resistor resistance change over frequency

Mike, I'm sorry I haven't followed along very well, so feel free to ignore me.

I'm suggesting taking your resistors out of the resonant circuit and measuring R some other way. I guess it's hard 'cause there's going to be some phase shift too, you've gotta put the phase shift into the math.

George H.

Why would I have phase shift at resonance? At resonance the circuit is reduced to just the loss resistance of the coil and my resistor. I understand there are parasitics, from my pcb, but I shorted the resistors and they were barely measurable if at all. I would see any phase shift as a change of inductance or capacitance and I don't. All I see is an increased R.

Perplexed, Mikek

Again, I'm suggesting you take the resistor out of the resonant circuit.

George H.

Do you understand why I want the added series resistance?

Mikek

I think you are trying to reduce the Q by adding some R. Then you claim the R you are adding is not constant with frequency. (based on the Q measurements.) That doesn't seem right to me... I'm guessing there is something else going on. So I'm suggesting that you measure the frequency dependence of your R some other way.

But perhaps I've not understood.

George H.

You get it, except I'm NOT trying to measure the frequency dependence of my R, I'm measuring my R and finding it frequency dependent, but not reactive. Also with a major caveat shown below.

Here's the full story after multiple measurements to verify. Photo of Q meter with coil.

Side view showing Coil and connections

(The long leads are a temporary convenience.)

I connect the Rseries resistor on the HIGH side or the LOW side. High side is drive from the osc. Low side is 0.02 ohms above ground.

There is a large difference when I switch from high side to low side. That I will show below.

This coil has a Q of 350 to 420.

Here is the Rloss of my coil.

kHz

Here is the (Rloss+Rseries)-Rloss = Rseries at frequency. This is mounted on the LOW side.

kHz > 500 2.13 > 600 2.17 > 700 2.19 > 800 2.10 > 900 2.27 > 1000 2.28 As you can see the 2.15 ohm (Rseries) resistor > 1100 2.33 increases with frequency > 1200 2.21 > 1300 2.45 > 1400 2.45 > 1500 2.54 > 1600 2.69

Here is the (Rloss+Rseries)-Rloss = Rseries at frequency. This is mounted on the HIGH side.

kHz > 500 2.09

I have ran this multiple times and that's the way the data falls.

I'm going to try running a high side scan with the coil much closer to the 2.15 ohm resistor.

If anyone has info to explain this please comment.

Thanks all, Mikek

PS. Here is the change in my resonating cap without the Rseries vs with Rseries on the HIGH and LOW side. In Picofarads. Negligible!

500 0.3 0.2 600 0.2 0.1 700 0.1 0 800 0 -0.2 900 0.2 -0.1 1000 0.2 0 1100 0.1 -0.1 1200 0.3 0.1 1300 0.2 0.1 1400 0.2 0 1500 0.2 0 1600 0.3 0.1

High Side Low side Cap Delta Cap Delta

Empirically since one is trending up and one down then by a judicious choice of two resistors in the right ratio you should be able to make the system resistive component more or less frequency independent.

The ratio should be about 3:1 but which way around...

BTW I don't like the look of that dip at 1200kHz on both traces.

--
Regards, 
Martin Brown

Ya, I thought about that, but why is the resistor changing?

Looks like it goes back to the first measurement without a series resistor. I compared it to a couple other scans and this value is higher at 1200kHz. The value used was 4.5 ohms, I think 4.37 to 4.4 ohms is more accurate.

I get goofy after 3 hours of setting the freq, adjusting the multiplier knob, setting the big cap, checking the multiplier, setting the little cap for maximum Q, checking the frequency and multiplier and tweaking the little cap for maximum Q, Then logging Multiplier, Q reading and Capacitance reading. Then adjust frequency up 100kHz and start all over!

He should still be ok; Google can tell us all about mean, median, average, interquartile range, etc. :D

Michael