Hi John,
I've said similar things in the past myself. I do agree, but understand that I do not manufacture anything. I service and most of the things I find out are loosely related to what I may read and 100% based on what I see happen in the physical world. Besides, I have a reputation for that kind of work.
In fact, a 1% match between the compliments in a complimentary differential pair represents a lot of work. I don't usually get them that close, but certainly much closer than the production line delivers them. To say I was 1% or closer on pairs and probably around 5% on the compliments might be fair. I do know from experimenting that a closer match does a couple things for the design. It makes the bias current a little more stable (many depend on the current in the diff pair), and it does make the DC offset both lower and more stable. Take an amp with 60 mV DC offset that drifts and turn out one with less than 5 mV that is stable. No more clicks and pops and it will both sound a little better (not dramatically as I would describe it, some feel that way) and measure slightly better. The sound quality difference is greater than a total distortion number might suggest.
There are amplifiers that this effort is wasted on, so I don't waste my time or the customer's money on those. Everything in balance.
This might also be reflected in your efforts to reduce the seventh order distortion if you optimized everything about that issue. Wire types and connections could be construed that way as well. My experience comes from years working on other designs and also prototypes and experiments. I follow those things that actually do make a difference that can be heard by several different people in different rooms and systems. Something I can measure repeatedly also counts for a real change, could be good or bad. Either way, it's noted.
I think we are on the same page there John. We are saying the same thing.
I wonder. John, have you ever tried a super close match to see if there is any effect on the seventh harmonic? Given that this work is done in the error detector / amplifier of an amplifier, is it worth one experiment on your own? Yes, I do know we are mostly looking at even harmonics and that odd harmonics may be affected more in the Vas / current sources / or output stages, but one never knows until the numbers come in. BTW, numbers are only a way to measure or quantify a performance change. They are valid when taken along with the rest of the data and information. Otherwise, none of us would have any sort of distortion measuring apparatus (including spectrum analyzers or wave meters).
-Chris
I've said similar things in the past myself. I do agree, but understand that I do not manufacture anything. I service and most of the things I find out are loosely related to what I may read and 100% based on what I see happen in the physical world. Besides, I have a reputation for that kind of work.
In fact, a 1% match between the compliments in a complimentary differential pair represents a lot of work. I don't usually get them that close, but certainly much closer than the production line delivers them. To say I was 1% or closer on pairs and probably around 5% on the compliments might be fair. I do know from experimenting that a closer match does a couple things for the design. It makes the bias current a little more stable (many depend on the current in the diff pair), and it does make the DC offset both lower and more stable. Take an amp with 60 mV DC offset that drifts and turn out one with less than 5 mV that is stable. No more clicks and pops and it will both sound a little better (not dramatically as I would describe it, some feel that way) and measure slightly better. The sound quality difference is greater than a total distortion number might suggest.
There are amplifiers that this effort is wasted on, so I don't waste my time or the customer's money on those. Everything in balance.
This might also be reflected in your efforts to reduce the seventh order distortion if you optimized everything about that issue. Wire types and connections could be construed that way as well. My experience comes from years working on other designs and also prototypes and experiments. I follow those things that actually do make a difference that can be heard by several different people in different rooms and systems. Something I can measure repeatedly also counts for a real change, could be good or bad. Either way, it's noted.
I think we are on the same page there John. We are saying the same thing.
I wonder. John, have you ever tried a super close match to see if there is any effect on the seventh harmonic? Given that this work is done in the error detector / amplifier of an amplifier, is it worth one experiment on your own? Yes, I do know we are mostly looking at even harmonics and that odd harmonics may be affected more in the Vas / current sources / or output stages, but one never knows until the numbers come in. BTW, numbers are only a way to measure or quantify a performance change. They are valid when taken along with the rest of the data and information. Otherwise, none of us would have any sort of distortion measuring apparatus (including spectrum analyzers or wave meters).
-Chris
anatech said:
Hi Chris;
did not you see color coded and letter coded devices already?
We had such a process back in 1970'th in Russia: the same devices were automatically measured and sorted out right on conveyor, before their automated labeling. However, sometimes production people used to pour some very good devices to the bin with worse ones, because there was a State Plan, to produce certain amounts of devices with particular letter codes. That's why, knowing how the manufacturing works in reality, I often used a characteriograph to find real jewels among almost rejects. Such a way I got many matched pairs of 60W complementary transistors with equal betas close to 200 and breakdown voltages more than 150V, from a batch with Beta > 25 and Vcb <50.
Selecting matching pairs such a way don't mean rejecting other devices. Single device does not mean unhappy one, like in case of human beings.
http://www.diyhifi.org/forums/viewtopic.php?p=40359#p40359
Hhm, looks like he's tied up with something great to come...
Hhm, looks like he's tied up with something great to come...
The only surprise is JC getting into such (otherwise, it was done 20 years ago). I wonder what the (hobbyst and professional) purists will comment about. And of course, I can't wait to read a M. Fremer rave review. The same guy that was complaining about the digital circuitry (MCU, remote, etc...) in a preamp, as deeply impacting the sound.
Why not?
Class D may sound really nice. My class D recording amp in a tape recorder sounded extremely nice. It was a bias frequency square signal modulated by width by recording signal.
I'm sure it's an open loop class D. Everybody in the high end audio knows about negative feedback being badbad, even for digital stuff.
Open loop class D, the best thing since sliced bread.
Sure, it was open loop! Audiophiles would be happy. I was thinking about power amps, but there were no good transistors available then. But a tape recorder sounded really, really well! Even noise level of a tape went down significantly.
Interesting.
A truly new technology. Class "D" and linear combination sounds a great deal like ...... a Carver Lightstar (series one of course)?? !
The hard part is to keep emissions down on the "down converters". I wonder what John will call these?
Anyway, it's sure to be a secret and well beyond what mere mortals can deal with. Bob Carver is not really of this world you know, that's why his technology is so far beyond ours.
I wonder if Bob Carver helped? If it works without throwing a shower of sparks, then Bob Carver did help out.
-Chris
A truly new technology. Class "D" and linear combination sounds a great deal like ...... a Carver Lightstar (series one of course)?? !
The hard part is to keep emissions down on the "down converters". I wonder what John will call these?
Anyway, it's sure to be a secret and well beyond what mere mortals can deal with. Bob Carver is not really of this world you know, that's why his technology is so far beyond ours.
I wonder if Bob Carver helped? If it works without throwing a shower of sparks, then Bob Carver did help out.
-Chris
It can be serial (voltage source) or paralel(current source) combination of linear and class D amps.Comercial example for first metod e.g. LabGruppen amps (tracking bipolar power supply and analog filter). Both methods are well known. And both combined problems from class D and analog amps .
.
Last edited:
Interesting.
Anyway, it's sure to be a secret and well beyond what mere mortals can deal with. Bob Carver is not really of this world you know, that's why his technology is so far beyond ours.
I wonder if Bob Carver helped? If it works without throwing a shower of sparks, then Bob Carver did help out.
-Chris
We already know it could be tailored for any sound.
- Status
- Not open for further replies.