开云体育

Hi all,
On another group there is a discussion on Q meters, then we went to the loss the air cap added to the loss of the inductor during measurement.
So all we know is the combined loss not the individual losses.
This article purports to measure with a VNA capacitor losses.


But,
1) I can't see how the author separates the inductor and capacitor losses, as he does use a coil to resonate.
2) I want to test an air capacitor, at several points from 20pf to 400pf over a frequency range of 500kHz to 1700kHz, probable requiring a 240uh inductor.
That 240uh inductor will have about 1.5Ω of loss and more over that frequency. I don't see measuring 0.1Ω and 0.2Ω capacitor losses if the coil losses aren't separated.
Can you get my thinking straightened out?

Mikek

On Sun, Oct 9, 2022 at 12:54 PM, Mikek wrote:

2) I want to test an air capacitor, at several points from 20pf to 400pf over
a frequency range of 500kHz to 1700kHz, probable requiring a 240uh inductor.
That 240uh inductor will have about 1.5Ω of loss and more over that
frequency.

What type of 240 uH inductor will you be using to get such a low R over that frequency range?

Roger

On Sun, Oct 9, 2022 at 01:05 PM, Roger Need wrote:

What type of 240 uH inductor will you be using to get such a low R over that
frequency range?

Well, it won't be that low over the whole range. But I have coils that reach 1400Q of the top of my head.
I'll post a picture and some graphs of measurements made.
Mikek

If you use a resonant circuit to find the Q the loss with be combination of both the L and C. You can only separate it if you can independently measure the loss of the coil or capacitor without the resonance. For low frequencies you can use an ESR meter to measure the loss if the capacitor. Alternatively I would parallel the capacitor with a series of high value resistors R. Measure the Z. Then extrapolate to zero R to find residual R in Z. This method can be used at any frequency.

On Mon, Oct 10, 2022 at 12:20 AM, tuckvk3cca wrote:

If you use a resonant circuit to find the Q, the loss with be combination of
both the L and C. You can only separate it if you can independently measure
the loss of the coil or capacitor without the resonance.

That is why I have the question, from my quick reading of the paper, it seemed like the author separated them, but I agree with you,
at resonance you can't separate them.

Alternatively I would parallel the capacitor with a series of high value resistors R. Measure the Z. Then extrapolate to zero R to find residual R in Z.

By extrapolate to zero, here is about as far as I understand that concept, my example would be finding the output impedance of an amp by taking two measurements with two different value of resistor and extrapolate to find the output impedance.
I expect values under 0.2Ω and hopefully less if I have any good air caps.

Is there a math expression that you can tell me to do the extrapolation?
I don't have the math skills to figure that out.

If have that, I can take many measurements over frequency, plug everything into Excel and make graphs of my better capacitors.
Thanks, Mikek

My observation/experience with inductors that large (~240uH) at those frequencies is caused by two effects: 1) skin effect 2) proximity effect. The latter being less known about yet a significant contributor to resistive losses in the inductor. I suspect you will find the capacitor resistive losses are *lower* than the inductor’s resistive losses. How did you calculate the 1.5 ohms for the inductor?

-Charlie
W5CDT

On Mon, Oct 10, 2022 at 07:09 AM, Charlie Thompson wrote:

My observation/experience with inductors that large (~240uH) at those
frequencies is caused by two effects: 1) skin effect 2) proximity effect. The
latter being less known about yet a significant contributor to resistive
losses in the inductor. I suspect you will find the capacitor resistive losses
are *lower* than the inductor’s resistive losses. How did you calculate the
1.5 ohms for the inductor?

Yes, the capacitor loses will be much less than the inductor. A very good air cap can have Qs from 8,000 to 20,000.
A really good Q for an inductor is 1400, but some have built BCB inductors near 2000, using 1162/46 litz wire.
Proximity effect was the reason I did the 5 coil experiment to find the optimum spacing for 660/43 litz.
I measured Q on my Boonton 260A* and R(loss) = XL/Q, the 1.5Ω was a guestimate using, XL @1MHz = 1508Ω, Q =1000 R = 1.508Ω.
Real measurements, looking at my graphs, best case, XL @ 800kHz = 1,191Ω, Q = 1440, R = 0.827Ω.
I see R(loss) of 0.577Ω at 500kHz for the 12 Turns per inch coil. The worst case of my 5 coils, has over 3Ω R(loss) at 1700kHz.
I expect Capacitor R(loss) under 0.2Ω.
I still need the math for extrapolation if anyone can help.
* also did 3db Q test to verify.
Thanks, Mikek
P.S. My graphs were an average of 4 measurements at each frequency.

Coil forms contribute dielectric losses. It's fairly common to see coils that use plastic rods or strips rods to keep the windings appropriately spaced while minimizing dielectric losses.

On Mon, Oct 10, 2022 at 07:47 AM, Lou W7HV wrote:

Coil forms contribute dielectric losses. It's fairly common to see coils that
use plastic rods or strips rods to keep the windings appropriately spaced
while minimizing dielectric losses.

Absolutely, the form I showed in the picture uses a styrene form, styrene is high on the list of low loss materials.
I was borrowing someone else's time and lathe work or I would had the form turned for a thinner wall, before cutting the grove for TPI.
I tested 5 wire spacings to make the graph shown in my previous post. I had a very shallow groove cut to get proper, consistent spacing
of the wire for the 5 different spacings tested.
I did build a 3d coil from polystyrene sheet, but have yet to wind it with litz wire.
Here's a picture, sorry, it does not show the 3D layering very well. it is 6 wires wide and 6 layers deep.
Mikek

Hello Mikek,

From my point of view it will be better to use thick silver-plated copper wire for high-Q inductors.

--
Bert W

On Mon, Oct 10, 2022 at 12:53 PM, Bert W wrote:

From my point of view it will be better to use thick silver-plated copper wire
for high-Q inductors.

That is true in some circumstances, it depends on the inductor in question and the frequency.
For a 240uh air core inductor at BCB frequencies, I doubt silver would beat litz, because of skin effect.
In copper at 1MHz, the resistance is 6.7 times the DC resistance. It is similar in silver.
But it all comes down to the diameters. 660/43 litz is about 0.065", if you went with a
silver wire 0.065", because of skin effect, 37% of the current flows in the wires outer 0.026".
This causes the AC resistance to be higher that the DC resistance.
These pages has some explanation.


Mikek

1. Use this program to calculate coil inductance and Q. It models skin effect, proximity effect, and more subtle effects for both solid and Litz wire:



Results agree well with measurements made with an HP 4342A Q meter. Use the program to verify that Q measurements made with a NanoVNA are not grossly off.

2. Silver plating annealed copper increases Q about 4%. This is usually not worth the trouble.

3. Evidently the styrene pipe couplings popular with crystal set builders use PVC Type 1. This is not a particularly low-loss dielectric. See the attached data sheet.

4. If you have a Q meter, it's easy to measure capacitor Q. See the Boonton 260A or HP 4342A manual for instructions. These Q meter utilities contain the instructions and will do the calculations:



5. The 3-D coil is a novel idea. However, the inner windings contribute much less inductance because of their smaller diameter. I get similar Q at 1 MHz by having the coil program automatically optimize the design of a conventional solenoid of #24 wire.

6. Q > 2000 is readily achievable with Litz wire at 1 MHz. The design shown below uses a 9.9" OD styrofoam cake form.

Brian

On Mon, Oct 10, 2022 at 03:25 PM, Brian Beezley wrote:

1. Use this program to calculate coil inductance and Q. It models skin effect,
proximity effect, and more subtle effects for both solid and Litz wire:

I can open the above page but the Coil.Zip keeps say connection reset while loading.

3. Evidently the styrene pipe couplings popular with crystal set builders use
PVC Type 1. This is not a particularly low-loss dielectric. See the attached
data sheet.

I think that would depend on the crystal set builder. I was active on the now *defunct
'The Radioboard' crystal radio section, stryene was discussed , it is well known
in the crystal radio community that PVC is lossy.
You can tell the difference in the feel, styrene is harder than PVC.
It has more ring to it (like bell), PVC has a damped thud. I doubt you would get a
Q of 1400 with PVC. It is much harder to find a source for Styrene couplers now than it was 5 years ago.
I recently went on a search for a friend and found some, I gave him him the site,
he called and they said they don't carry them anymore. So it may be difficult to find them now.
You can also tell styrene from PVC by a burn test, flame and smell.

4. If you have a Q meter, it's easy to measure capacitor Q. See the Boonton
260A or HP 4342A manual for instructions. These Q meter utilities contain the
instructions and will do the calculations:

I'll look it to that, but, are you actually measuring the capacitor Q, and not the capacitor and inductor Q.
I suspect it's both.

6. Q > 2000 is readily achievable with Litz wire at 1 MHz. The design shown
below uses a 9.9" OD styrofoam cake form.

As I said, a couple people did it on the older Rap n Tap crystal radio forum, using double and triple
660/46 litz wire.

Thanks, Mikek

*Sadly, the fairly new administrator got covid, had real issues and didn't pay the
server fee and we lost years of information. It was very active and productive group.

Your antivirus program is probably keeping you from opening the ZIP file. Send the file to to assure yourself that it's virus-free and then disable your antivirus program.

The pipe coupling datasheet has a standard listed for the styrene, but I was unable to find it for free. However, it is listed as a withdrawn standard. That may explain why styrene couplers are now hard to find. I wonder if it like ABS, which has styrene in the name. PVC and ABS have similar loss.

The capacitance measurement procedure uses a coil, but its Q is irrelevant. Read the instructions and you'll see what they're up to.

The example coil with Q > 2000 used a single 660/46 Litz wire. Form wall should have been 4.95", but styrofoam has such low loss that it makes no difference. I ran a model with a 6.67" PVC pipe coupler, the largest size available. Q was something like 1880, which isn't all that bad. Optimum Q occurs near 9.3". The program will find that value automatically if you let it vary coil diameter.

Measuring large coils can be difficult because they couple to everything, including the Q meter enclosure and your body. See README.TXT for a simple way to overcome this problem.

Brian

Excel does allow you to curve fit and then find an extrapolated value. It will provide you with the curve fitted formula.

After much searching, I was able to locate D-2852, the ASTM standard for the styrene pipe coupling. As expected from the application, it does not contain any electrical specs. Here's the description of its composition:

5.1 Materials—The pipe and fittings shall be made of styrene-rubber (SR) plastics meeting the following requirements:

5.1.1 The SR plastics compound shall contain at least 50 % styrene plastics, combined with rubbers to a minimum rubber content of 5 %, and compounding materials such as antioxidants and lubricants, and may contain up to 15 % acrylonitrile combined in the styrene plastics or rubbers, or both. The rubbers shall be of the polybutadiene or butadiene-styrene type, or both, with a maximum styrene content of 25 % or nitrile type. The combined styrene plastics and rubber content shall be not less than 90 %. No fillers may be used.

It sounds like a form of ABS, acrylonitrile butadiene styrene. Here are the dielectric constant and dissipation factor for ABS and PVC, both at 1 MHz:

ABS 3.20 .008
PVC 2.88 .016

The higher dielectric constant of ABS pretty much negates its loss advantage. See the two models for a coil using a 4.5" pipe coupling below.

Brian

Hi Brian,
Those Qs are higher than anything I have made, However, I would like to see a real world case on PVC that has a higher Q than
my so called styrene coupler.
Can you run my coil? Dimensions are, 6.5" in diameter, Coil length is 3.3125", wall thickness 0.138", it is 37 turns,
23" leads (total), (Wire diameter is 0.067", TPI is 11.)
Thanks, Mikek

On 10/11/22 7:21 AM, Brian Beezley wrote:

After much searching, I was able to locate D-2852, the ASTM standard for the styrene pipe coupling. As expected from the application, it does not contain any electrical specs. Here's the description of its composition:
5.1 Materials—The pipe and fittings shall be made of styrene-rubber (SR) plastics meeting the following requirements:
5.1.1 The SR plastics compound shall contain at least 50 % styrene plastics, combined with rubbers to a minimum rubber content of 5 %, and compounding materials such as antioxidants and lubricants, and may contain up to 15 % acrylonitrile combined in the styrene plastics or rubbers, or both. The rubbers shall be of the polybutadiene or butadiene-styrene type, or both, with a maximum styrene content of 25 % or nitrile type. The combined styrene plastics and rubber content shall be not less than 90 %. No fillers may be used.
It sounds like a form of ABS, acrylonitrile butadiene styrene. Here are the dielectric constant and dissipation factor for ABS and PVC, both at 1 MHz:
ABS 3.20 .008
PVC 2.88 .016
The higher dielectric constant of ABS pretty much negates its loss advantage. See the two models for a coil using a 4.5" pipe coupling below.
Brian

pipe and tubing will have wild variations in EM properties. Not specified, so therefore not controlled.

I used to build a lot of HV gear using plastic pipe in various forms. One problem is that seemingly pristine PVC pipe might actually have all sorts of inclusions and other materials - if you cut off the outside on a lathe (e.g. to turn grooves) you might find that the nice white pipe is grey or black inside. And that might be from carbon black (a cheap additive that makes it UV resistant, among other things). PVC is also pretty hydrophilic so it picks up atmospheric moisture, which also increases the loss and epsilon.

I'm sure you're right, Jim. I think most experimenters grab whatever is at hand and press it into service as a coil form. No telling what its characteristics really are. At one time or another I've used beverage bottles and cardboard boxes. I did find the dielectric constant and dissipation factor for a particular cardboard material. It's on the bottom of the dielectric list for a good reason. I expect it too attracts water.

My favorite writeup on wet coils:



I think this kind of loss would be easy to measure with a NanoVNA.

Brian

On Mon, Oct 10, 2022 at 05:53 PM, Brian Beezley wrote:

Measuring large coils can be difficult because they couple to everything,
including the Q meter enclosure and your body. See README.TXT for a simple way
to overcome this problem.

Yes, If I get within two feet of my coil at resonance, I can see the Q start to drop.
I use about 10" of styrofoam to get the coil away from the Q meter, probably not far enough, but
it's the best I can do.
There is a low loss ferrite available that if wound right will get you a Q over 1200 through the BCB.

And into the high 1400s If with contra wound.
Mikek