely will go with Ill TTAA type, Rubycon, or Nichicon.

Would you go with smaller values in parallel or one? Any benefit?

[Chuck H]

Would you go with smaller values in parallel or one? Any benefit?

That is actually a question that "I've" asked here!!! The answer I got was that smaller value caps of the same manufacture and type generally have proportionately higher impedance. Ergo, two 100uf caps in parallel has about the same impedance as a single 200uf cap (or 220uf as a standard value). So in this case it's a wash. Don't trouble with the parallel caps unless you can't get a single cap in a high enough value.