This is simply poetic license (to be polite). It leads to the old feedback goes round and round argument. A closed loop amplifier is a continuous time system, the signal does not go through it and come out later with a correction applied. I wish folks would let go of this concept.
I either case consider an error feed forward system, if well implemented your "tainted" signal has the taint subtracted at the output.
Thanks for your gracious reply Jan. I do have several comments which I feel apply.
1) the differences between our device distortions. Yours is 1-10%, mine is in the .01% in my preamplifiers.
2) No offense meant, but you have performed no testing, just reading.
3) I see no proof that those are all the distortions. How about masking or frequency response distortions? A Caddock sounds different than a Mills non-inductive (which has a very small amount of inductance etc)
4) What "controlled" tests are you referring to, measurements or listening? With measurements one must deal with all forms of distortions. With listening tests, dbts have been shown to be unreliable at best.
Sorry about the distraction. Of course I believe fish was on the menu some pages back.
Cheers.
HI Bob,
Really not surprising to me since the average listening spl was close to 86db at a meter, give or take. Peaks 104db give or take. Was it a large room?
Thanks Bob.
Cheers.
Last edited:
Indeed. And all physical systems have resonances and nonlinearities, and all the talk about dogs chasing tails by the anti-feedback brigade seems to imply that a non-feedback structure is somehow free of these supposedly evil dynamical effects.
Now, if one is after blinding speed (oscilloscope vertical amplifiers for instance, back before sampling approaches prevailed) you won't resort to simple global feedback, usually. But these applications are far beyond the highest-frequency signals encountered in audio, unless something has gone horribly wrong.
Yes, the cascomp pioneered at TEK is a good example of feeding a replica of the error forward and subtracting.
It is important to distinguish between the simple discrimination of a waveform shape and what that waveform shape change would do to the discrimination of a virtual sound field image.
The last research I read on the topic seemed to be focussing on absolute zero crossings and localization. Mods to the waveform shape would indeed alter that.
How is this assertion supported? Anecdotal?
Also, how does a Mills reduce it's inductance?
jn
And indeed it does! Not only on localization. Phase coherence of sounds with subwoofer (<40 Hz!) matters as well, in terms of "natuallness" of such sounds like drums and even triangles!
Speaking of "error correction" and "feedbacks", I am playing now with an output stage using TV sweep output tube. It has dense screen grid and relatively rare control grid, so when controlled by screen grid it is more linear, when controlled by control grid it is more sensitive. What I am trying to do now, to find balance between sensitivity and linearity driving both grids, with additional feedback from anode to the control grid. Looks promising!
Like in power wires we discussed. Bifilar winding.
However, it increases parasitic capacitance...
there is a difference between poetic license and polemic
Artistic license - Wikipedia, the free encyclopedia
Polemic - Wikipedia, the free encyclopedia
the "Tsunami of distortion" is the latter
Artistic license - Wikipedia, the free encyclopedia
Polemic - Wikipedia, the free encyclopedia
the "Tsunami of distortion" is the latter
Last edited:
Meany. You took all the fun outta it. I was waiting for positron to answer.Like in power wires we discussed. Bifilar winding.
However, it increases parasitic capacitance...
I've screened over 10 thousand of the non bifilar style for liquid nitrogen dunking/oven bake at 150 C, looking to verify stability of resistance. They are all operating in liquid helium now, none have failed, it's been 14 years now. The MTBF for this construction resistor even at 4.5K is just unbelievable.
The real reason I asked, is because despite the bifilar winding topology, the Mills resistor still has a length, it still has a nominal current path centroid, and as a result of being physically longer than a typical caddock, will have a higher inductance. The only way around that would be to lower the loop between the current centroid of the resistor and the return trace (if there is one in proximity.
So the statement that a Mills has a lower inductance than a caddock (thick film) is certainly unsupported.
And any reasonable method of reducing the inductance of either requires bringing the return current path into intimate proximity to the resistor.
To make a truly non inductive resistor requires folding the current path back through the current centroid. This is a trivial thing to do. ***
It is a simple thing to get some power while dropping inductance below 250 picohenries. And, there will be no external magnetic field (of course, that is a consequence of almost no inductance.
jn
***edit: for example, if you take a mills bifilar, and use a copper braid over the body such that it connects to the far end bringing the current back over the body, you create a coaxial "resistor structure. There will be no magfield outside the braid, and the inductance will be limited to magnetic flux within the body of the resistor...probably in the range of 20 to 50 nanohenries per foot, a half inch long resistor would probably be in the range of .8 to 2 nanohenries total. Below the test capability of most people.
Last edited:
Unsupported statements? In this thread? Please, John, tell me it ain't so!
Jan,
.001% !
JN
Mills v Caddock big measured difference!
SY
Analyzer = hardware item. Think goldline, sencore, neutral. Dick probably has an HP which will show phase.
Scott
As I wrote it is not an adaptive filter!
DF
I had trouble with Boltzmann etc. at first because I was 13 at the time.
Waves
If you don't like Boltzmann stay away from resistors!
.001% !
JN
Mills v Caddock big measured difference!
SY
Analyzer = hardware item. Think goldline, sencore, neutral. Dick probably has an HP which will show phase.
Scott
As I wrote it is not an adaptive filter!
DF
I had trouble with Boltzmann etc. at first because I was 13 at the time.
Waves
If you don't like Boltzmann stay away from resistors!
My own research via comparitive listening testing over the years.
From Mills:
I doubt that but see no inductance specs.
Cheers.
Just making sure that this was an anecdotal statement, not one based on verifiable measurement.
I did not state that your opinion is incorrect, just anecdotal. In point of fact, your design topology in using both types could very well have caused a difference you discerned. But your thinking that it is because the mills is lower inductance is incorrect.
Actually, I don't doubt what they've stated. Most of it is quite accurate, the bifilar is indeed called a "non inductive" construction.
My point is, the physical size and current path of the Mills can be higher inductance than a thick film caddock. Your assertion that it is the opposite is not correct.
But either can be made far less inductive if one is creative.
Sub nanohenry power resistors are trivially made by the diy'er, but they are not very marketable.
jn
Actually I was not thinking of inductance, but of element materials used. The inductance is too small for me to conceive of sonic differences. (Small resistances used.)
However, it is interesting since I checked and saw no "non-inductive" comments on the Caddocks while Mills is listed as non-inductive. I take it you measured both, Mk-132s and Mills RA types?
Cheers John.
Pos.
ps. I am the one from IL we emailed about Pin 1 sometime back.
Last edited:
Big being relative of course.
I do not believe you have the capability to measure either for inductance to within 10%. It is neither easy nor obvious how to measure below half a nH.
jn
- Status
- Not open for further replies.