TheRadioBoard

Seems like a ring down Q meter wouldn't be that hard to build. Anyone up for trying something like this?

In the circuit below, a 50 KHz or 100KHz square wave generator causes the coil under test to ring. The higher the Q the longer the ring. The ringing waveform is rectified and filtered by a high impedance precision rectifier, yielding a ramp wave. At some point the ramp will be .707 of the original peak of the wave.

With a high Q coil, the .707 point will occur later than it will with a low Q coil.
This time differential is output as differing pulse widths depending on the Q. This can be read directly by a calibrated meter, or digitally, with the additional circuitry shown in blue.

This seems too simple to be true, so I'm looking for flaws in this concept/execution. Thoughts?



Macrohenry

uhm, I am afraid the meter would have to work at 500-1700khz, or whatever frequency range is of interest for the coil in question.
this is because Q changes over frequency, ideally we would take samples over several frequencies (say every 100khz or so ) and draw a curve of the derived Q.
otherwise it sounds like a interesting project.
even more interesting if we let a computer draw the graph and do the counting...

Note that the LC cap is variable. A frequency counter would be needed to monitor that.

The meter is intended to integrate the pulses, similar to how speaker coils integrate pulse waveforms in Class D audio amplifiers. I saw one where the pulse train was at 200 KHz, yet as far as the speaker cone was concerned, the coil was behaving at audio frequencies.

Admittedly, there might need to be an IC buffer to make the meter work correctly, so there may be some sub-blocks missing in the block diagram.

I agree that a computer might be the easiest way to output the results. It still would be nice to have a little stand alone unit.

One cool thing I realized is that for a given Q the .707 point occurs at the same time length for 500 KHz as it does for 2000 KHz.

Freq_____Bandwidth__Q___Time to decay to .707
500000____500 _____1000____1/1000 sec
1000000___1000____1000____1/1000 sec
2000000___2000____1000____1/1000 sec

It seems to me that this serendipity is just begging to be exploited for measuring Q. Maybe that's the way the big boys (read expensive) do it, I don't know.

Mac

Hi Macro,
My only input would be to go for it. "Suck it and see."

First thing to keep in mind is that IMHO it does not have to be exceptionally accurate. If you could measure coils close to the accuracy of the HP 'Q' meter which is +/- 15% you would be doing OK. 'Q' results vary on the day or your proximity while reading the meter etc. so no great store can be placed on isolated readings.

To my mind, all that is needed is something that can indicate the potential quality of the coil and mainly at the upper end of the freq. range.
It is an eyeopener when winding some of these "odd-bod" unidentified toroids to see the poor 'Q'. Some as low as a few points WRT their measured inductance.

One would be confident that any coil exhibiting 'Q' of around 1000 is way better than one of a few hundred, even with a 10% error. It would quickly indicate if it's 'good' or 'bad' or 'getting better' dependant on your own criteria.

Don't forget that you may need a very Hi-'Q' cap. to resonate your coil or your 'Q' readings could be compromised. Which goes to show that the readings you get are only ever relevant to the components, the setup, the environment, (how you hold your tongue), etc.

WOW.

It's been established that when doing Q measurements that one must use as high quality varicap as possible. The following pictures dramatically demonstrate it. Here's the setup:

555 timer producing a slightly distorted square wave at 171 KHz is routed through a 1K resistor into a parallel RC combination. This produces a ringing waveform. I used a Dick Kliejer scope amplifier which loads the circuit about as lightly as it can be loaded.

First picture: Quality TRW varicap (Holy Grail) in parallel with Litz on styrene form. It looks like the ringing is never going to stop. To be able to see any ringing down, I'll need to stretch the pulses.



Second picture: Broadcast takeout varicap in parallel with Litz on styrene form. Note that the square wave has disappeared, ringing is just a few cycles.



Clearly, a low Q capacitor damps the circuit enough to prevent ringing.

Macrohenry

Interesting to witness it in 'visual' mode.

interesting indeed. keep researching

Actually you do NOT need to punch the coil with a higher frequency. You just need a square wave with a sufficiently fast rise time to stimulate the LC circuit at its natural resonant frequency. The square wave also needs to have a sufficiently long width to last beyond the decay of the sine wave. Remember, a fast rise time square has harmonics that go far beyond the broadcast band frequencies. You can calculate the Q of the LC with a simple formula (that escapes me just now). From this, if you know the Q of your capacitor, you can calculate the Q of the coil, and the frequency of the ring down sine wave is the resonant frequency of your LC. From this you can calculate the inductance of your coil under test.

In short, Tom is right on the money. I am surprised this isn't done more often. A simple-to-use piece of test equipment could be made using the techniques Tom has described.

Bruce