if you don't see the connection between the two issues that Robert has stated then yes, maybe design camp is appropriate?
Hi mmerrill99,
I didn't say there were no connections between the two. Robert introduced each concept separately.
The entire design is important, something I've always said. I think you will find that the assembled talent here is well beyond those basic concepts. The tone taken should be more respectful, wouldn't you say?
-Chris
I didn't say there were no connections between the two. Robert introduced each concept separately.
The entire design is important, something I've always said. I think you will find that the assembled talent here is well beyond those basic concepts. The tone taken should be more respectful, wouldn't you say?
-Chris
Possibly my tone was due to the tone of your post to Robert - maybe it was said in jest & I missed this nuance (this word seems to be my mot de jour)?
Maybe you missed the details of what KSTR does?
Or maybe not & you can tell us the details of your testing?
Nope, didn't miss the message in your post that both concepts were introduced separately - I just didn't think it important, needing any reply/comment
Hi mmerrill99,
I will say that a differential probe across the bases of the diff pair (or gates) can be instructive. Of course, you can also use an op amp or another diff pair to amplify the difference signal, saving the cost of a differential probe (they aren't cheap). Mine is a Tektronix amplified differential probe, P6046.
Can you tell us any details of your studies along the same general idea? You should be able to guess what I've been looking at from my last few posts.
-Chris
Edit:
I haven't the time for going into detail. I was pointing out that the sole purpose of the differential pair is to subtract input from a scaled down output sample. The result being inverted distortion products and a low amplitude desired signal.
I will say that a differential probe across the bases of the diff pair (or gates) can be instructive. Of course, you can also use an op amp or another diff pair to amplify the difference signal, saving the cost of a differential probe (they aren't cheap). Mine is a Tektronix amplified differential probe, P6046.
Can you tell us any details of your studies along the same general idea? You should be able to guess what I've been looking at from my last few posts.
-Chris
Edit:
... and yet you did anyway.
Neither have I got the time but I do recognize that the usual amplifier configuration of diff input pair & global feedback is very different to KSTR's method & I recognize the usefulness of KSTR's method having tired to use Diffmaker & finding it unreliable with test signals.Hi mmerrill99,
I haven't the time for going into detail. I was pointing out that the sole purpose of the differential pair is to subtract input from a scaled down output sample. The result being inverted distortion products and a low amplitude desired signal.
I will say that a differential probe across the bases of the diff pair (or gates) can be instructive. Of course, you can also use an op amp or another diff pair to amplify the difference signal, saving the cost of a differential probe (they aren't cheap). Mine is a Tektronix amplified differential probe, P6046.
Can you tell us any details of your studies along the same general idea? You should be able to guess what I've been looking at from my last few posts.
-Chris
Nope, incorrect - I replied to your suggestion that I "missed the message in your post" not to your original point - two very different things.
Who, when, what, why er what was the subject?
Scott did you notice in Howie's connector problem discussion about intermittent low level signals?
Sure, but I suspect that unavailability of published measurement of complete audio chain distortion contributes to major misconceptions by both user and designer. I highly doubt that most amplifier designers have sufficiently clear global system view because the last chain of sound reproduction is seldom given enough attention.
Eduardo de Lima clearly indicated years ago that there is a specific way to design an amplifier that will reduce generated distortion of the whole chain. I noted short remarks like illusion, not accurate and such on my bringing the paper to attention with no clear reason, no supporting data and no answer yet on further question.
I missed that paper & needed to do a search as the PDF name needed changing
http://www.audiopax.com/e/pdfs/artigos/Whysingle-endedtubeamplifiers.pdf
I'll have a read when I get a chance, thanks!
http://www.audiopax.com/e/pdfs/artigos/Whysingle-endedtubeamplifiers.pdf
I'll have a read when I get a chance, thanks!
Intentionally creating 2nd harmonic distortion, in a 1979 product (!)
I found this schematic in Audio Magazine, February 1979. The article was a product review of a phonostage sold by PS Audio, a company that is still in business today. See attachment below.
Each of the two Long Tailed Pairs is intentionally unbalanced, by adding an emitter degeneration resistor (330 ohms) to only one of the two legs of the Long Tailed Pair. This creates 2nd order harmonic distortion and apparently that was a design goal. Somebody knew something, long long ago.
One way to think about the circuit is this: Imagine that the 330 ohm resistor is a potentiometer, whose wiper connects to the 39K bias resistor. With the wiper at the midpoint (165 ohms left, wiper, 165 ohms right), the circuit is balanced. But the designer has intentionally chosen to turn the pot all the way clockwise, and now it's (330 ohms left, wiper, 0 ohms right). The circuit is no longer balanced.
By the way, with ±32V supply rails, the circuit ought to have huge amounts of overload margin, even though the equalization is passive.
I found this schematic in Audio Magazine, February 1979. The article was a product review of a phonostage sold by PS Audio, a company that is still in business today. See attachment below.
Each of the two Long Tailed Pairs is intentionally unbalanced, by adding an emitter degeneration resistor (330 ohms) to only one of the two legs of the Long Tailed Pair. This creates 2nd order harmonic distortion and apparently that was a design goal. Somebody knew something, long long ago.
One way to think about the circuit is this: Imagine that the 330 ohm resistor is a potentiometer, whose wiper connects to the 39K bias resistor. With the wiper at the midpoint (165 ohms left, wiper, 165 ohms right), the circuit is balanced. But the designer has intentionally chosen to turn the pot all the way clockwise, and now it's (330 ohms left, wiper, 0 ohms right). The circuit is no longer balanced.
By the way, with ±32V supply rails, the circuit ought to have huge amounts of overload margin, even though the equalization is passive.
Attachments
Last edited:
Full agreement to that. If you have access to a direct error signal (diff of +in and -in of any op-amp type circuit) it is a perfect testpoint as the amp already minimises the error signal. It may not actually be trivial to measure things this way but the method is proven and useful, as is looking at the residual output of a distortion analyzer (limited to sines, though).
The method I have described is conceptually the same, just that the "diff-amp" and any required trimming is done off-line by software/processing after sequential digitizing using only one DAC/ADC-channel+hardware. That's the main difference and for some applications this adds some flexibility and increased resolution.
Last edited:
Isn't this just an ersatz Apt-Holman input with degenerated NPN instead of FET?
The power supply is all important; way, way more than most suspect. Even great designers don’t see it. Wanna’ go over specifics?
In the sense that both of them use ThingX on the left side of an LTP and ThingY on the right side of the LTP, where X!=Y and more importantly where gm_X != gm_Y , yes they both do that.
However I favor the explanation that there is a balance potentiometer which the designer has deliberately twisted all the way to the endstop, "with malice aforethought", to intentionally unbalance its operation.
But that also hints that there is something about deliberately unbalanced inputs. Maybe?
Oh yes.
Some people hold opinions such as these:
Those who wish to read more are encouraged to use the box called "this exact word or phrase" in Google Advanced Search
Some people hold opinions such as these:
"For reproducing music as naturally as possible, push-pull operation is not the best approach. Air is not symmetric and does not have a push-pull characteristic. ... Only one linear circuit topology delivers the appropriate characteristic, and that is the single-ended amplifier. Single ended amplification only comes in pure class A ..."
Those who wish to read more are encouraged to use the box called "this exact word or phrase" in Google Advanced Search
- Status
- Not open for further replies.