john curl said:
GAS evolved to a few poor souls sitting in front of a test fixture measuring and dipping the various transistors in paint to mark their grade. The original amps used MPQ-07 and 057's, later versions went with the 2n5551-5401's; lost a bit of resolution in the process (to my ear).
MikeBettinger said:
Thank you for the informative post. I agree with everything you wrote except for the above comment on the various configurations of output transistors. (I also disagree with Bongiornio's lack of respect for passive parts quality, but perhaps you do as well as you didn't mention that in your post.)
In the article he compared four different ways to connect three series-connected output stage transistors (ie, pre-driver, driver, output). One way was an emitter follower triple (like the T-circuit that was invented by Locanthi). The others were either various combinations of complementary feedback pairs, or in one case a complementary feedback triple (to coin a phrase). So far, so good. Everything is fine, and it makes sense to do an analysis like this.
Bongiornio claimed the emitter follower triple had the highest distortion, while the complementary feedback triple had the lowest distortion. The two variations of complementary feedback pairs fell somewhere in the middle, and he chose one of these for his project. The complementary feedback circuits are called that because they create a small feedback loop around two (or three in the case of the CF-triple) of the transistors. As feedback normally does, it then lowers the measured steady-state distortion of a given circuit.
But in my estimation he made a mistake in his analysis. If you think about it for a moment it is quite obvious that the CF-triple had the highest amount of feedback and hence the lowest amount of measured steady-state distortion. But instead of looking at things in a straightforward fashion (ie, more feedback gives lower measured distortion), he came up with some strange explanation regarding "base-emitter junctions". In his explanation, a base-emitter junction was the source of the distortion. So an emitter follower triple was the worst design, as it had three of these nasty things in a row. The CF-triple was the best, as it had only one B-E junction, but it was too hard to stabilize for the "amateurs" who were going to build this project. So he compromised by choosing one of the CF-pairs -- it had only two B-E junctions to mess things up.
To my way of thinking, Bongiornio's explanation is not only not helpful, it is misleading. To this day, there are still people who believe that there is something inherently bad about a "base-emitter junction" and that a good circuit should avoid them to whatever degree is possible. It was such a posting on this thread that brought up the whole topic, and I wanted to go on record as saying that I don't believe Bongiornio's explanation, nor do I believe it to be useful.
We now return you to your regularly scheduled programming.
SY said:
No.
With an emitter follower triple, each device is running (just below) unity gain. But with a complementary feedback pair (or triple), then one or more of the transistors is run as a common emitter stage with gain. Then the output is fed back (gee, feedback!) to the input stage such that the overall circuit has unity gain. But the steady-state distortion as measured with sine waves on the test bench will be reduced by however many dB of gain is in the common emitter stage.
Probably the best book to read on this is Self's book on power amps. He does a lot of comparisons between emitter follower doubles and complementary feedback pairs. I don't agree with his conclusions at all (he doesn't even really mention triples, which is ridiculous, and of course he concludes that CFPs are better than EF doubles) but he does a good job of explaining things as he goes.
Charles, thanks for the clarification. In my naive way, I think, hmmm, if I stick the 8 ohm load in the collector, I'll get gain, if I stick it in the emitter, the gain drops to (essentially) unity. And that difference is the feedback and the degree of linearization compared to the common emitter circuit. I'll read the treatment in Self that you suggested.
BTW, for anyone interested, I posted both the infamous Curl, Leach, and Jung letter and the Cordell response, all unedited, over on my nascent web site. My thanks to a little birdie that flew by and dropped these in my lap.
http://home.comcast.net/~syaniger/Cordell 3_31_1980_Letter.pdf
http://home.comcast.net/~syaniger/TIM Forum 1980.pdf
Any discussion of these here, please keep stuff above the belt, folks.
BTW, for anyone interested, I posted both the infamous Curl, Leach, and Jung letter and the Cordell response, all unedited, over on my nascent web site. My thanks to a little birdie that flew by and dropped these in my lap.
http://home.comcast.net/~syaniger/Cordell 3_31_1980_Letter.pdf
http://home.comcast.net/~syaniger/TIM Forum 1980.pdf
Any discussion of these here, please keep stuff above the belt, folks.
Okay, I have read both letters and will for the moment restrict myself to two comments:
1) Of the two letters, it seems to me that Bob's is more...emotional, shall we say, in content. I'm not saying that he doesn't include his rationale for the points he is trying to make, but between those points there is more angst evident. I'll be interested to see if others get the same impression.
2) Is there an accepted, formal, test for slew rate? One of the letters (I think it was Curl, et. al.) mentioned rate of change from 0 to -, from 0 to +, from + to -, and square waves, but without specifying a single, universal bench measurement. I seem also to recall Nelson mentioning something about slew rate recently, but I got the impression that he was running from rail to rail, but only looking at the region around 0 crossover.
Grey
1) Of the two letters, it seems to me that Bob's is more...emotional, shall we say, in content. I'm not saying that he doesn't include his rationale for the points he is trying to make, but between those points there is more angst evident. I'll be interested to see if others get the same impression.
2) Is there an accepted, formal, test for slew rate? One of the letters (I think it was Curl, et. al.) mentioned rate of change from 0 to -, from 0 to +, from + to -, and square waves, but without specifying a single, universal bench measurement. I seem also to recall Nelson mentioning something about slew rate recently, but I got the impression that he was running from rail to rail, but only looking at the region around 0 crossover.
Grey
Hi, Glen,
To be balanced, we should consider at what time/condition a certain thing happens. No internet or diversed text books back then.
Given the time weighing, I think Wright brothers seeing and flying their aeroplane has the same pride and importance with a man who make and see his space shuttle going to the orbit.
To be balanced, we should consider at what time/condition a certain thing happens. No internet or diversed text books back then.
Given the time weighing, I think Wright brothers seeing and flying their aeroplane has the same pride and importance with a man who make and see his space shuttle going to the orbit.
Luman, they were still using biplanes in commercial service 28 years after the Wright Brothers flew their first plane. Keep that in mind when comparing what was happening then, when we made our arguments and today. It is only younger people that think that audio has changed so much or that things have changed, in general, in important ways.
The measurement questions were exactly the same then, as now. However, with MP-3 and even CD taking over the middle class, the slew rate requirements have probably reduced rather than increased in importance.
Just think, the standard mid fi IC is NOW the 4558, compared to the 741, when we first did TIM measurement. That is all the real progress that has been made for the general public.
The measurement questions were exactly the same then, as now. However, with MP-3 and even CD taking over the middle class, the slew rate requirements have probably reduced rather than increased in importance.
Just think, the standard mid fi IC is NOW the 4558, compared to the 741, when we first did TIM measurement. That is all the real progress that has been made for the general public.
john curl said:
It is kind of sad.
Recently I was at an airport. Here is an environment surrounded by probably millions of dollars of exotic electronics, RADAR, high tech communications, automated weather systems, micro-burst wind shear detection, computerized scheduling and ticketing machines.
Yet, the PA system was utterly un-intelligible
- Status
- Not open for further replies.