If you simply invert our hearing bandwidth, that is the number you get.
However, researchers have tested, reported, duplicated results where humans are indeed capable of discerning ITD at the 1.2 uSec level midband out to 12Khz with a dithered signal, and at the 5 uSec level for undithered midband.
A quick literature search will show your numbers entirely incorrect.
jn
I would appreciate the link as well. I think sorting out the hidden ground currents and loops is one of the most vexing audio challenges and possibly why equipment that measures orders of magnitudes better than any hearing experiment validates can sound different. Even connecting a probe to look at voltages disturbs the matrix of currents.
Do you have links to any non-headphone studies? BTW one of the keynotes at this years ISSCC will be a guy from CERN telling us how chips helped find Higgs. He managed to get "atto-world" into the title. I still vote for zepto-world, hail Fredonia!
No. No links to other forums.
I note with chagrin
that my diyaudio gallery has NO entries whatsoever. I guess I've been downloading pics without storing them.I apologize for that. I will put all I have into my diy gallery very shortly. Unfortunately, they are not on the computer I am currently on, so it will be later today. All I gots here is a phase diagram overlay between tin/silver and lead/tin, and measurements of some air core inductors vs frequency both on and off a copper pcboard plane.. nothing of interest for this forum..
No links, I tend to print them out. I do note that Greisinger appears to be the farthest along in understanding, and I believe his best to date was put online on 11/11/11. Check his site. He discusses his localization abilities given a string quartet I believe, and ITD based side to side differential from a specific distance in the audience and a discernible instrument horizontal spacing.
I've not seen him attempt to apply nor map human hearing for closely spaced images given ear to ear crosstalk, where 4 images are trying to establish. He tends for the movie special effects, like sound effects far outside the angular span of the speakers.
I also recall a paper discussing human responsed to quick ITD shifts such as a car zooming right to left, but again, not a typical home listening enviro.
I'll look for what I can find, as I am not where those papers are.
All those guys working with particle accelerators/colliders are yahoos. Don't listen to em, they are just plain nuts.
jn
ps..AHA!! found em. Be patient, most are jpegs but some are pdf's
pps. they're in for the most part, I'll update the comment fields when I can.
Last edited:
Amazing stuff they can do with the fake images, even from ordinary stereo.
It can get really interesting. The guys here designed a chassis to amplify a 500 MHz pulse stream. After the input low level high gain section, they high passed it heavily before doing the a/d conversions. They found that switchmode supply artifacts on other systems made it through everything, and couldn't understand it.
They had a ground loop, and the ground loop currents were going from the BNC's to ground through the chassis, and the Analog circuitry after the high pass was directly in line between the BNC's and the chassis IEC ground. So the currents coupled into the signal after all the filtering.
Winding the input coax cables through a toroidal ferrite multiple times attenuated the shield currents, but not the signal currents.
jn
The drawings are great. I think I have most of what is necessary for the testing in the last drawing. Only the excitation core/coil is not in my collection. What do you propose? Its similar to some of the standard conducted interference tests. Is 20 KHz enough? Switching power supplies and amps go way higher and dump the energy everywhere.
Unfortunately the real world has more stuff plugged in so it gets far more complex fast.
Unfortunately the real world has more stuff plugged in so it gets far more complex fast.
One thing not obvious usually is the nature of the source of the noise. If its a current with a high enough impedance (1k +) the common mode chokes don't help much. This is the case with leakage through a transformer. The source voltage is high enough and the impedance high enough that you have essentially a constant current excitation of the loop. And then the filter you add to the power input just serves to share the noise with more conductors.
I drew the core I used. It was opportunistic for me as it was on hand.
All you need it a toroid you can put the line cord plug through. It also needs to be of sufficient BW, so ferrite is best, or a thin laminated output transformer toroid. I tested a plitron 200 VI core, but it rolled off way too fast.
Base the core on what you're trying to do. I used a QSC rmx 1450 for the drive, so limited it to 20K. I suspect there will be need of higher bw in many applications.
I used the toroids to get away from using a current viewing resistor in the path, as the loop resistance change will directly impact what you are trying to measure.
jn
Ah, couldn't figure out your verbage for a moment. That is because I neglected an important thing.
The chassis currents did not cause a ground IR drop. The a/d printed circuit board was close to the chassis bottom, and the currents flowing underneath the printed circuit board inductively coupled to the board circuits.
jn
If you design them, or just have to deal with them, get tom van doren's presentation. Lots of good things in it.
jn
davym,
Or anyone else could you explain to me how an optical link can remove the ground plane induced noises or any other noises that I would assume are superimpose over the original signal earlier in the chain? It would almost seem like a Bybee device type of situation where the question is how does the optical link distinguish what to reproduce and what to remove? Is this just a bandwidth limiting situation removing any extraneous noise outside the chosen bandwidth?
Or anyone else could you explain to me how an optical link can remove the ground plane induced noises or any other noises that I would assume are superimpose over the original signal earlier in the chain? It would almost seem like a Bybee device type of situation where the question is how does the optical link distinguish what to reproduce and what to remove? Is this just a bandwidth limiting situation removing any extraneous noise outside the chosen bandwidth?
- Status
- Not open for further replies.