Announcement

**WolfeMacleod** · 06-17-2011, 07:47 PM

Scissors? To cut wire?
I hate to be critical, but....

**imaradiostar** · 06-18-2011, 04:59 AM

They're only a few percent different. You'd have to wind a large number of each (technically at least 7) to obtain a true statistical sampling.

The real question: How do they sound?

jamie

**StarryNight** · 06-18-2011, 06:31 AM

great post! so perhaps a question would be what is the expected range of reasonant peak to be with all things being equal execept the winding method and look at the percentage of difference between the two. For example if the range might be between 8k and 9k then the difference of 238hz is almost a 24% difference. If the range would only fall between a 500hz range, then the difference in RP between the two methods would almost be 50%.

**freefrog** · 06-18-2011, 07:31 AM

My 2 c of hobbyist:

if the two coils have the same inductance (let's say 1.8 H, that I've typically measured with 5000 turns of wire), the hand made coil has a lower stray capacitance (10pf lower, precisely).

At least it's the result that I find when I use the typical "resonant frequency" formula : Rz = 1 / [square root of (inductance X capacitance)] x 2 x pi.

In this case, if the inductance is 1.8H, the stray capacitance of the hand made coil + other wires and testing gear = roughly 195 pf and it's roughly 205 pf with the cnc coil. That explains the location of the resonant peaks (and it's not surprising since hand wiring is supposed to lower stray capacitance).

I don't understand why the frequency response rises again above 15 khz with the cnc coil, though. Dave Stephens might have something to say about it. :-)

**Joe Gwinn** · 06-18-2011, 03:02 PM

Both coils have 5,000 turns and are identical in all save how they were wound, which raises the question of why the resonant peaks differ, the CNC peak being 8,248 Hz while the hand-wound peak being at 8,486 Hz, this being 238 Hz higher, which is 2.9%. The CNC coil is tightly wound, while the hand-wound coil is scatter-wound to a fair degree.

There are two things that will contribute to the difference in resonant frequency:

The self-capacitance of a random-wound coil is lower than that of a tightly-wound coil.

The inductance will also differ, even though the number of turns is the same, as the physical size and shape of the winding of an air-core inductor will differ slightly: the random-wound coil is slightly larger than the tightly-wound coil.

The DC resistances also differ, the tightly-wound (CNC) coil being 3,748 Ohms, while the scatter-wound coil is 3,726 Ohms; their ratio is 1.0059 to 1, a 0.6% difference, which is well within the error band of handheld ohmmeters. While scatter-wound coils use more wire for the same number of turns as for a tightly-wound coil, machine winders tend to stretch the wire more, and it seems that these two effects largely cancelled.

What would be useful at this point would be direct measurement of the inductances of the two coils at 1,000 Hz, to assess the relative importance of these effects.

**Mike Sulzer** · 06-18-2011, 04:58 PM

Originally posted by Joe Gwinn View Post

B
The inductance will also differ, even though the number of turns is the same, as the physical size and shape of the winding of an air-core inductor will differ slightly: the random-wound coil is slightly larger than the tightly-wound coil.

I agree that they are air core; the view at the end shows that well. This brings up the question of why the resonance is so low in frequency. I would expect above 10,000 Hz for a 5,000 turn humbucker coil with no cores. But I do not understand the set up. Why not just the coil under test and a driver coil? Why that bundle of coils? Two coils in series should have the same resonant frequency as one alone, and so that should not be a problem, but maybe there is some kind of excessive coupling going on?

It is important to do both kinds of measurements, a driver coil, and an impedance measurement and make sure that the results are consistent.

**marku52** · 06-18-2011, 05:08 PM

One potential error source with the measurement is the scope probe that is providing the 10Mohm input Z. These probes are "compensated" voltage dividers, typically consisting of a 9Mohm input series R with a small shunting capacitor working into the 1Mohm to ground resistance provided by the input Z of the measuring device, usually the input of a scope. There is also a way to "compensate" the voltage divider, since to have a flat frequency response, the RC time constant of the series element needs to be equal to the RC time constant of the shunt element. And the C of the input element will vary from scope to scope. The fix for this is a variable cap in one of the 2 divider elements, so that time constant of the 2 RCs can be equalized. this is usually a trimmer cap in the "comp box" of the probe, tho the one in the video looks like an older type with the trim cap in the probe head.

If this compensation is incorrect, then the probe/scope combo will have a bump or dip in the frequency response, usually in the audio region. The 1kHz square wave response will be correspondingly non-flat as well. Perhaps the compensation trim cap got bumped between measurements. Also, because of the extremely high impedances, odd things like moisture on or in a circuit board in the probe can disrupt this compensation across frequency.

If this were this my test jig I'd get rid of the scope probe-- really, 1Mohm input Z is plenty enough, and you don't need the potential error of the probe not being rigorously calibrated between measurements, and the typical scope 1Mohm has the advantage of equalling the input R of most amps.

Also, I'd run a flatness test of the test setup itself by temporarily replacing the driving coil with a low resistance, 1K or whatever suits the output of the current amp to provide an adequate signal level to the computer/scope. The response of this simple system had better be flat across the frequencies of interest, which is to about 20kHz here

**David Schwab** · 06-18-2011, 10:02 PM

I think if you want to compare hand wound to machine wound, you shouldn't scatter the hand wound coil. I don't use any scatter at all when I hand wind now. I try and get it as neat and even as possible.

**MrCandy** · 06-20-2011, 02:43 PM

Originally posted by WolfeMacleod View Post

Scissors? To cut wire?
I hate to be critical, but....

For 42 AWG it's best to cut with scissors.PCB soldering uses special liquid fluxes that you can see on the attched screenshot.
For drawoffs from the coil we use silver-plated wire in teflon insulation. If you ever hold this kind of wire in your hands,
then you have to understand what we are talking about. The winding wire itself is not oxidized because it is isolated, thats why it is soldered with a conventional rosin.

Attached Files

**Mike Sulzer** · 06-20-2011, 05:53 PM

Originally posted by MrCandy View Post

For 42 AWG it's best to cut with scissors.PCB soldering uses special liquid fluxes that you can see on the attched screenshot.
For drawoffs from the coil we use silver-plated wire in teflon insulation. If you ever hold this kind of wire in your hands,
then you have to understand what we are talking about. The winding wire itself is not oxidized because it is isolated, thats why it is soldered with a conventional rosin.

Yes, scissors are very nice for cutting small wire. I very much like teflon insulated silver wire for this purpose, although it is a bit expensive. Thanks for the explanation of the soldering. I thought it would be something like that.

**David Schwab** · 06-20-2011, 06:13 PM

I use a small pair of surgical scissors. However, after I solder the end lead on I just pull off the wire.

Seems silly cutting it.

**jason lollar** · 06-20-2011, 06:53 PM

david has a good point its not a definitive test- that is a "modern" winding machine very accurate and programmable- what kind of pitch did you use and how many trurns per layer does that equate to? Did you try any other pitch settings? That would be interesting. My experience is turns per layer has a very noticeable effect. Also that type of machine works very differently than the old mechanical traverse types so any results would only apply to that type of machine. One could set up the machine and the handwound process to either be as different as possible or as similar as possible- I think it would be more useful to set them up as similar as possible and see what difference there is.
" 3,748 Ohms, while the scatter-wound coil is 3,726" i wouldnt trust those readings to have anything to do with the winding- thats so small it could be wire variance.

**MrCandy** · 06-20-2011, 09:47 PM

Summing all the questions we would like to respond maximally with 1 post:
1) To compare the sound of these two coils makes no sense, will sound almost identical.
The only thing hand-wound by contrast will sound more sharp then CNC, as seen from the spectrogram. With respect to a note by David the coil is not scattered, wound qualitatively, providing the same number of turns in one layer as CNC.
About their sound: the coils we used had a bad color and a little bit bad spectrum,
they were made exclusively for testing and exclusively for resonance peak to be exactly in the middle of scale for clarity

After the tests the samples were successfully sent to the trash

2) About the measurement procedure below you can see some pictures with other method of measuring. Both of the methods of of measuring which we have provided you successfully used by us for over 17 years. Even in those days, when there was no computer gadget programs and measurements were only special devices using these two methods of measuring we obtained annalogichny result.On one of the pictures attached you can see a spectrogram of Lace Sensor Red we made more than ten years ago.
As for measurement techniques using a probe Rin=10 mOm - this is the only true test method for measuring the resonant peak. Even better to use AD 624 that would minimize the drawdown in the measurement at the beginning of the spectrum, although this part of the curve a little informative to pickup sound modeling on the spectrum analyzer.
Using Rin=1 mOm is erroneously because this gives a reduction of the spectrum акщь 1 ещ 2.5 kHz that would not be a true value. There are a number of scientific papers on this topic in which it shows the true values of the resonance peaks, that do not always coincide with the stated, that should cause a number of issues.
3) As for the drive coil used by us its DCR=6 Om and has 13 turns in each coil. Measurement directly at the amplifier output with connected coil and shows a satisfactory linearity. Because we use not traditional type booster, it measures with current sensor current flowing in the coil and it provides a constant. Voltage measurement (sweep) акщь 250mV - 2V not reflected in the testimony of the spectrogram.

With respect
MrCandy

Attached Files