Testing the TK Tar Baby

[MileHigh]

TK:

I don't think I have ever used a capacitance meter or an inductance meter.  It's funny I vaguely remember doing the engineering labs where you look at RC and L/R time constants to measure capacitance and inductance, "like a man."  So I kind of frown to myself when I hear about capacitance and inductance meters.

I think we are back to the always interesting issue of the limits for your measuring apparatus.  What I did on the bench with a scope had it's limits.  Similarly for both capacitance meters and inductance meters, eventually the values you are trying to look at get so low that either the measurement device was never designed to go that low, or the values that you are trying to measure fall under the "background noise level" and the capacitance and inductance of the leads of the instrument are larger than what you are actually trying to measure.

You mentioned that most inductance meters use about 900 Hz.  Right away I am thinking forget it, it will never be able to measure nanohenries, it was designed for milihenries and possibly microhenries.  Any lower than that forget it.  There is always "RTFM" but a lot of these commodity meters don't actually give you the real specifications.  That's a Marketing decision to increase sales (or to prevent reduced sales).  The Dark Side of Capitalism.  They know that the average Joe Blow might not buy and say, "What?  It can't measure picohenries??!!"

What I think PW may be suggesting is the following:  I assume that your signal generator has a 50-ohm output impedance.  Just solder your resistor array to a 1/2" length of coax connected to a male BNC connector and then jack it straight into your function generator.  Then sweep up the frequency and observe the voltage across the resistor array.  Assuming that the inductance dominates as you go up towards 1-2 MHz, you should see the AC voltage across the resistor array increase as you go higher in frequency.  Obviously at lower frequencies you have 50 ohms in series with 0.25 ohms, so you should meaure a very low output voltage.   But at 1.5 MHz if the impedance is 16 ohms you should see a higher AC voltage.

So I suppose that this method will work out to much higher frequencies.  But there are limits of course.  Eventually the inductance inside the function generator itself will start to come into play and the stray capacitance across the leads of the resistor array itself will come into play.  As a certain frequency you should see a roll-off and the AC voltage across the resistor array will start to decrease again.

Somebody might correct me here but I think I am in the ballpark.

MileHigh

[MileHigh]

You know another interesting tidbit here is about when to use 10X attenuation on your scope probes.  I think that some people might always defer to 10X attenuation thinking that it's better all the time.  I believe it's preferable to use 10X attenuation if your signal is very high in voltage or you are looking at a signal that has a really high inherent impedance and you want your probes to be as high an impedance as possible to reduce the disturbance that they cause to the device under test.  The trade-off is that your signal seen by the scope is weaker and it's own impedance is higher as subject to disturbance, and I believe that you lose some bandwidth.

So I would suggest that you don't keep the probes on 10X attenuation in this case.  The signal source impedance is a very robust 50 ohms, and you won't disturb it with the normal probe setting and you will get a "clearer" signal and hence a more accurate measurement of the AC voltage across the resistor array.  In other words, you won't have to worry about any possible frequency roll-off.  I am pretty sure that there are no roll-off issues for 10X probes at 1.5 MHz but using non-attenuated setting just "feels better."  I am sure you can relate to that.

Note one more thing.  If you assume inductive wire-wound 10-watt resistors, they are wound like like little inductors.  When four of them are in close proximity the mutual inductance between resistors may come into play.   I am not really sure, but it implies that you could arrange 2 + 2 such that there is some self-cancellation, hence reducing the inductance of the array.

Again, I am not an analog guy so what I am saying may be subject to correction.

MileHigh

[picowatt]

TK,

If you have another resistor similar to the ones you used for your CSR, try placing it directly across the FG terminals with the FG set to 1.2MHz (I am assuming an FG with a 50R output).  Scope across the FG before and after connecting the resistor.

If they truly are 7uHy, and therefore close to 52R at 1.2MHz, you should see the FG output go down by 6dB (half).  If the output drops much more than this, they are not 7uHy.  You could sweep the freq to get an idea of what the resistor is "acting like" at higher frequencies as well.  There are a few issues that make this method unable to tell if the observed reactance is inductive or capacitive, but should suffice for the way the resistors are to be used.  If you have no spare, use the four in your circuit (you will have to free the connections at least on one end) and modify calculations/frequencies accordingly.  You can calculate the equivalent resistance based on the observed drop at a given frequency (with the Rgen 50R as part of the divider) and calculate an inductance from there.  There may be a capacitive component, so I would rather just say that at 1.5MHz, their ESR is the observed/calculated value

Possibly, the low resistance or some internal capacitance of the resistor are fooling the LCR meter.

The lower values your alternate inductance tests indicate are more inline with other values given.

I would want to know how the resistor performs at the fundamental and the second and third harmonic, as these are most prominent in the waveforms.

PW



 

[TinselKoala]

I've done some work on the spreadsheet, incorporating power computations for various CVR reactances. This affects only the absolute magnitude of the power values, not the sign or the shape of the power curve of course.

All "positive" values of shunt inductances still yield NEGATIVE mean power values.

If anyone has a better idea as to the overall impedance of this CVR at 2 MHz please let me know so I can revise the spreadsheet again.

I am happy to use 125 nHy as the total inductance of the CVR based on my resonance measurement instead of the ProsKit meter.

The reactance at 2 MHz will then be
2*pi*f*L == 2(3.142)(2,000,000)(0.000000125) == about 1.57 Ohms,
and the  impedance then is sqrt(1.6^2 + 0.25^2) == about 1.59 Ohms.

With that as the AC impedance at 2 MHz, the mean power shown on TarBaby's scopeshot appears to be about _negative_ 14 Watts.

Right?

Please feel free to check my work -- those of you with the knowledge to do so -- and go over the spreadsheet with a finetooth comb, so that I MAY CORRECT MY ERRORS AND POST CORRECTIONS AS SOON AS POSSIBLE, like a good little scientist should always do. This, after all, is a major reason why RAW DATA and computations are always made available to reviewers of experimental work.

[picowatt]

TK,

If the values being used for the inductance/ESR are closer to the values your alternate measurements arrived at, you will have to crank up the FG output to get an accurate measurement of the drop with the FG's 50R.  If they are closer to 1R at 1.5MHz, with your FG at 40volts (amazing!), you should be able to read the divided voltage fairly well.

PW