janneman said:
Hi Jan.
Your statement regarding capacitance is correct, I also do not see caps in the .05 uf..
This is not what I speak of..
Completely disregard capacitance..neglect it, it doesn't exist.
Disregard the cable inductance (for now), it's prop delay, it's impedance, the input impedance of the amp, the output impedance of the source..everything...
Ok, now..
Plug an amp into the wall.
Plug a source into the wall.
Connect the source to the amp using an IC.
You have formed a loop. Actually, you have formed two, one of them is the shield through the grounds/neutrals/xfmr capacitance, whatever reactances are there if not a hard ground loop.
The voltage that is generated by this loop is governed by faraday's law of induction.. The voltage generated is equal to
- rate of change of flux per unit time that is trapped by the loop..
If you have an area of 1 meter square formed by the loop, it will generate 1 volt around that loop when the magnetic flux that goes through that loop changes at a rate of 1 tesla per second.
Earth's magnetic field is half gauss, or 1/20,000th of a tesla, (for scale).
This loop is what picks up the hum when you have a ground loop induced hum. This loop is what picks up the pop of a sub when you hear that. Do not underestimate the contribution simply because of the "Tesla per second" stuff. It is an effect of significance because we have all had the issue..
Now, Lenz's law states that the polarity of induced emf in a loop will fight the rate of change of flux through the loop. This is what the shield of the IC will support. This is the current that is intercepted by the blue region in the drawing I put up way back. This current is the deriviative of the flux through the loop..the conductivity of the shield of the IC will supress the actual voltage that will be generated across the IC shield...
Wait...didn't I say, two loops??
The hot wire in the ic, what's the story with that?
It also intercepts the time rate of change of the flux in the loop..but, it's in series with the 100 ohm output and 10 k input. So, where does that induced voltage show up??? Simple..at the input of the amp, voltage divider style, 10k and 100 ohm..
That voltage will depend on the trapped flux in the loop. Moreso, it will depend on the rate of change of the trapped flux in that loop.
That is what I am talking about..not cable capacitance.
Cable inductance..(eventually I got to it..)
Inductance is defined as the relationship between the energy stored in an entity as a result of the current through it.. E = 1/2 L I2. More inductance means more energy is stored. Space wires apart, more inductance..because the field is able to express itself farther out from the wires physically.
One consequence of increasing the inductance by increasing spacing and allowing the field to express farther, is that it is now more susceptible to intercepting external time varying flux. A better transmitter is a better receiver.
Take a large power amp, an AC powered source, and a portable source, common outlet for the line stuff..
Push major audio power into the load using the battery powered source in the amp...measure the change in potential between the amp ground and the not connected source ground, with a 10K resistor connected between them.. If there is a voltage present, it is a result of the loop intercept. A shield WILL NOT stop that voltage because it does not shield the inner conductor of an IC from Faraday's law of induction directly.
If any of the audio signal can be found there, it is because the audio signal is causing the flux through the loop to change at the audio rate..THAT is either positive feedback, or negative, or some phase shifted artifact..it is slew rate dependent, and that is what will cause the 5, 10, 100, whatever, shift in ITD...don't forget, two IC's, two somewhat independent loops..
Cheers, John
PS..In my office at work, I have two outlets on opposite walls. There is 1.5 volts AC between the ground pins of the outlets, this is not IR drop, it is loop induced, by all the emf garbage at 60hz in the environment.. Others can easily do this test.
Perhaps the balance control is useless because the effectivepath length is more unequal than 3mm. You might measure from the seating planes of your headphones to your eardrum. In this way you could optimize things by reversing the way you wear the headphones... despite our inherent beauty, the human head is generally asymmetrical. This might be used to your advantage.
poobah said:
When I first encountered this goop, after the initial 😕 😕 thing, I tried several things.
Shorted tip and ring, image mono and centered now. 'Ducers seem identical.
Reversed the phones...effect followed the phones..so, not my ears....maybe drivers still??
So, reversed the tip and ring at the jack.. effect followed the electrical connection...so it was not the headphone drivers. Then swapped phones again, as a check..it moved in response..
Far as I could evaluate, the effect is strictly channel to channel something, possibly just internal balance level?
So, used the on screen balance control to center the image..
Blah, it doesn't make it better..yes, the image grossly centers, it is repeatable, swap phones and balance from other direction, same amount of balance gives gross centering.. This, I conjecture is because we do not respond to the timing shift uniformly across the entire audio band, as fcseri stated. So, some of the image is solidly centered by equal level without regard to timing, like the lows and the highs... but where we are sensitive to timing, mids, they seem to one side..so I use the balance control to move the mids back...and the whole kaboodle slides to the left..great...the mid vocals are centered, but low vocal and sibilance are still off.
So, I checked the audio wave file.
I had a swept sine file, looked at it. I found a timing asymmetry in the waves. Very hard to spot, but it's there. And it looks to be very close to what is expected when the conversion is mux'd.
Cheers, John
poobah said:
UH OH....ground loop...
It was very interesting to me..I kinda geek out at these kinda problems..
It'd be great if anyone else could repeat this stuff themselves, as I'm feelin kinda insane..
Luckily for me, it's only a darn comp sound card in my office..I chill at lunch, listen to various music like inna gadda da vida..so really, the source material ain't exactly um, audiophile quality. No biggie.. but veddy interrrresting...
Cheers, John
instead.....easy enough to test your soundcard, create .wav with 2 sines at different freq in r,l chan respectively; add dac r.l out with equal value resistors, and put into single adc line input
then a dft of the captured sum will show the phase diff between say 1 KHz in right chan vs 2KHz in left
LtSpice does wave file i/o, I think RMAA will play your test wav and capture for you - you should be able to make a long enough stimulus file that you could manualy initate a readwave while it plays using LtSpice alone
this sim shows you could read off 7.8 degrees phase diff @2KHz (diff from 1 KHz ref) which calculates nicely out to 11 US delay with plenty of resolution:
then a dft of the captured sum will show the phase diff between say 1 KHz in right chan vs 2KHz in left
LtSpice does wave file i/o, I think RMAA will play your test wav and capture for you - you should be able to make a long enough stimulus file that you could manualy initate a readwave while it plays using LtSpice alone
this sim shows you could read off 7.8 degrees phase diff @2KHz (diff from 1 KHz ref) which calculates nicely out to 11 US delay with plenty of resolution:
Attachments
jcx said:
The reconstruction algorithms in my sound card are old generation stuff..coupla years old...ancient...
When I play back a swept sine, I can hear the intermodulation products resulting from a "not too deep in Z" reconstruction algorithm..
Your test seems good, but my sound card is not worth the effort, I'd rather listen to James Brown, "get up off a that thing" while I type this..
I gave the information as a little background into how interchannel delay can manifest. There's a whole lot more people here better than I who can take that further...I'm chillin and boppin ta da beat.... OOOOOWWWWWWW....UUUGGGGHHHHH..HOOOO..Somebody stop me...
Cheers, John
janneman said:
If the cable is a first-order system (it isn't, really, but inductance will be a negligeable problem for the frequencies and impedances involved) then we can assume the delay will be due to high freq rolloff caused by the combination of driving impedance and parallel capacitance. We can further derive the corner frequency by knowing that a first-order filter has 45 degrees of phase shift at the -3dB point, and a spot of math shows us the -3 dB point of 25 kHz.
Let's consider further a preamp with an output impedance of 1 kohm (high for solid-state, kind of low for tubes). A bit more math shows us the cable would need 7000 pf (7 nF) to cause a rolloff at the aforementioned 25 kHz. Most cables have about 30 pf/ft, so only a pathologically bad setup or wire would cause this much phase delay between 500 Hz and 10 kHz.
It should be mentioned localisation works on the same signal presented in different ways to both ears; I haven't heard of localisation working for a frequency in one ear and another for the other side.
Cables also are used symmetrically in stereo rigs: it's not usual to find a 1 foot interconnect on one side and a 200 foot length on the other. In other words, localisation has nothing to do with cables.
Cheers,
Francois.
DSP_Geek said:
While your analysis seems good, it has nothing to do with what I have been talking about. Go back to post #321, that has nothing to do with what you analyzed.
DSP_Geek said:
You misunderstood. Perhaps I did not explain well.
Take a 1khz tone and delay it 100 uSec. It will appear at point "A" in the soundstage.
Take a 12 Khz tone and do the same...where will it seem to be??? At the same point A? Why would one assume that?
Take a 50 hz tone and do the same. Should it also be at point A? Nope.
This is what I mean. Varying the ITD or the IID of the sound does not guarantee the virtual image of all of the frequency components will appear to move in concert. There will be a function between frequency, and total shift of image vs ITD and vs IID.
DSP_Geek said:
Agreed, cables are generally symmetrical....at least in length, and vicariously, by total emf coupled voltage generated....but, perhaps not. Nobody is aware of the relationship of this loop coupling other than I (well, ok, I embellish...let me have some fun😀 ..).. I can run unbalanced signal runs 100 feet with multi-megawatt noisy loads within a quarter mile, HVAC systems in the building, 20 kilowatts of light dimmers, and a quarter million volt van degraf 5 feet away, without noise.. because I understand this goop.
As for your statement localization has nothing to do with cables.. go back and read #321. It is not what you spoke on.
Cheers, John
jneutron said:
Hello John,
OK, I read you. Of course the C and L is there but it is not a factor that would cause the delay that would lead to audible localisation differences. That is a good conclusion, I feel we are making progress here. Let's keep chipping away on all these things until we come, hopefully, to the piece de resistance .
Now if you'll excuse me, I need to study loop and field theory...
Jan Didden
Hi Jan
did a search, came up with some good links. The info is clear enough that even I can understand it.
Faraday's law of induction, applied to a real problem, not that integral or derivative goop.
http://hyperphysics.phy-astr.gsu.edu/hbase/electric/farlaw.html#c1
Lenz's law generic form is also underneath Faraday's. Good, they even state that the effect fights to keep the flux constant.
Good luck with the e/m theory stuff..it's not that hard, but it sometimes it can be very counter-intuitive. Drop me a line if ya has any questions.. I am lucky in that I work with some guys who really understand this, as opposed to me..😉
Ah, piece de resistance...ok..
I have included a picture of the snake I used in a 450 seat venue. This cable has a 100 foot #12 extension cord, a balanced mike cable, and the line level stereo run back to the amplifier on stage, a qsc rmx1450. I used a 120 foot length of balanced mike cable for the stereo audio run in order to get the snake to fit on the spool you see. While the RMX-1450 has balanced input capabilities, the rack I use on the stage has an unbalanced (rca) input equilizer that feeds the 1450. So, I used the blue wire of the mike cable for right, the white wire for left. Yes, it degrades the hf channel separation, but that is not significant for this application, and I really didn't notice it anyway..
But of significance is the fact that even though I have 100 feet of unbalanced feed, there is not a whit of noise pickup, no hum, nuttin, in a very hostile e/m environment. The reason is this: The mixing rack/cd player I have on the balcony uses the extension cord for it's power. In that way, the source ground provided by the extension cord is very close to the line feed back to the amp, so that there is no loop formed that will catch stray magnetic field garbage. If I connect the building ground system at the source rack, to the rack in any way, the hum and noise is unbelievably loud.
Luckily, using the one unbalanced mike cord worked like a charm so I didn't have to run two balanced feeds to the amp. This is an old picture, I have since added a 120 foot length of RG-6 as a video feed to a 39 inch tv on stage, and the spool is difficult to wind the cables onto. Also added a dvd player to the balcony rack, and tied that audio into my main system. Plugged the tv into the same outlet strip the amp and snake are on, and still, no noise out of the system...(unless you count the spongebob video clip as noise
...(don't ask)..
The indicators of ground loop, by severity:
1. Steady state hum.
2. Music power modulated hum.
3. External noise pickup, like sub pops.
4. Alteration of audio phase by induced voltages at input.
My setup eliminated 1-3, the setup is incapable of resolving 4, especially with 450 people laughing and having a good time..that's all that counts..
Cheers, John
did a search, came up with some good links. The info is clear enough that even I can understand it.
Faraday's law of induction, applied to a real problem, not that integral or derivative goop.
http://hyperphysics.phy-astr.gsu.edu/hbase/electric/farlaw.html#c1
Lenz's law generic form is also underneath Faraday's. Good, they even state that the effect fights to keep the flux constant.
Good luck with the e/m theory stuff..it's not that hard, but it sometimes it can be very counter-intuitive. Drop me a line if ya has any questions.. I am lucky in that I work with some guys who really understand this, as opposed to me..😉
Ah, piece de resistance...ok..
I have included a picture of the snake I used in a 450 seat venue. This cable has a 100 foot #12 extension cord, a balanced mike cable, and the line level stereo run back to the amplifier on stage, a qsc rmx1450. I used a 120 foot length of balanced mike cable for the stereo audio run in order to get the snake to fit on the spool you see. While the RMX-1450 has balanced input capabilities, the rack I use on the stage has an unbalanced (rca) input equilizer that feeds the 1450. So, I used the blue wire of the mike cable for right, the white wire for left. Yes, it degrades the hf channel separation, but that is not significant for this application, and I really didn't notice it anyway..
But of significance is the fact that even though I have 100 feet of unbalanced feed, there is not a whit of noise pickup, no hum, nuttin, in a very hostile e/m environment. The reason is this: The mixing rack/cd player I have on the balcony uses the extension cord for it's power. In that way, the source ground provided by the extension cord is very close to the line feed back to the amp, so that there is no loop formed that will catch stray magnetic field garbage. If I connect the building ground system at the source rack, to the rack in any way, the hum and noise is unbelievably loud.
Luckily, using the one unbalanced mike cord worked like a charm so I didn't have to run two balanced feeds to the amp. This is an old picture, I have since added a 120 foot length of RG-6 as a video feed to a 39 inch tv on stage, and the spool is difficult to wind the cables onto. Also added a dvd player to the balcony rack, and tied that audio into my main system. Plugged the tv into the same outlet strip the amp and snake are on, and still, no noise out of the system...(unless you count the spongebob video clip as noise
...(don't ask)..The indicators of ground loop, by severity:
1. Steady state hum.
2. Music power modulated hum.
3. External noise pickup, like sub pops.
4. Alteration of audio phase by induced voltages at input.
My setup eliminated 1-3, the setup is incapable of resolving 4, especially with 450 people laughing and having a good time..that's all that counts..
Cheers, John
Attachments
Get sick for a couple of days and miss lots of fun stuff 🙂
jneutron. I skimmed some of the previous posts but I might have a contribution about relative delay and headphones on computers. Cheap sound cards use one DAC, better ones use two (or more). One DAC always causes a time delay between channels because it is multiplexed and by necessity decodes only one channel at a time. So depending on the audio file and the sample rate used there will be an easily measurable time offset between two stereo channels.
jneutron. I skimmed some of the previous posts but I might have a contribution about relative delay and headphones on computers. Cheap sound cards use one DAC, better ones use two (or more). One DAC always causes a time delay between channels because it is multiplexed and by necessity decodes only one channel at a time. So depending on the audio file and the sample rate used there will be an easily measurable time offset between two stereo channels.
Hey, Ya snooze, ya lose...hermanv said:
😉 You need a wireless laptop to bring to bed wit ya..😀
hermanv said:
That is what I had suspected with my sound card...viewing the wave files seems to have confirmed it.
In further playing with my headphones, I found the following:
A vocal which is supposed to be centered and isn't, is just slightly to the right of center. By lifting the right piece away from my ear in an attempt to change the delay, the sound gets weirder. It seems like the position is kinda anchored by the delay, and the level difference caused by moving the set away from the ear is not as strong at moving the image for the main vocal. The gratey, harsh hf part of the male vocal seems to move to the left faster than the main vocal content, it's weird.
Not a very good test, I admit, but fun to try. I am sure that by moving the cup away from the ear, the amplitude vs freq gets all messed up...bass is the first to go.
From all this, it seems like the ITD seems to play a stronger role in image position than the IID. And that to move the image to the left so it is centered, I can sense that the levels are not correct. And, it is, when moving the phones in this fashion, all the frequencies do not move equally.
At least, in these unscientifically controlled playings. And, of course, with the song "hair", by the cowsills..(and again, don't ask...it relates to the van de graf..)
Fun stuff..
Cheers, John
Imaging - volume vs delay.
No much to do with cables but an interesting diversion:
On another forum (no names, no names!) there was a huge debate (tall flames) about sound staging, myself and some others adviced a newbie to get records recorded in a true accoustic format, like a Jazz group maybe, to set up his speaker placement.
An ex-recording engineer took exception, claiming that modern mixes where the sound stage is created at the console was as good and maybe better for setting up his speakers.
I certainly understand how a pan pot works, moving the apparent right to left position of a sound source but admit to no knowledge about mixing consoles and variable delay lines. Like you, I think that just right/left volume can only place an image on a flat two dimensional soundscape but to have realistic depth the time dimension is necessary , additonally it would seem that a time delay right to left would add a realism that only volume can not create. I don't know if consoles can do this.
Anyone out there with artificial imaging knowledge?
jneutron said:
No much to do with cables but an interesting diversion:
On another forum (no names, no names!) there was a huge debate (tall flames) about sound staging, myself and some others adviced a newbie to get records recorded in a true accoustic format, like a Jazz group maybe, to set up his speaker placement.
An ex-recording engineer took exception, claiming that modern mixes where the sound stage is created at the console was as good and maybe better for setting up his speakers.
I certainly understand how a pan pot works, moving the apparent right to left position of a sound source but admit to no knowledge about mixing consoles and variable delay lines. Like you, I think that just right/left volume can only place an image on a flat two dimensional soundscape but to have realistic depth the time dimension is necessary , additonally it would seem that a time delay right to left would add a realism that only volume can not create. I don't know if consoles can do this.
Anyone out there with artificial imaging knowledge?
Re: Imaging - volume vs delay.
At least I know that I am still waiting to hear one that doesn't sound artificial.
hermanv said:
At least I know that I am still waiting to hear one that doesn't sound artificial.
Perhaps we should remain on focus.
How about keeping this thread as strictly IC's, and a new thread started on localization parameters. This can remain set on measurement of effects, as a measurement can be related to the input via percentage distortion, without worrying about how much it will affect localization parameters.
As of this time, we've gotten to the point where testing of the interchassis voltage between a source component and a poweramp running into a good load means something. The audio feed signal for the power amp has to be a battery operated one, so that no loops are formed.
Has anybody taken the time to try this experiment? It'd be nice to see some results. Don't forget, it should be a differential scope measurement to eliminate that ground loop problem.
What I'd look for is residual voltages generated by the haversine component of the line cord currents feeding the full wave bridge supply, residual voltages generated by the audio currents making their way back through the supply, and the switching noise generated by the supply rectifiers.
External influence, motors, lights, should be 60 hz. Haversines line draw will be odd harmonics, supply ripple feeding back into the line will be even harmonics, diode switching will be spikes every 8.3 mSec, and audio feeding back will be modulation of the odd order haversine components, as they typically get back only when the diodes are conducting.
All this is actually the derivative of the currents, not the currents directly.
The first thing I would do, is use an IEC line cord that is made with parallel wires, not twisted. This will create the largest coupling coeff with respect to the ground loop.
Anyone game?
They just poured my basement floor day before yesterday, I can't get to anything.
Cheers, John
How about keeping this thread as strictly IC's, and a new thread started on localization parameters. This can remain set on measurement of effects, as a measurement can be related to the input via percentage distortion, without worrying about how much it will affect localization parameters.
As of this time, we've gotten to the point where testing of the interchassis voltage between a source component and a poweramp running into a good load means something. The audio feed signal for the power amp has to be a battery operated one, so that no loops are formed.
Has anybody taken the time to try this experiment? It'd be nice to see some results. Don't forget, it should be a differential scope measurement to eliminate that ground loop problem.
What I'd look for is residual voltages generated by the haversine component of the line cord currents feeding the full wave bridge supply, residual voltages generated by the audio currents making their way back through the supply, and the switching noise generated by the supply rectifiers.
External influence, motors, lights, should be 60 hz. Haversines line draw will be odd harmonics, supply ripple feeding back into the line will be even harmonics, diode switching will be spikes every 8.3 mSec, and audio feeding back will be modulation of the odd order haversine components, as they typically get back only when the diodes are conducting.
All this is actually the derivative of the currents, not the currents directly.
The first thing I would do, is use an IEC line cord that is made with parallel wires, not twisted. This will create the largest coupling coeff with respect to the ground loop.
Anyone game?
They just poured my basement floor day before yesterday, I can't get to anything.
Cheers, John
Hi John,
You have an excellent idea about starting another thread on localization. It will help keep this thread on track.
As for your basement, you could leave hand prints. Foot prints? Did you bury any wiring you needed?
-Chris
You have an excellent idea about starting another thread on localization. It will help keep this thread on track.
As for your basement, you could leave hand prints. Foot prints? Did you bury any wiring you needed?
-Chris
John,
I think testing IC's can be done in separate steps. No external field influence would be one. Influence of adjacent conductive objects (use a pipe?) and then field rejection tests.
This may help quantify different aspects of the IC's under test. I don't think it can be boiled down to a best design without looking at how these factors may interact.
-Chris
I think testing IC's can be done in separate steps. No external field influence would be one. Influence of adjacent conductive objects (use a pipe?) and then field rejection tests.
This may help quantify different aspects of the IC's under test. I don't think it can be boiled down to a best design without looking at how these factors may interact.
-Chris
- Status
- Not open for further replies.