Which brings up........ In this entire thread, I'm surprised nobody has mentioned skin effect.
In an inductor, the skin effect (which appears as a FREQUENCY DEPENDANT RESISTANCE, as opposed to a reactance) has a profound effect on coil Q. You can't measure skin resistance with an ohmmeter, which is why you need something like a Q meter to calculate it.
Skin effect losses in a fat-conductor roller inductor are negligible at HF frequencies, but can be large in "wire" inductors such as Airdux and Miniductor, such as used in low end MFJ tuners and such.
Here's where we run across another odd paradox. Silver plating a small diameter wire doesn't do you any good, while silver plating a FAT conductor isn't necessary!
For ages and ages big inductors in A.M. broadcast transmitters were silver plated just on principle. When silver prices went through the ceiling in the 80s, manufacturers revisited the necessity of this. As it turns out, the gains made by silver plating inductors at broadcast frequencies were unmeasurable.
So...even the big boys make mistakes once in a while.
"A minute of measurement trumps a decade of debate."
Eric, Thanks for the explanations! I will be re-running the loss tests that I ran on the same transmatch. I hope to get to it this weekend. I am not convinced that min L will be result in min loss in the transmatch for all cases. I suspect the loss will be a function of what the impedance of the load being matched is. In other words, if the Z is low there will be lower loss with min L than will be the case with a Hi Z. With Hi Z I think you want to minimize the current thru the inductor. Why drop the impedance down in the middle of the transmatch when you are trying to raise it up a higher Z at the load. The latter might be the case that I remembered.
In regard to measuring the Q of an inductor, the Q meter pictured in this thread is an old Boonton Radio Corp job. I have this one and the version that is useable to 220 MHz ( I think they are models 160 and 190 respectively). These work quite simply by injecting a signal of low and known voltage into a circuit that is comprised of the unknown L and a calibrated variable C. At resonance the Q is simply the ratio of the voltage across the tuned ckt divided to the injected voltage.
I also have the BRC / HP 250 RX Meter (Scherring Bridge)that will measure the impedance of a coil in terms of its parallel reactance and resistance. Then Q = Rp/Xp.
I just acquired a GR 821A (Twin T Impedance Bridge). This also returns the impedance parameters so you can calculate Q.
The Palomar Noise bridge should be the useable as well it the bridge can be balanced (the Z must be in the range of the unit). I just obtained this old bridge and only had a moment to glance at it. However, I built the Wilfred Caron (SK)and the K2BT(SK) bridges that operate the same way.
When you measure the Q of a circuit it doesn't have to be the Q of a component. It is simply the Xs / Rs. Often the Q of the inductor is swamped out by the circuit Q so that the coil may not have much influence on the on the outcome whether you are using the full coil or tapping down. Eric pointed out the induction heating of the case and so forth so the inductors surroundings need to be factored in as well.
The definition Q is the amount of energy stored per cycle divided by the energy lost per cycle. No matter how the energy was lost (heat or radiation, etc.)
I have direct experence with silver plated conductors in a 40KW HF Xmitter, And will verify your statement as 100% accurate.
The Transmitter used a length of silver plated copper about .020 thick and 10 or 12 inches wide as the connection between the anode of a 4CX35000 and the pi-L network.
The strap was partially torn from being connected/disconnected, only about 2" was left connected.
Figuring a poor connection was better than no connection at all, I overlapped the ends and soft soldered them together.
Later, the "correct" replacement was obtained.
There was absolutly no measurable difference between the 3 cases!.
The big issue with minimum L/ Maximun C is not so much the difference of a few % efficency with absolute minimum possible L/C vs. most efficent L/C, But instead the much more likly possibility that using a ordinary Ham transmatch that you pick a grossly wrong L tap, and end up with a low VSWR but mucho circulating current, no power out to the antenna, and you burn up the transmatch.
Hence the warning to allways look for a match from the minimum L position.
Now my mistakes travel at the speed of light!:cool:
Last edited by KA5S; 12-06-2009 at 06:22 PM.
The link is broke , at least on my end .
PDF from a Google search (lazy me) but I have made it more direct.
And fixed the first one too.
Last edited by KA5S; 12-06-2009 at 06:23 PM.
If interested in the re-test results- loss of transmatch with max L vs min L for a T type transmatch:
I used the AI1H method where I adjusted the transmatch for a return loss of about 40 dB (the best I could achieve as my transmatch is not well shielded) into a resistance and then doubled the resistance then measured the return loss without touching the transmatch controls. I used a pair of 220 Ohm resistors in parallel then removed one for the doubling of the load resistance. Other resistances can be used but this took long enough given the limited time that I have
If the transmatch is lossless, the return loss will drop to 9.54 dB which is equivalent to a 2:1 SWR (the SWR was 1:1 (rho = 0.0) with 110 Ohms and then doubling the resistance to 220 ohms on the output should change the SWR on the input to 2:1 as well). The loss in the transmatch results in the return loss changing to something greater than 9.54 dB (rho = .33333).
The loss = 10 log (Rho after doubling the resistance / Rho = an SWR of 2:1 or 0.33333). In this case with min L Rho = 0.3126 so the loss = 0.28 dB.
In summary, the loss went up with decreasing L as I remembered from years ago. This still makes sense to me because with min L the circulating currents go up and there is more resistive losses in the circuit.
The 2:1 SWR bandwidth was also interesting to look at. With min L the bandwidth was about 150 kHz and with max L the 2:1 SWR bandwidth is about 900 KHz !
There is much more that can be tested here. I could have pushed my transmatch to higher inductances but the loss was already in the mud so it didn't make any sense to go further. Also, it would be good to use much lower resistances say 5 ohms doubling to 10 ohms for the 2:1 or starting with 10 Ohms and halving it. Frank Witt, AI1H did this by building his Geometric Resistance Box which had resistors that were related 2:1 (and 1:2 depending on how you run the test). Maybe using very low resistances, the min L has lower losses ? During the Xmas-New Year break, I might check this out.
Last edited by WB2UAQ; 12-07-2009 at 02:30 AM. Reason: Mistake