Total posts : 45366
I was particularly interested in Item 6 in Macrohenry’s post, which dealt with Q measurement using Ben Tongue’s method. Ben Tongue’s method has been discussed recently in another thread. To reiterate, the method consists of applying a small amount of RF power from a signal generator to a small coil. The signal from this small coil is loosely coupled to the cold end of the coil under test. The coil under test is connected to an air-variable tuning capacitor, and the coil and capacitor are tuned to resonance at the desired test frequency. The ground clip lead of a 1X oscilloscope probe is connected to the cold end of the coil under test. The probe tip is clipped to the hot end of the coil under test, but a direct conductive connection is not made. The probe tip is clipped to an insulated wire at the hot end of the coil under test. This provides for very low capacitive coupling between the hot end of the coil and and the 1X oscilloscope probe. The signal generator either has a digital frequency readout, or it is connected to a frequency counter. Initially, the signal generator frequency is adjusted for a peak reading on the oscilloscope, and the amplitude at resonance is noted. Then the bandwidth between the 3 dB points (1/Sqrt(2) below the amplitude at resonace) is found. This bandwidth is divided into the resonant frequency to give the Q.
Macrohenry states that Ben Tongue’s method does not accurately measure the Q of a coil if the Q is very high. Macrohenry recommends using an FET buffer amplifier at the hot end of the coil under test to get more accurate Q measurements. I did not use the circuit Macrohenry recommended, but I have another similar circuit available. It consists of a 2N4416 source follower with a 727 ohm source resistor to ground. I used this circuit, which is constructed on a small copper-clad fiberglass board, to isolate the hot end of the coil from a 10X oscilloscope probe. Based solely on circuit analysis, I determined that the input impedance of the buffer is about 30 M ohms in parallel with about 3 pF. This input impedanci is enough to significantly reduce the Q reading of a loading coil. For a particular coil that I had tested using Ben Tongue’s method, I obtained a Q of 438. Using the FET buffer reduced the Q reading to 349. The FET buffer just applied too much of a load to the coil under test. This additional load was not present when using Ben Tongue’s method.
There is a counterintuitive aspect of Ben’s method that I wish to point out. The resistance in series with the signal coil needs to be higher than one would expect. I originally used a small scramble-wound signal coil, with an inductance of 67.3 uH, with a 27 ohm resistor between the signal generator and the coil. Increasing the resistance to 820 ohms resulted in considerably higher Q readings. It at first seemed to me that the the higher resistance in series with the signal coil would reduce the Q of the test system, but the opposite turned out to be the case. I verified this conclusion by performing a circuit analysis of the loosely-coupled RF transformer that constitutes the test setup. The signal coil should be located as far from the coil under test as possible in order to maximize Q readings. The mutual inductance must be low to get accurate results. I use so much separation between coils that I see appreciable 60 Hz interference. This is not a problem when the oscilloscope is used with line sync.
Macrohenry mentioned obtaining poor results in Q measurements when using a Heathkit oscilloscope. I have never used a Heathkit oscilloscope, but I have used a Heathkit PK-1 probe with an EICO Model 460 oscilloscope many years ago. After I upgraded to a Telequipment D52 oscilloscope (which I still have, but only as a backup for my “good” scope), I modified my Heathkit scope probe from banana plugs to a coaxial connector, and continued to use it. I still have the probe, and I tried to obtan a Q measurement with it. I got a Q reading of 380 using the Heathkit probe compared to 438 when I used an old Textronix 1X probe. A slight lowering of the resonant frequency I obtained indicated that using the Heathkit probe added about 2 pF to the test circuit. Also, a higher signal on the oscilloscope indicated that there was more capacitive coupling from the hot end of the coil under test when using the Heathkit probe than with the Tektronix probe, causing the Heathkit probe to function as a heavier load than the Tektronix probe. The Heathkit probe does not have a clip at the tip (just a sharp point). I tied a loop of insulated wire around the tip to hold the probe in place.
Ben Tongue recommends measuring Q over a sheet of metal used as a ground plane. This practice reduces the Q measurement, as Macrohenry’s recommendation of keeping the coil under test away from metal would suggest. I, myself, perform Q measurements keeping the coil under test away from metal, but I can see why Ben recommends testing above a ground plane. Testing above a ground plane keeps the Q measurement more consistent. Rattan explained the situation very well in this insightful comment in his last post in this thread.
“I’m pretty sure that most of the conditions for measuring the Q of very high Q coils will be nullified to some degree in mounting it as part of a physical antenna.”
Using a metallic ground plane when measuring the unloaded Q of a coil measures said unloaded Q under more realistic conditions. The self-capacitance of the coil, which affects Q a great deal, does not depend only upon the the properties of the isolated coil itself, but upon its surroundings. For the coil I measured for this post, I used a dip meter to measure the self-resonant frequency of the isolated coil. The self-capacitancce of the coil measured to be only .76 pF. When measuring the Q using Ben Tongue’s method, the self-capacitance measured to be 11.3 pF. I was not using a ground plane, but the cold end of the coil was connected to ground through the ground lead of the oscilloscope probe. When using a ground plane, the self-capacitance of the coil increases, depending upon how far above the ground plane the coil is located. More consistency in results can be obtained if a consensus can be reached about a standard height (like a foot, for example) above a ground plane at which a Q test should be performed.
Because of the variables that are involved, Q is, at least partially, in the eye of the beholder. Different people get different Q measurements while testing the same coil. Some people consistently get high Q, and others consistently get low Q. Wishful thinking plays a part in Q measurement. For high-Q coils, the tests are not very robust because the measurements are at the threshold of detectability of the test instrumentation. It is not easy to get the exact resonant frequency of the coil because the peak of the resonance curve is nearly flat. There is a considerable region near resonance where the amplitude seen on the oscilloscope does not change much. Similarly, it is not easy to detect the 3 dB points because the resonance curve is quite smooth at these points. This is where wishful thinking becomes a significant part of measurement error. If the tester is completely objective, the average of many tests would give an accurate Q measurement. But testers, being people, are likely to (perhaps unconsciously) bias the tests.
I noticed that several Q measurements on the web sites linked by Macrohenry report measurements above 1000 using the HP (Agilent) 4342A Q meter, even though the upper limit of this meter is 1000. In principle, higher Qs can be measured by adding a known loss resistance to the coil under test. This practice can result in unsatisfactory results. Adding loss resistance broadens the resonance peak, making the resulting corrected Q measurement inaccurate.