Measuring the Disspation of the self capacitance of a thru hole resistor.

The reason for the measurement error has been explained to my satisfaction.
First I had the skin effect backwards, that would increase resistance not decrease.

  A fine fellow on another group ran VNA tests of a 1/4W resistor, (no info whether thru hole or smd). 
 The graph shows a drop of the 10MΩ resistor drops to 6MΩ at 1Mhz. 
The explanation is, the dielectric loss of the resistor body in parallel with the 10MΩ.
 If you are interested, here's a link to the explanation and VNA graphs of a 1MΩ and a 10MΩ resistance, capacitance and Q.
https://groups.io/g/Test-Equipment-Design-Construction/message/1236

                                       Mikek

On Saturday, November 19, 2022 at 4:27:05 PM UTC-5, Lamont Cranston wrote:
> The reason for the measurement error has been explained to my satisfaction. 
> First I had the skin effect backwards, that would increase resistance not decrease. 
> 
> A fine fellow on another group ran VNA tests of a 1/4W resistor, (no info whether thru hole or smd). 
> The graph shows a drop of the 10MΩ resistor drops to 6MΩ at 1Mhz. 
> The explanation is, the dielectric loss of the resistor body in parallel with the 10MΩ. 
> If you are interested, here's a link to the explanation and VNA graphs of a 1MΩ and a 10MΩ resistance, capacitance and Q. 
> https://groups.io/g/Test-Equipment-Design-Construction/message/1236 

That is interesting.  I'm surprised this has significant impact at 1 MHz.  

I'm not convinced there isn't something else going on too.  Notice that even at 300 kHz the 10 Mohm resistor is 7.2 Mohms.  I think that's a bit hard to explain.  You can see from the graph that at 300 kHz, the effect is almost gone and the curve is nearly level.  But maybe that's my bias. 

-- 

Rick C.

---- Get 1,000 miles of free Supercharging
---- Tesla referral code - https://ts.la/richard11209

On Saturday, November 19, 2022 at 4:08:58 PM UTC-6, Ricky wrote:

> > If you are interested, here's a link to the explanation and VNA graphs of a 1MΩ and a 10MΩ resistance, capacitance and Q. 
> > https://groups.io/g/Test-Equipment-Design-Construction/message/1236

> That is interesting. I'm surprised this has significant impact at 1 MHz. 

 I thought the resistance was 10MΩ at DC. but it's not, now I'm unclear, I thought the R and C were completely separated in the graph. 
 
> 
> I'm not convinced there isn't something else going on too. Notice that even at 300 kHz the 10 Mohm resistor is 7.2 Mohms. I think that's a bit hard to explain. You can see from the graph that at 300 kHz, the effect is almost gone and the curve is nearly level. But maybe that's my bias. 

 Are you looking at graph number three?

 I'll see if I can get anymore info,
                                                  Mikek

On Saturday, November 19, 2022 at 10:00:56 PM UTC-5, Lamont Cranston wrote:
> On Saturday, November 19, 2022 at 4:08:58 PM UTC-6, Ricky wrote: 
> 
> > > If you are interested, here's a link to the explanation and VNA graphs of a 1MΩ and a 10MΩ resistance, capacitance and Q. 
> > > https://groups.io/g/Test-Equipment-Design-Construction/message/1236 
> 
> > That is interesting. I'm surprised this has significant impact at 1 MHz.
> I thought the resistance was 10MΩ at DC. but it's not, now I'm unclear, I thought the R and C were completely separated in the graph.
> > 
> > I'm not convinced there isn't something else going on too. Notice that even at 300 kHz the 10 Mohm resistor is 7.2 Mohms. I think that's a bit hard to explain. You can see from the graph that at 300 kHz, the effect is almost gone and the curve is nearly level. But maybe that's my bias.
> Are you looking at graph number three? 

Three or four, both have the same resistance curves. 


> I'll see if I can get anymore info, 

You have a lot more experience in this than I do.  I just came to the realization that Q depends on the frequency it is measured at.  Since Q = XL / RL, and XL depends on frequency, the Q will vary over frequency even if nothing else changes.  Higher frequencies will give a higher Q... at least until other parasitic values kick in, like the self capacitance.  I expect you already knew that. 

As to the change in value of the resistor, I though there might be something going on with the dielectric effect, but I expected the impact would be very slight.  However, when I calculated the impedance of the 0.12 pF at 1 MHz, it comes to 1.3 Mohms, so clearly this can have significant impact if the dielectric is very lossy.  

-- 

Rick C.

---+ Get 1,000 miles of free Supercharging
---+ Tesla referral code - https://ts.la/richard11209

On Sunday, November 20, 2022 at 12:34:37 AM UTC-6, Ricky wrote:
> On Saturday, November 19, 2022 at 10:00:56 PM UTC-5, Lamont Cranston wrote: 
> > On Saturday, November 19, 2022 at 4:08:58 PM UTC-6, Ricky wrote: 
> > 
> > > > If you are interested, here's a link to the explanation and VNA graphs of a 1MΩ and a 10MΩ resistance, capacitance and Q. 
> > > > https://groups.io/g/Test-Equipment-Design-Construction/message/1236 
> > 
> > > That is interesting. I'm surprised this has significant impact at 1 MHz. 
> > I thought the resistance was 10MΩ at DC. but it's not, now I'm unclear, I thought the R and C were completely separated in the graph. 
> > > 
> > > I'm not convinced there isn't something else going on too. Notice that even at 300 kHz the 10 Mohm resistor is 7.2 Mohms. I think that's a bit hard to explain. You can see from the graph that at 300 kHz, the effect is almost gone and the curve is nearly level. But maybe that's my bias. 
> > Are you looking at graph number three?
> Three or four, both have the same resistance curves.
> > I'll see if I can get anymore info,
> You have a lot more experience in this than I do. I just came to the realization that Q depends on the frequency it is measured at. Since Q = XL / RL, and XL depends on frequency, the Q will vary over frequency even if nothing else changes. Higher frequencies will give a higher Q... at least until other parasitic values kick in, like the self capacitance. I expect you already knew that. 

 My experience is mostly with AM Broadcast frequency Inductors, both Air core and a very low loss ferrite R40C1, that can match or exceed the Qs of a good air cap. Most inductors optimized for the BCB have a peak Q around 800kHz and drop as frequency increases or decreases. There are some tricks to flatten this.
 Below are some graphs I made. Five years ago I wanted to see what the optimum wire spacing was for a 6" diameter coil, and wound 5 coils with different wire spacing. All were wound with 660/46 Litz wire.

Graph.
https://www.dropbox.com/s/eapq2fnpzpesn9v/5%20Coil%20Q%20Graph%20with%20labels.jpg?dl=0
An actual Coil as wound.
https://www.dropbox.com/s/ru9im2fecdi0g3f/9%20TPI%20coil.jpg?dl=0

This is a graph of a coil made with the R40C1 ferrite material. Measured on 3 different Q meters.
https://www.dropbox.com/s/au2sp79rgta1d80/Elusive%20Q.jpg?dl=0
As I say "Q is Elusive"

> 
> As to the change in value of the resistor, I though there might be something going on with the dielectric effect, but I expected the impact would be very slight. However, when I calculated the impedance of the 0.12 pF at 1 MHz, it comes to 1.3 Mohms, so clearly this can have significant impact if the dielectric is very lossy. 

 I can't make the numbers work out, It seems with the Q and the value of the Capacitance, it should all just shake out 
and find a 40MΩ resistor in parallel with the 10MΩ.  i.e. 10MΩ // 40MΩ = 8MΩ. 
I thought it was a series parallel conversion situation, but that doesn't work either.
btw, here is a nice series to parallel, and vise versa calculator, otherwise I wouldn't have a clue.
https://daycounter.com/Calculators/Parallel-Series-Impedance-Conversion-Calculator.phtml

                                       Mikek