Michael Gerzon Audio Pioneer
Ed very good post, thank you.
I experience Eugene’s case every day with my mother (only a few seconds memory retention time) and everything you point out is correct.
Studying pathological behavioural cases is like looking at normal cases through a powerful microscope
George
We are short of critical readers (while critical listeners are in abundance)
Which of all examples (illustrated in which plot?) you are referring to?
George
PS I have a relevant question (exercising all DAC codes) but it’s totally off topic. I’ll ask you in another circumstance.
The 1kHz sampled at 44.1K a 32K length record can't contain all 64k codes at one obvious level, but an interesting thought experiment. At a 64k record a ramp from plus to minus full scale does, it there a sine wave at another frequency that touches exactly each code once in 64/44.1 seconds?
I should be more careful with my references, and a bit less trusting, if that is possible. I assumed the reference below would work because it was published in classic-era Electronics World, but it appears I was wrong.
35] Lawson, J "Rumble Filter Preserves Bass" Letter to Electronics & Wireless World, April 1992, p317
I've spent an hour prodding about with it in SPICE and can get nothing sensible out of it. A major stumbling block is that the circuit appears to be wrong- two of the 47k feedback resistors are connected directly to the input. Presumably these are meant to be driven with a low impedance so those feedback paths will do nothing. If anyone can make sense of it I would be glad to know.
My apologies if that reference turns out to be a time-waster.
I do hope I'm not the only concerned about how to implement the oh-so-simple-sounding 'LF crossfeed'. It is not simple at all, and I'm not wholly satisfied with my 4th-order solution.
Work continues, but let's have others joining in.
35] Lawson, J "Rumble Filter Preserves Bass" Letter to Electronics & Wireless World, April 1992, p317
I've spent an hour prodding about with it in SPICE and can get nothing sensible out of it. A major stumbling block is that the circuit appears to be wrong- two of the 47k feedback resistors are connected directly to the input. Presumably these are meant to be driven with a low impedance so those feedback paths will do nothing. If anyone can make sense of it I would be glad to know.
My apologies if that reference turns out to be a time-waster.
I do hope I'm not the only concerned about how to implement the oh-so-simple-sounding 'LF crossfeed'. It is not simple at all, and I'm not wholly satisfied with my 4th-order solution.
Work continues, but let's have others joining in.
Had an interesting experience last evening. My wife was recording some voice readings, and had edited her breathing out by replacing it with silence. I could hear the room ambiance going away each time. It was an interesting experience. Let me know if you want an example, I'll post if you do.
Both.
The removal of antiphase signals might be called Devinylising. It a) allows experiments on the 'ambience' hypothesis which started this thread, and b) would solve very real difficulties in reproducing non-flat disks, especially in a world now full of reflex speakers (made possible by CDs) that have no VLF control of the bass cone.
On the other hand, Vinylising is going to require the addition of VLF anti-phase noise to a CD signal. The amount, frequency, and slope of this required are currently unknown.
Right at the minute I'm concentrating on Devinylising as the more useful process.
A possible snag is that it doesn't work.
I have modelled just the central crossover part of the circuit, and I get a 55dB deep notch at 126 Hz (bandwidth a couple of octaves) for in-phase signals, which is unhelpful.
The anti-phase suppression has a slope of 17 dB/oct at 2 Hz, but only 11 dB/oct at 20 - 40Hz.
I may have screwed up something, but I don't think so. Comments?
(I am aware that more complex crossfeeders are described later in the thread, but I am not enthusiastic about taking the time to try them out, after the initial debacle)
Last edited:
The post 1 circuit. The one in post 30 has no component values and does not seem to be a fully worked out design.
Last edited:
Scott
If I am not bothering you much, what these 64k and 64/ stand for?
I had a day dealing with numbers and FFTs and now I am even more slow than I am during a good day
Sorry for corrupting the thread guys
George
Going back to this reference:
35] Lawson, J "Rumble Filter Preserves Bass" Letter to Electronics & Wireless World, April 1992, p317
Lawson gives two earlier WW references for the LF crossfeed idea:
Macaulay J P, Circuit Ideas WW Sept 1979, p75
Lagvad J, Letters WW Mar 1980, p61
and it would probably be easy to find earlier ones. You could argue it's a fairly obvious idea. I wonder if these two work?
35] Lawson, J "Rumble Filter Preserves Bass" Letter to Electronics & Wireless World, April 1992, p317
Lawson gives two earlier WW references for the LF crossfeed idea:
Macaulay J P, Circuit Ideas WW Sept 1979, p75
Lagvad J, Letters WW Mar 1980, p61
and it would probably be easy to find earlier ones. You could argue it's a fairly obvious idea. I wonder if these two work?
I have now checked the Macauly circuit. It's a beauty, and I expected no less from Macaulay, who haunted the Letter columns of WW for many years.
Firstly, there is a +0.5 dB step below 100 Hz for in-phase signals. Not dreadful, but very clearly not flat.
Of greater moment is that it shows 6 dB of gain for anti-phase signals below 100 Hz. It's worse than nothing.
The original circuit used discrete transistors and I have converted it to opamps, keeping the significant values the same.
I think the sims are sound, but I do wish someone else would check.
I'm starting to understand why this notion has never seen commercial use. Now, for Mr Lagvad...
Firstly, there is a +0.5 dB step below 100 Hz for in-phase signals. Not dreadful, but very clearly not flat.
Of greater moment is that it shows 6 dB of gain for anti-phase signals below 100 Hz. It's worse than nothing.
The original circuit used discrete transistors and I have converted it to opamps, keeping the significant values the same.
I think the sims are sound, but I do wish someone else would check.
I'm starting to understand why this notion has never seen commercial use. Now, for Mr Lagvad...
So far Mr Lagvad wins (I'm not including my effort in the contest) as he published something that more or less works.
It has a damn-near flat response to in-phase signals. (+0.02dB peaking)
Unfortunately the slope of the anti-phase suppression is only -6dB/octave, and there is +2 dB of anti-phase gain around 135Hz. There's the trap...
I've got a fourth-order Devinyliser, and I'm not afraid to use it.
It has a damn-near flat response to in-phase signals. (+0.02dB peaking)
Unfortunately the slope of the anti-phase suppression is only -6dB/octave, and there is +2 dB of anti-phase gain around 135Hz. There's the trap...
I've got a fourth-order Devinyliser, and I'm not afraid to use it.
Sorry for being dim, is there a general WW archive were we can find the letters?
I'm zero use on the SIM front but trying to wrestle a re-installed audacity into submission to see if there is anything that can be done on some real files for testing.*
I do need to mimble off and reseach something that is troubling me and too stupid for me to ask here though.
*Note that whilst this is opening up all sorts of avenues for experimentation for which I am grateful for the nudge I have to admit that all the best 'ambient soundstage' I have heard in my system has been off CD.
I'm zero use on the SIM front but trying to wrestle a re-installed audacity into submission to see if there is anything that can be done on some real files for testing.*
I do need to mimble off and reseach something that is troubling me and too stupid for me to ask here though.
*Note that whilst this is opening up all sorts of avenues for experimentation for which I am grateful for the nudge I have to admit that all the best 'ambient soundstage' I have heard in my system has been off CD.
So k in this case means kibibyte(kiB)=1024byte. Therefore 64k (binary) means 65536byte (decimal).
Now we communicate (on another thread)
Thanks for the patience.
George
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.