John Fields wrote in message . ..
On 14 Feb 2004 16:07:44 -0800, (Svante)
wrote:

John Fields wrote in message . ..
On 13 Feb 2004 17:36:14 -0800, (Svante)
wrote:

---
Precisely, and that's what would account for the broadening of the peak
at resonance, which is what I thought you were talking about.

So what's left?

Perhaps the output impedance of your signal generator is varying?

If possible, how about a schematic of your setup?

As you may have noticed in my other posts I have reralised that my
calculations of the impedance SHOULD not work for anything but
resistors. :-( Silly me. I am working on fixing it, and also to
compensate for the left-right delay of my soundcard.

I posted the schematic previously

Lout ----------
|
Lin ----------*
|
Rs=47 ohm
|
Rin ----------*
|
Z
|
GND ----------

---
That's not really what I meant.

What I'm trying to get at is what your test equipment looks like.

Do you have a signal generator?

Ok, so the signal generator is my computer, and of course that has an
output impedance.

If you do, what is its output impedance and how flat is its output with
frequency when you are using it to make measurements?

Don't know, but I compensate for it by measuring "Lin".

I noticed from your photos that you were coupling direcly into the
vertical inputs of your oscilloscope. Do you know what the impedance of
those inputs looks like?

That was a different setup, but it says on the front, "1 Mohm ~22pF"

---

and this should work regardlessly of the output impedance of the
generator, since I *measure* the voltage fed to the circuit by Lin. If
I had only done the calculations correctly with complex math, and
compensated for the ~1/4 sample delay between Lin and Rin channels.

---
Since you haven't described _how_ you make your measurements, that's
confusing in that it could mean a 1/4 sample time delay if you have a
fixed time between samples or, presupposing a priori knowledge of the
frequency, for test, 90° of that. However, since you say:
---

I
feed the network a complex signal and FFT the two inputs, and get a
magnitude and phase for each frequency. I used the magnitudes as I
posted and thought I would get the impedance. Of course I need to
apply the (compensated) phase as well.

---
Indeed. You would be well advised to use your 'old scope' to make phase
measurements in real time and use those results to determine the
impedances in question.

I will try the "old scope" method at work where I have a decent signal
generator, I don't have that at home. Meanwhile I try to fix the
computer method too, and if I manage to get a good match between model
and measurement by doing the math right and introducing a small delay,
I am satisfied.

To make the test setup clear:

SBLive! Left out ----------
|
SBLive! Left in ----------*
|
Rs=47 ohm
|
SBLive! Right in ----------*
|
Z
|
SBLive! GND ----------

Each of the in/out connections are made with a cheap RCA-3.5mm cable,
possibly adding some 100pF of parallel capacitance. The impedance of
100 pF at 20 kHz is about 80 kohm, so that should not be a problem at
these impedance levels and frequencies.

Certainliy there is an output impedance from the SBLive!, but my idea
is to compensate for that by actually measuring the voltage with the
left input. This circuit would also compensate for amplitude and phase
response issues of the sound card as long as the two channels of the
soundcard are identical and the amplitude is reasonably much greater
than zero. It turns out that the "phase" condition does not hold for
higher frequencies due to a hard wired delay between the Lin and Rin
channels in the soundcard, so I'll have to calibrate for that.

The second setup was with the oscilloscope, and you saw photos of
that.

I am too *a little* surprised about the delay between channels.
However I guess that is what you get when you buy a $20 soundcard
(SBLive!), I suppose it means that there is in fact only *one* A/D
converter and that it multiplexes the two channels.

Thanks again for your help.

So... Finally I have understood what happened. The cause was no
mysterious skin effect, neither was it a parallel capacitance. It was
simply bad programming.

As you may have seen in my other posts I was on the track of it
earlier, thanks to John Fields and chung who pointed out that my
program of course need to take phase into account as it calculates the
impedance. Silly me.

Anyway, once I had the math right, that was not enough. It turned out
that there was a delay between left and right channels of my soundcard
of about 17 us, or 0.818 samples at 48kHz sampling frequency. This was
perfectly enough to ruin measurements above, say 5 kHz. After
compensating for the delay too, I end up with these curves, a nearly
perfect fit with a model where Z=0.45ohm + jw*0.21mH.

http://www.tolvan.com/coil3.gif

Even though it would have been exciting finding that it was the skin
effect, I am relieved that it was just me being stupid, and that the
old fashioned model of the coil works so well.

Thank you all for helping me out with this. I just wish I had started
a separate thread for it.

/Svante

So... Finally I have understood what happened. The cause was no
mysterious skin effect, neither was it a parallel capacitance. It was
simply bad programming.

As you may have seen in my other posts I was on the track of it
earlier, thanks to John Fields and chung who pointed out that my
program of course need to take phase into account as it calculates the
impedance. Silly me.

Anyway, once I had the math right, that was not enough. It turned out
that there was a delay between left and right channels of my soundcard
of about 17 us, or 0.818 samples at 48kHz sampling frequency. This was
perfectly enough to ruin measurements above, say 5 kHz. After
compensating for the delay too, I end up with these curves, a nearly
perfect fit with a model where Z=0.45ohm + jw*0.21mH.

http://www.tolvan.com/coil3.gif

Even though it would have been exciting finding that it was the skin
effect, I am relieved that it was just me being stupid, and that the
old fashioned model of the coil works so well.

Thank you all for helping me out with this. I just wish I had started
a separate thread for it.

/Svante

So... Finally I have understood what happened. The cause was no
mysterious skin effect, neither was it a parallel capacitance. It was
simply bad programming.

As you may have seen in my other posts I was on the track of it
earlier, thanks to John Fields and chung who pointed out that my
program of course need to take phase into account as it calculates the
impedance. Silly me.

Anyway, once I had the math right, that was not enough. It turned out
that there was a delay between left and right channels of my soundcard
of about 17 us, or 0.818 samples at 48kHz sampling frequency. This was
perfectly enough to ruin measurements above, say 5 kHz. After
compensating for the delay too, I end up with these curves, a nearly
perfect fit with a model where Z=0.45ohm + jw*0.21mH.

http://www.tolvan.com/coil3.gif

Even though it would have been exciting finding that it was the skin
effect, I am relieved that it was just me being stupid, and that the
old fashioned model of the coil works so well.

Thank you all for helping me out with this. I just wish I had started
a separate thread for it.

/Svante

So... Finally I have understood what happened. The cause was no
mysterious skin effect, neither was it a parallel capacitance. It was
simply bad programming.

As you may have seen in my other posts I was on the track of it
earlier, thanks to John Fields and chung who pointed out that my
program of course need to take phase into account as it calculates the
impedance. Silly me.

Anyway, once I had the math right, that was not enough. It turned out
that there was a delay between left and right channels of my soundcard
of about 17 us, or 0.818 samples at 48kHz sampling frequency. This was
perfectly enough to ruin measurements above, say 5 kHz. After
compensating for the delay too, I end up with these curves, a nearly
perfect fit with a model where Z=0.45ohm + jw*0.21mH.

http://www.tolvan.com/coil3.gif

Even though it would have been exciting finding that it was the skin
effect, I am relieved that it was just me being stupid, and that the
old fashioned model of the coil works so well.

Thank you all for helping me out with this. I just wish I had started
a separate thread for it.

/Svante

On 14 Feb 2004 07:40:07 -0800, (Svante)
wrote:

(Goofball_star_dot_etal) wrote in message ...
Here are some numbers that came out for copper 0.9mm:
Freq Multiply resistance by.
10 kHz 1.24
20 kHz 1.35
100 kHz. 1.84
1 MHz.. 4.04
10 MHz... 11.5
100MHz:-) 35
1GHz :--)) 109
Probably ********!

Great! So if the original 0.7 ohm resistance turns into 0.945 ohms at
20 kHz it would be no way near the effect I saw. It feels comforting
to know that skin is for the radio guys.

The golden ears are not going to be happy with the losses at 100 kHz
in 1cm dia. wire.

You were right in the first place: 'John's dad say: "Never trust
measurement that give wrong answer."'

One does not talk abot one's psychic powers. . .

As you may have seen in a
parallel post of mine the program does silly things, probably due to a
silly programmer. It is a miracle that it worked at all. I'll do my
homework and get back.

At least you avoided becoming a Tweako-freako(tm).

On 14 Feb 2004 07:40:07 -0800, (Svante)
wrote:

(Goofball_star_dot_etal) wrote in message ...
Here are some numbers that came out for copper 0.9mm:
Freq Multiply resistance by.
10 kHz 1.24
20 kHz 1.35
100 kHz. 1.84
1 MHz.. 4.04
10 MHz... 11.5
100MHz:-) 35
1GHz :--)) 109
Probably ********!

Great! So if the original 0.7 ohm resistance turns into 0.945 ohms at
20 kHz it would be no way near the effect I saw. It feels comforting
to know that skin is for the radio guys.

The golden ears are not going to be happy with the losses at 100 kHz
in 1cm dia. wire.

You were right in the first place: 'John's dad say: "Never trust
measurement that give wrong answer."'

One does not talk abot one's psychic powers. . .

As you may have seen in a
parallel post of mine the program does silly things, probably due to a
silly programmer. It is a miracle that it worked at all. I'll do my
homework and get back.

At least you avoided becoming a Tweako-freako(tm).

On 14 Feb 2004 07:40:07 -0800, (Svante)
wrote:

(Goofball_star_dot_etal) wrote in message ...
Here are some numbers that came out for copper 0.9mm:
Freq Multiply resistance by.
10 kHz 1.24
20 kHz 1.35
100 kHz. 1.84
1 MHz.. 4.04
10 MHz... 11.5
100MHz:-) 35
1GHz :--)) 109
Probably ********!

Great! So if the original 0.7 ohm resistance turns into 0.945 ohms at
20 kHz it would be no way near the effect I saw. It feels comforting
to know that skin is for the radio guys.

The golden ears are not going to be happy with the losses at 100 kHz
in 1cm dia. wire.

You were right in the first place: 'John's dad say: "Never trust
measurement that give wrong answer."'

One does not talk abot one's psychic powers. . .

As you may have seen in a
parallel post of mine the program does silly things, probably due to a
silly programmer. It is a miracle that it worked at all. I'll do my
homework and get back.

At least you avoided becoming a Tweako-freako(tm).

On 14 Feb 2004 07:40:07 -0800, (Svante)
wrote:

(Goofball_star_dot_etal) wrote in message ...
Here are some numbers that came out for copper 0.9mm:
Freq Multiply resistance by.
10 kHz 1.24
20 kHz 1.35
100 kHz. 1.84
1 MHz.. 4.04
10 MHz... 11.5
100MHz:-) 35
1GHz :--)) 109
Probably ********!

Great! So if the original 0.7 ohm resistance turns into 0.945 ohms at
20 kHz it would be no way near the effect I saw. It feels comforting
to know that skin is for the radio guys.

The golden ears are not going to be happy with the losses at 100 kHz
in 1cm dia. wire.

You were right in the first place: 'John's dad say: "Never trust
measurement that give wrong answer."'

One does not talk abot one's psychic powers. . .

As you may have seen in a
parallel post of mine the program does silly things, probably due to a
silly programmer. It is a miracle that it worked at all. I'll do my
homework and get back.

At least you avoided becoming a Tweako-freako(tm).

Goofball_star_dot_etal wrote:

On 14 Feb 2004 07:40:07 -0800, (Svante)
wrote:


(Goofball_star_dot_etal) wrote in message ...

Here are some numbers that came out for copper 0.9mm:
Freq Multiply resistance by.
10 kHz 1.24
20 kHz 1.35
100 kHz. 1.84
1 MHz.. 4.04
10 MHz... 11.5
100MHz:-) 35
1GHz :--)) 109
Probably ********!

Great! So if the original 0.7 ohm resistance turns into 0.945 ohms at
20 kHz it would be no way near the effect I saw. It feels comforting
to know that skin is for the radio guys.


The golden ears are not going to be happy with the losses at 100 kHz
in 1cm dia. wire.


I think you mean 1 mm.


You were right in the first place: 'John's dad say: "Never trust
measurement that give wrong answer."'


One does not talk abot one's psychic powers. . .


As you may have seen in a
parallel post of mine the program does silly things, probably due to a
silly programmer. It is a miracle that it worked at all. I'll do my
homework and get back.


At least you avoided becoming a Tweako-freako(tm).



--
After being targeted with gigabytes of trash by the "SWEN" worm, I have
concluded we must conceal our e-mail address. Our true address is the
mirror image of what you see before the "@" symbol. It's a shame such
steps are necessary. ...Charlie

Goofball_star_dot_etal wrote:

On 14 Feb 2004 07:40:07 -0800, (Svante)
wrote:


(Goofball_star_dot_etal) wrote in message ...

Here are some numbers that came out for copper 0.9mm:
Freq Multiply resistance by.
10 kHz 1.24
20 kHz 1.35
100 kHz. 1.84
1 MHz.. 4.04
10 MHz... 11.5
100MHz:-) 35
1GHz :--)) 109
Probably ********!

Great! So if the original 0.7 ohm resistance turns into 0.945 ohms at
20 kHz it would be no way near the effect I saw. It feels comforting
to know that skin is for the radio guys.


The golden ears are not going to be happy with the losses at 100 kHz
in 1cm dia. wire.


I think you mean 1 mm.


You were right in the first place: 'John's dad say: "Never trust
measurement that give wrong answer."'


One does not talk abot one's psychic powers. . .


As you may have seen in a
parallel post of mine the program does silly things, probably due to a
silly programmer. It is a miracle that it worked at all. I'll do my
homework and get back.


At least you avoided becoming a Tweako-freako(tm).



--
After being targeted with gigabytes of trash by the "SWEN" worm, I have
concluded we must conceal our e-mail address. Our true address is the
mirror image of what you see before the "@" symbol. It's a shame such
steps are necessary. ...Charlie

Goofball_star_dot_etal wrote:

On 14 Feb 2004 07:40:07 -0800, (Svante)
wrote:


(Goofball_star_dot_etal) wrote in message ...

Here are some numbers that came out for copper 0.9mm:
Freq Multiply resistance by.
10 kHz 1.24
20 kHz 1.35
100 kHz. 1.84
1 MHz.. 4.04
10 MHz... 11.5
100MHz:-) 35
1GHz :--)) 109
Probably ********!

Great! So if the original 0.7 ohm resistance turns into 0.945 ohms at
20 kHz it would be no way near the effect I saw. It feels comforting
to know that skin is for the radio guys.


The golden ears are not going to be happy with the losses at 100 kHz
in 1cm dia. wire.


I think you mean 1 mm.


You were right in the first place: 'John's dad say: "Never trust
measurement that give wrong answer."'


One does not talk abot one's psychic powers. . .


As you may have seen in a
parallel post of mine the program does silly things, probably due to a
silly programmer. It is a miracle that it worked at all. I'll do my
homework and get back.


At least you avoided becoming a Tweako-freako(tm).



--
After being targeted with gigabytes of trash by the "SWEN" worm, I have
concluded we must conceal our e-mail address. Our true address is the
mirror image of what you see before the "@" symbol. It's a shame such
steps are necessary. ...Charlie

Goofball_star_dot_etal wrote:

On 14 Feb 2004 07:40:07 -0800, (Svante)
wrote:


(Goofball_star_dot_etal) wrote in message ...

Here are some numbers that came out for copper 0.9mm:
Freq Multiply resistance by.
10 kHz 1.24
20 kHz 1.35
100 kHz. 1.84
1 MHz.. 4.04
10 MHz... 11.5
100MHz:-) 35
1GHz :--)) 109
Probably ********!

Great! So if the original 0.7 ohm resistance turns into 0.945 ohms at
20 kHz it would be no way near the effect I saw. It feels comforting
to know that skin is for the radio guys.


The golden ears are not going to be happy with the losses at 100 kHz
in 1cm dia. wire.


I think you mean 1 mm.


You were right in the first place: 'John's dad say: "Never trust
measurement that give wrong answer."'


One does not talk abot one's psychic powers. . .


As you may have seen in a
parallel post of mine the program does silly things, probably due to a
silly programmer. It is a miracle that it worked at all. I'll do my
homework and get back.


At least you avoided becoming a Tweako-freako(tm).



--
After being targeted with gigabytes of trash by the "SWEN" worm, I have
concluded we must conceal our e-mail address. Our true address is the
mirror image of what you see before the "@" symbol. It's a shame such
steps are necessary. ...Charlie

On Sun, 15 Feb 2004 22:42:32 GMT, CJT wrote:


The golden ears are not going to be happy with the losses at 100 kHz
in 1cm dia. wire.


I think you mean 1 mm.


Heck, I don't know what they use but 10mm gives a facor of 13 which is
bad luck.

On Sun, 15 Feb 2004 22:42:32 GMT, CJT wrote:


The golden ears are not going to be happy with the losses at 100 kHz
in 1cm dia. wire.


I think you mean 1 mm.


Heck, I don't know what they use but 10mm gives a facor of 13 which is
bad luck.

On Sun, 15 Feb 2004 22:42:32 GMT, CJT wrote:


The golden ears are not going to be happy with the losses at 100 kHz
in 1cm dia. wire.


I think you mean 1 mm.


Heck, I don't know what they use but 10mm gives a facor of 13 which is
bad luck.

On Sun, 15 Feb 2004 22:42:32 GMT, CJT wrote:


The golden ears are not going to be happy with the losses at 100 kHz
in 1cm dia. wire.


I think you mean 1 mm.


Heck, I don't know what they use but 10mm gives a facor of 13 which is
bad luck.