Will do my best. At my age my rise-time is totally insufficient for any class-D operation anyway.
(Now-now-now folks! That is not what I meant!)
(Now-now-now folks! That is not what I meant!)
Re: Re: looking for pp amp schematic without nfb
It would seem that English is not robertc's first language either. No problem there. Allow me to rephrase it and add a tiny addition of my own:
"Is it true that an amp without NFB can sound more natural than one with NFB?"
It might be seen as a different question and it is one that is not (in my opinion) answered well by the textbooks. Just to be clear, I am not attempting to change the laws of physics. In fact I have a lot of formal training in both physics and electronics. I have designed servo motion control systems as a profession, so I am not afraid of feedback - negative or positive.
I have been designing and building audio systems as a hobby for over twenty years. The first amplifiers that I built were all good textbook designs. Obviously they were PP with carefully applied NFB because nothing else makes sense in a good textbook design. After a while I decided to pay some attention to those with more emotion than data [see this post ] I built a PP 6CK4 amp (without NFB) that, compared to the good textbook designs, sounded more than a little flabby and "colored." However, I did notice that there was something very, very good about the way it sounded. "Natural" might be a good word to describe it ... "real" might be better. It made the the good textbook design that I had been listening to sound alarmingly flat and two dimensional. (The old amp was PP EL34 very similar to a classic Mullard design. Honestly, I just didn't want to listen to it any more.)
I tried applying NFB to the 6CK4 amp. The NFB did tighten it up and reduce the coloration, but it also killed the ability of the amp to sound "real." I tried varying amounts if NFB. It was always a compromise. I can easily imagine that some people might prefer the sound with some (or a lot) of NFB, but no matter how little was applied there was always a cost in terms of realness or naturalness.
That is just an example, but it is a good one. At first I didn't want to give up on NFB, but eventually I had no choice. Instead I went searching for designs that could sound CLEAN without NFB. It can be accomplished, but obviously not without cost. Unfortunately, those attractive easy to drive pentodes don't work. UL just doesn't come through on its promises; judging purely by the sound it's worse than pentodes with NFB (maybe just different, but certainly not right.) On the other hand, those antique directly heated triodes showed promise. They can sound clean and uncolored without NFB, and the naturalness that I got a hint of with the PP 6CK4 amp is there in all of its fantastic glory. The are not as simple to use and they generally produce much less power, but if "natural" is your goal ...
Please note that it's not just a matter of fashion for me. I did not start with SE 300B because that happened to be popular with the crowd that I wanted to hang with. In fact, after all of those years and all that I've learned, I recently (over the past year) decided to try to build a good sounding PP EL84 amp. What I've managed to come up with is listenable; certainly much better than any of the plastic boxes you can buy at any big name store. But it falls far short in terms of the realism that I've become accustomed to.
You can call it a matter of taste and you would be correct; everyone has different priorities. As I said earlier in this post, some people might prefer some (or a lot) of NFB for a tighter sound, but in my experience it never makes an amp sound more natural and it is completely unnecessary (my opinion) if everything else is done right.
-- Dave
Johan Potgieter said:
It would seem that English is not robertc's first language either. No problem there. Allow me to rephrase it and add a tiny addition of my own:
"Is it true that an amp without NFB can sound more natural than one with NFB?"
It might be seen as a different question and it is one that is not (in my opinion) answered well by the textbooks. Just to be clear, I am not attempting to change the laws of physics. In fact I have a lot of formal training in both physics and electronics. I have designed servo motion control systems as a profession, so I am not afraid of feedback - negative or positive.
I have been designing and building audio systems as a hobby for over twenty years. The first amplifiers that I built were all good textbook designs. Obviously they were PP with carefully applied NFB because nothing else makes sense in a good textbook design. After a while I decided to pay some attention to those with more emotion than data [see this post ] I built a PP 6CK4 amp (without NFB) that, compared to the good textbook designs, sounded more than a little flabby and "colored." However, I did notice that there was something very, very good about the way it sounded. "Natural" might be a good word to describe it ... "real" might be better. It made the the good textbook design that I had been listening to sound alarmingly flat and two dimensional. (The old amp was PP EL34 very similar to a classic Mullard design. Honestly, I just didn't want to listen to it any more.)
I tried applying NFB to the 6CK4 amp. The NFB did tighten it up and reduce the coloration, but it also killed the ability of the amp to sound "real." I tried varying amounts if NFB. It was always a compromise. I can easily imagine that some people might prefer the sound with some (or a lot) of NFB, but no matter how little was applied there was always a cost in terms of realness or naturalness.
That is just an example, but it is a good one. At first I didn't want to give up on NFB, but eventually I had no choice. Instead I went searching for designs that could sound CLEAN without NFB. It can be accomplished, but obviously not without cost. Unfortunately, those attractive easy to drive pentodes don't work. UL just doesn't come through on its promises; judging purely by the sound it's worse than pentodes with NFB (maybe just different, but certainly not right.) On the other hand, those antique directly heated triodes showed promise. They can sound clean and uncolored without NFB, and the naturalness that I got a hint of with the PP 6CK4 amp is there in all of its fantastic glory. The are not as simple to use and they generally produce much less power, but if "natural" is your goal ...
Please note that it's not just a matter of fashion for me. I did not start with SE 300B because that happened to be popular with the crowd that I wanted to hang with. In fact, after all of those years and all that I've learned, I recently (over the past year) decided to try to build a good sounding PP EL84 amp. What I've managed to come up with is listenable; certainly much better than any of the plastic boxes you can buy at any big name store. But it falls far short in terms of the realism that I've become accustomed to.
You can call it a matter of taste and you would be correct; everyone has different priorities. As I said earlier in this post, some people might prefer some (or a lot) of NFB for a tighter sound, but in my experience it never makes an amp sound more natural and it is completely unnecessary (my opinion) if everything else is done right.
-- Dave
This is absolutely true. So why are you so hung up on objective proof? Test equipment is fine for design. I own lots of high quality Hewlitt-Packard, Tektronix, B&K, Gen-Rad, etc. stuff that I use myself. But it has only helped prove to myself that a good measuring amp isn't necessarily the best sounding one. But for the type of music I listen to, a feedbackless (if there is such a word) always sounds more natural, lifelike, and LESS DISTORTED than one with feedback. Maybe that will answer Robert's original question. I haven't read a post yet in which someone has been bold enough to say flatly that a real triode (845, 211, 300B, 2A3, GM-70) amp with feedback sounds better, based on their own listening experience, than one without.
John
jlsem said:
John;
I am not bold, but 1-triode power amp driven by a CD player sounds for my taste better when it's gain is reduced by a parallel feedback (inverting amp) rather than by a voltage divider.
Now you've heard that. 😀
You guys should all play on a Hammond organ for awhile... be a bit less mystified by harmonics.
Some of the best stuff I've ever heard was Dave Matthew's, on CD, played through Sony junk, with JBL speakers...
If you don't play an instrument, you might be at a disadvantage. After all... when you listen to live music... you are listening to PEAVEY.
😉
Some of the best stuff I've ever heard was Dave Matthew's, on CD, played through Sony junk, with JBL speakers...
If you don't play an instrument, you might be at a disadvantage. After all... when you listen to live music... you are listening to PEAVEY.
😉
poobah said:
If each instrument is connected to the own amp with own speaker it is all the one instrument in each particular case, like tha amp and the speaker is the part of the instrument. If to mix all together and play through the same amp, one of them, it will be a mess. 😀
PS: I like a Hammond organ with Leslie milk-shaking amp distortions over the room... 😉
Great post Dave. Though I've a fraction your time in DIY my experiences parallel yours closely. From Mc240 to horrid little high GNFB EL84SE that still just did something right (another Mullard design), to grinding away at circuits trying to replicate the 'right' and minimize the 'wrong'. Haven't made the DHT jump yet though, still much to learn first about the way I hear things. One question: when you say no NFB, does that include local degeneration? No cathode followers for example?
Re: Re: Re: looking for pp amp schematic without nfb
For whatever reason, NFB does seem to have that effect. It does seem to be flatter and less spatial. If using pentodes, it's necessary to trade off something since they do other things that are even worse sonically. If you can hold the gNFB down to reasonable levels, it's still quite OK. Who knows what that's all about? There are always design trade-offs. Perhaps making the level of gNFB variable might be something to consider?
This guy did.
Dave Cigna said:
For whatever reason, NFB does seem to have that effect. It does seem to be flatter and less spatial. If using pentodes, it's necessary to trade off something since they do other things that are even worse sonically. If you can hold the gNFB down to reasonable levels, it's still quite OK. Who knows what that's all about? There are always design trade-offs. Perhaps making the level of gNFB variable might be something to consider?
This guy did.
An externally hosted image should be here but it was not working when we last tested it.
Re: Re: Re: Re: looking for pp amp schematic without nfb
Indeed I did!
Sometimes NFB is needed to tailor harshness. GOT knows a 300B needs it for my tastes 😉
Miles Prower said:
Indeed I did!
Sometimes NFB is needed to tailor harshness. GOT knows a 300B needs it for my tastes 😉
rdf said:
Well, I have to admit that I often leave cathode resistors unbypassed; it usually sounds better to me. Maybe if I could get REALLY good capacitors I would feel differently, maybe not...
I am not fond of cathode followers and have not found any place where they are a real advantage. The calculated output impedance is very low, but they only work right when driving a high impedance. In other words, if you parallel enough of them to get the output impedance down to 8 ohms, then try to drive a speaker directly they willl fall on their face due to current limiting. They need a load more like 10x or even 100x of their output Z to work as expected. In that case, why bother with them?
For years I have been wanting to build a PP triode amp with a McIntosh style (unity coupled) output transformer. Maybe with 6AS7's. They are not my favorite tube but I think they would work well in that application. Anyway, that configuration makes heavy use of NFB. (Think of it as a pair of PP concertinas. The gain is always less than two.) Anyway, the whole idea flies in the face of everything I've just said, but I still would like to try it if I ever get the time and the inclination. Who knows ...
-- Dave
Hi Dave!
Just a bit on your problem with cathode followers - later to our other topic.
You correctly state that the necessary current capability is also required for the cathode follower to work - like any other amplifier stage. Their low impedance is of value when (1) you drive a power amplifier with admittedly a high input impedance, but whose characteristics are only guaranteed when there is no further undefined driver impedance in series with the input (e.g. Miller effect with a high mu triode first stage as so many amps have); and (2) for driving e.g. tone compensation or other filters with substantially variable input impedance with frequency in order to have predictable characteristics. A cathode follower's relatively high input impedance may also be desirable following such a stage.
Drive a power amp with several meters of interconnect from a triode driver with say 47K as anode load, compared to the same in the cathode (Zout perhaps = 1K2), and you are likely to notice a difference in noise pickup. But the matter you mention of designers sometimes not paying attention to whether their cathode followers can cope with what they attach, is certainly a real one! I have seen some real non-linear cathode followers even despite the feedback in my day.
Regards.
Just a bit on your problem with cathode followers - later to our other topic.
You correctly state that the necessary current capability is also required for the cathode follower to work - like any other amplifier stage. Their low impedance is of value when (1) you drive a power amplifier with admittedly a high input impedance, but whose characteristics are only guaranteed when there is no further undefined driver impedance in series with the input (e.g. Miller effect with a high mu triode first stage as so many amps have); and (2) for driving e.g. tone compensation or other filters with substantially variable input impedance with frequency in order to have predictable characteristics. A cathode follower's relatively high input impedance may also be desirable following such a stage.
Drive a power amp with several meters of interconnect from a triode driver with say 47K as anode load, compared to the same in the cathode (Zout perhaps = 1K2), and you are likely to notice a difference in noise pickup. But the matter you mention of designers sometimes not paying attention to whether their cathode followers can cope with what they attach, is certainly a real one! I have seen some real non-linear cathode followers even despite the feedback in my day.
Regards.
Sorry Dave,
Wanted to add a further remark re the McIntosh (time of the pm I guess).
I would encourage you in that venture. I have chickened out of that one to something half-there, by using a UL output but with the common output transformer "G2 - B+" part of the primary moved down to the cahtode (Quad II type of output configuration). I have had success with that (also listening tests by others!), but one must design one's own output transformer. I am not aware of anything on the market. I presume you will have to do the same for the McIntosh.
Wanted to add a further remark re the McIntosh (time of the pm I guess).
I would encourage you in that venture. I have chickened out of that one to something half-there, by using a UL output but with the common output transformer "G2 - B+" part of the primary moved down to the cahtode (Quad II type of output configuration). I have had success with that (also listening tests by others!), but one must design one's own output transformer. I am not aware of anything on the market. I presume you will have to do the same for the McIntosh.
Johan, you make good points regarding cathode followers. I suppose it is just my personal preference to do without them. At this point I find that I can use a suitably low rp tube plate loaded to get the job done. If I need a VERY low impedance source then I think about a stepdown transformer. Despite the fact that a CF looks better on paper, I have found that a transformer (when needed) just sounds better.
Having said that, your comment about using them to drive a tone network is interesting. I have been helping a friend design a phono stage (really just following along) and the importance of a well defined source impedance has become clear. There are a variety of solutions to choose from. A cathode follower with a relatively large series resistance is something to think about.
Regarding the McIntosh style transformer, you are right that it would need to be custom made, but I wind my own iron, so that is not a problem. 🙂
-- Dave
Having said that, your comment about using them to drive a tone network is interesting. I have been helping a friend design a phono stage (really just following along) and the importance of a well defined source impedance has become clear. There are a variety of solutions to choose from. A cathode follower with a relatively large series resistance is something to think about.
Regarding the McIntosh style transformer, you are right that it would need to be custom made, but I wind my own iron, so that is not a problem. 🙂
-- Dave
Dave, look at Ampeg SVT. It is full of followers (for tone control, to drive output toobs).
But it is not Hi-Fi, it is "the patr of the instrument".
...and it has a bad feedback loop through tone switch and ground wires that causes oscillations...
But it is not Hi-Fi, it is "the patr of the instrument".
...and it has a bad feedback loop through tone switch and ground wires that causes oscillations...
robertc said:
There is a distinct difference between 'zero-nfb', or open loop, and no loop nfb.
I'll have to say that when ever I have built with loop NFB, I have not liked the results. THe same circuit with local FB has always sounded bettter. None of the measurements could point to a specific quantity to define this quality...but I hesitate to conclude that they did not exist.
No loop FB is one I can do very easily. There are a few ways of accomplishing it, depending on what you have available for output TX's.
cheers,
Douglas
Thanks for the reply. I want to investigate cathode followers not so much for the low output impedance but as a means to completely decouple the preceding CCS-loaded gain stage from any current variation. It's for a triode-813 SE and 6C45-pe seems up to the task in both functions.
Re: cathode feedback wrapped around the output stage I also scored an AMI-Rowe 7868 P-P jukebox amp with separate 70 volt windings just for that experiment. 😉
Re: cathode feedback wrapped around the output stage I also scored an AMI-Rowe 7868 P-P jukebox amp with separate 70 volt windings just for that experiment. 😉
- Status
- Not open for further replies.