开云体育

I've been following this article on how to accurately measure the Q of a coil:



I measured a series LC circuit and converted the S11 file to R+jX. When I plotted R against frequency I observed a peak at the resonant frequency. Any idea what is causing this?

--
Mike

Sorry, heres the graph with numbers!

--
Mike

OK if I may answer my own question, I repeated the measurement using the S21 method and got a sensible answer.

--
Mike

Hi MIke
What component type are you using for the test? Trying to understand why the R is changing myself. I would think the resistance of a fixed series resistor would remain fairly constant if it has low reactance itself. PA3A's article shows an example of measuring the Q of an inductor so it is understandable that he would measure a Q of over 200 and a low resistance for the iron powder Micrometals core. His R stays flat as would be expected. I am not familiar with the 4C65.
Vector network analyzers were not the preferred instrument to us for measuring the Q of a coil. The R is so small with respect to the reactance that small variations of the phase angle and magnitudes of the reflection coefficient would make the Q and the calculated R to vary considerably. The old Q meters remained in favor for many years because they used a parallel tuned circuit method (old Boonton Radio Corp 160 and the HP 4342 for instance). In any case it would be interesting to know what type of inductor you are testing. What is the core?
73, Pete

Hi Pete, it's an air cored coil with an inductance of around 108uH. I made the measurement with an 18pF COG capacitor in series so the plot is showing the resistance around the resonant frequency of the circuit. However at 15 ohms it's higher than I was expecting.

--
Mike

Hi Mike,

Please see message # 23812. This documents one approach.

Alan

Mike
Is the scale of your reactance graph supposed to be in kohms and not just ohms? I think so because at 3 MHz the impedance of the series L-C would be about -j911 ohms. Correct?
Pete

Mike, Scratch that idea about the scales because it doesn't work the same at 4 MHz where the Z should be about +j503 ohms.

Mike
15 ohms is not too far from what a long inductor made with small gauge wire could have. Q of about 143.
Pete

On Mon, Jul 3, 2023 at 12:23 AM, alan victor wrote:

Hi Mike,

Please see message # 23812. This documents one approach.

Alan

Thanks Alan I'll have a look at that.
--
Mike

If difficulty in finding the paper from EDN and the figures, here is a pdf abbreviated and a test case circuit screen capture. If you are careful in fixture construction depending on operating frequency, unloaded Q of 350 or more is easy to obtain. You really need to take care in any Q measurement to prevent series R losses from accumulating unless you can accurately de-embed the "stray" R value(s).

Alan

Mike,

the article is in this file /g/Test-Equipment-Design-Construction/files/Q-factor%20%28Q%20is%20the%20inverse%20of%20dissipation%20factor%20DF%29./Method%20simplifies%20testing%20high-Q%20devices;%20Alan%20Victor,%20EDN%20Feb%202002.pdf. [It's no longer available on the EDN website with the all-important diagrams!]

HTH, 73, happy (or otherwise) measuring,

Robin, G8DQX

Alan

I found the article from your earlier link and will give it a go, thanks.

In the meantime I repeated my measurement with NanoVNA-App calibrated for 6401points and got a slightly lower value of 13.6 ohms at resonance.

--
Mike

On Mon, Jul 3, 2023 at 03:27 PM, alan victor wrote:

If difficulty in finding the paper from EDN and the figures, here is a pdf
abbreviated and a test case circuit screen capture. If you are careful in
fixture construction depending on operating frequency, unloaded Q of 350 or
more is easy to obtain. You really need to take care in any Q measurement to
prevent series R losses from accumulating unless you can accurately de-embed
the "stray" R value(s).

Alan

OK I had a go at this with my NanoVNA and got the attached result, which if I follow your calculation correctly equates to Rq = 8.05 ohms.

Coil64 estimates 5.9 ohms which, given the less than ideal test jig I'm using, is not too far off.

Mike


--
Mike

Excellent Mike. So that is a Qul of 348. I spilt the difference between 6 and 8 ohms... lets say 7... From the appearance of your coil, that seems
reasonable. The material you wound the unit on has a loss tangent. Any idea what it is? Nice if it were .0005 or better.... Suspect it is not.

Alan, How does the loss factor of the dielectric that the coil form is made from impact the magnetic circuit? I have heard others mention this but with no explanation.
Thanks, Pete

Hi Pete.

The inductor is supporting a propagating field. It is both magnetic and electric. You are correct, while the magnetic properties of a core are key to control loss, so are the dielectric properties, as this impacts the electric field. A good way to approach all lumped components is not to consider them as lumped, but as distributed. An L and C are actually transmission line components. Particularly as the frequency increases. So the dielectric loss as it affects a transmission line is really no different in its impact when it supports a coil on an insulating form. Loss in the dielectric contributes a resistive shunt component and reduces the inductor Q. As a rule of thumb, try to use a material whose loss tangent provides a Q value (1/loss tan) of 5 x the inductor alone.

On 7/3/23 4:13 PM, WB2UAQ wrote:

Alan, How does the loss factor of the dielectric that the coil form is made from impact the magnetic circuit? I have heard others mention this but with no explanation.

Most coils have some parasitic C (it can be quite large). That C has a voltage on it, and a lossy dielectric leads to loss.

On Mon, Jul 3, 2023 at 11:24 PM, alan victor wrote:

Excellent Mike. So that is a Qul of 348. I spilt the difference between 6 and
8 ohms... lets say 7... From the appearance of your coil, that seems
reasonable. The material you wound the unit on has a loss tangent. Any idea
what it is? Nice if it were .0005 or better.... Suspect it is not.

It's PVC water pipe.

--
Mike

Mike and Alan,
I measured the loss factor for PVC and it was about 0.005 or so at 1 kHz. I did more measurements at RF but will have to go back to old note books. I made a simple fixture to test a number of plastics. I rebuilt a coil that plugs into a BC-610 transmitter and from that sprung all kinds of questions about dielectrics. Many local guys are using 3D printers and using PLA, I think it is called. They printed out a plate of it for me to test. I would not say that all of this material is made equally and permitivity and loss factor are probably not well controlled. Won't be able to get back to this for awhile as it is Independence Day and a lot going on to take care of this week. Can only say right now that none of this plastic seemed to jump out as being a terrible dielectric. I also tested plate glass because back in the early days of radio capacitors for spark transmitters used what I think was regular glass available at that time.
73