For those of you that have a (grid) dip meter and a spectrum analyser (or oscilloscope) -- could I ask you to do an experiment? The experiment is to observe how much the frequency of the dip meter is pulled, when it is brought near a resonant circuit.
I did this and was highly surprised by the results. I was wondering if my results are typical or not; hence the request for additional data.
The dip meter I am using is the two-transistor circuit here: https://www.b-kainka.de/bastel53.htm . This circuit has a number of peculiar properties which may complicate its use in a dip meter (peculiar properties such as oscillating at a frequency slightly, or even significantly, lower than the resonant frequency of the tank: see https://www.radiomuseum.org/forum/relaxation_oscillations_in_lc_oscillators.html). In spite of its disadvantages and difficulty of analysis, the circuit is easy to build and seems to work to some extent.

The device under test was a coil on a FT50-6 toroidal core resonated with a variable capacitor at about 14 MHz. I coupled the dip meter to the DUT as per method B on this page: https://www.qsl.net/iz7ath/web/02_brew/15_lab/02_dipper/english/pag02_eng.htm .

In my case, the (surprising) results were as follows.
Just by moving the dip meter coil near the resonant circuit being measured (the DUT or Device Under Test), the dip meter's oscillation frequency was pulled by several hundred kHz. I observed the dip meter's oscillation frequency via a small pickup loop (about 2 cm diameter) placed at a fixed location near the DUT. This pickup loop was then connected to my RTL-SDR, functioning as a crude spectrum analyser. This test was at about 14 MHz.
When the dip meter is coupled to a DUT, and when slowly adjusting the dip meter's oscillation frequency around the region of a dip, the dip meter's oscillation frequency can suddenly jump by several tens or hundreds of kHz! This causes the "snapping" behavior when slowly tuning lower in frequency across a deep dip, whereupon suddenly the deep dip disappears and the meter's needle snaps back up. At the instant of the needle's snapping back up, the oscillation frequency (as observed by the radiated signal picked up by the RTL-SDR) jumps down by several hundred kHz. This frequency jumping is probably caused by an over-coupled, double-humped response of the two resonant tanks (the dip meter tank and the DUT tank) forming a sort of "combined, bandpass resonator" that determines, in a complicated manner, the oscillator's frequency, and that pulls the dip oscillator's frequency away from its natural frequency of oscillation if it were not coupled to the DUT. As the dip meter's tank is adjusted, the properties of the "combined, bandpass resonator" probably change until the the point where the influence of the coupled DUT's tank becomes too weak and the dip meter's own single tank suddenly dominates the oscillator behavior, causing the meter's needle to snap back, and causing the frequency jump.
If the coupling between the dip meter's coil and the DUT is reduced to a very low level such that the frequency jumping no longer occurs, then unfortunately (with my simple dip meter) the dip is no longer visible on the meter.
Even with coupling reduced to produce the weakest possible dip, the adjustment of the dip oscillator's frequency still was not smooth and exhibited sudden jumps in frequency in the region of the dip.
In practice, for a given coupling between dip meter and DUT, I found that the frequency where the deepest dip occurred, even though it pulled the oscillator frequency away from its natural oscillation frequency, was the most accurate indicator of the DUT's frequency, verified by measuring the DUT's resonant frequency on my NanoVNA (with a single-turn link winding through the DUT's toroidal coil). In other words, the frequency pulling doesn't seem to matter (as long as you can measure the frequency with a spectrum analyser or frequency counter) -- the frequency of the deepest dip (even if that frequency has been pulled by over-coupling to the DUT) still seems to correctly represent the resonant frequency of the DUT.
It may be that the above odd behavior is due to the limitations of my dip meter circuit. Or, it may be that all dip meters -- even the best ones -- exhibit frequency pulling and frequency jumping when tuning the oscillator around the dip.
To summarise, the surprising thing to me was that my dip meter's oscillation frequency was far different that what I expected. It was heavily pulled by by coupling into the DUT, and tuning across the dip yielded unexpected jumps in the oscillation frequency.
So my questions to you, if you have a dip meter, and a spectrum analyser, are:
Frequency pulling: Use a small pickup loop to detect the radiated signal from the dip meter, and display it on the spectrum analyser (or oscilloscope). When a dip is found, take a note of the oscillation frequency on the spectrum analyser (or oscilloscope). Then, completely remove the DUT and leave the dip meter tuning untouched. Does the oscillation frequency of the dip meter (as seen on the spectrum analyser) change? By how much?
Frequency jumping: When tuning across a dip, continuously observe the oscillation frequency (as observed on the spectrum analyser or oscilloscope). Does the dip meter's oscillation frequency suddenly jump by several 10s or 100s of kHz when tuning slowly across the dip?
ll a GDO is , is an oscillator with some kind of indicator to give a reading of the current the oscillator is using from the battery. When it couples to an external resonant circuit, the absorption of energy reduces at resonance giving the dip. Because the GDO is free running oscillator designed to be effected by external influences , it can never be absolutely accurate with frequency, as others have said, its a ballpark figure. To minimise frequency shift at resonance, the coupling too should be minimal, this will prevent the measured resonant tuned circuit from pulling the oscillator too much to give a more accurate reading on the dial. The frequency pulling is a requirement for them to work , if they didnt, they would not work.
These simple devices are a fantastic aid for anyone winding coils, and the quickest and easiest way to design RF coils, it serves as a quick and dirty RF signal generator, even Q can be gauged by noticing the distance from the DUT, the dip will occur, higher Q equals greater distance. The effective permeability of tuning cores and ferrite rods can be gauged by noting the frequency shift with or without, and with ferrite rods, the upper frequency limits can be discerned by winding a rough coil on the rod and attempting to dip it, the dip will get weaker as the freq limit is approached. Variable and fixed Capacitors can be roughly measured too by using a fixed inductance and noting the freq shift when tuning the capacitor. Dipole center frequency can be measured by connecting directly to the coil terminals of most GDOs and tuning for a dip.... most will require a DC path through the aerial. RF field strength and local oscillator output can be checked using the GDO just below oscillation as a regenerative receiver. IF your GDO has a headphone output, most do, the GDO will serve as a very wideband regenerative RX. The GDO to anyone playing with RF is more use than a multimeter and they are often overlooked these days as a useful bit of test equipment. I have worn several out over the years..
Following up on my earlier comment:
My 1.2 volt battery has run down to 0.816 volts, and I am getting much larger dips now. Idle (no-dip) current through the meter is about 20 uA, and when dipping my toroidal tank at 12 MHz, the current dips all the way down to 10 uA, a deflection of a 10 uA or slightly more (but still with the needle snapping back up and the frequency jumping once the oscillator is detuned too far across the dip). I can also approach the dip slowly from both the high-frequency side and the low-frequency side, and observe deep dips when approaching from either direction.
In contrast, with a fresh battery providing 1.2 volts, I get much shallower dips. Idle current is about 42 uA and the dip (on the same toroid as before) can only move the needle about 5 uA downwards. Furthermore, I can only slowly approach the dip from the high-frequency side; approaching the dip from the low frequency side is very difficult and does not dip the needle very much, and then the frequency suddenly jumps above the dip and I have to tune back down again.
In conclusion, there is definitely a strong influence of the supply voltage (and hence oscillator strength) on the depth of the dip. Lloyd Butler addressed this issue in his dip meter by adding damping parallel resistances across the dip meter's coils (reference: http://users.tpg.com.au/users/ldbutler/NegResDipMeter.htm). I tried this previously in my dip meter, but noticed no difference. Probably, the damping resistance has to be very carefully adjusted to almost, but not completely, damp the oscillation.
So a promising approach would seem to be:
Supply voltage for the circuit through a potentiometer wired as a variable voltage divider. Adjust for minimum voltage that still allows oscillation.
Replace the fixed 10k series resistor ahead of the microammeter with a 10k potentiometer wired as variable resistance. With 10k of resistance and 0.816V supply voltage, the idle current through the meter is 20 uA, but the meter goes up to 100 uA, so 80% of the meter's deflection range is unused and wasted. In this case, reducing the variable resistance from 10k to, say, 5k, would increase the idle current through the meter and might increase the amount of meter deflection, making the dip more visible.
You caught my attention when you said you used an SDR as a cheap spectrum analyzer. Spectrum analyzers are expensive. So I looked into it, and you actually can use an SDR as a spectrum analyzer. I dont know much about them, but if you do, this looks cool. https://www.rtl-sdr.com/tag/spectrum-analyzer-2/
Found some more information. Maybe BJT and FET dip meters are inferior to tube types.
https://www.w4cwg.com/ngdip.html
Now, I do recall reading somewhere that adjusting the oscillator gain (as in a regenerative receiver), so that it is just barely oscillating and almost about to drop out of oscillation, is one way to improve dip meter sensitivity. I haven't tried this, but it makes sense. Of course, as in a regenerative receiver, the gain control (regeneration control) will need to be constantly adjusted as the oscillator is tuned over a wide frequency range.
Therefore, to improve the sensitivity of transistorised dip meters, maybe we can try adjusting the oscillator loop gain. That should be implementable in my current dip meter circuit (a cross-coupled bipolar transistor oscillator) with a pot, wired as a voltage divider, in the emitter, pulling the emitter either upwards towards +Vcc or downwards towards ground, hence regulating the emitter current and loop gain.
Increasing the sensitivity would hopefully allow for much looser coupling, which would then hopefully reduce the frequency "snapping" and frequency pulling effects.
Another way to avoid the pulling would be to buffer the oscillator output to isolate it from the DUT, and use a link coil on the buffered output to excite the DUT. Absorption of energy from the link coil would then need to be detected by a diode and meter.
http://www.agder.net/la8ak/m11.htm
The same concept is described under the heading of "dip meter adapter" in this book about GDOs: https://worldradiohistory.com/BOOKSHELF-ARH/Servicing/How-to-use-Grid-Dip-Oscillators-Rufus-P-Turner-1960.pdf .
This isolation approach is probably better than the previous approach of trying to improve the sensitivity, because the isolation approach addresses the root cause of the problem (unwanted interaction between DUT and oscillator) instead of trying to work around the problem with increased sensitivity and looser coupling.
For the isolation approach, I guess I could use a JFET source follower or BJT emitter follower on my oscillator's tank, and try to connect the output link winding onto the source or emitter. Then, test if the oscillator frequency is pulled when the link winding is brought near a DUT.
It has been a long time since I used my dip meter, but your observations are consistent with my memory.
I was never successful in using mine to get anything more than a ballpark idea of the resonant frequency of a tuned circuit that I was coupling into.
It was terrific, however, as a wavemeter for tuning up tuned circuits, and checking xmtr output and I used it frequently for that purpose.
In the old days, dip meters really were "grid" dip meters, but my device is one of the newer generic devices using BJT transistors, and I wonder what influence, if any, that has on the coupling. Your theory of creating a double tuned circuit makes sense and seems consistent with your observations.
I wonder if an FET circuit, which might better emulate a triode, would change the results.
By all means, publish your audio sweeper when you get time, here or in another thread.
73,
Win W5JAG
Made an interesting discovery:
When tuning the dip oscillator slowly up in frequency across a dip, we will reach a frequency F0 above which the dip oscillator suddenly jumps upwards in frequency.
Then, when tuning the dip oscillator slowly back down in frequency across the same dip, we will reach a frequency F1 (where F1 > F0) below which the dip oscillator suddenly jumps downwards in frequency.
The "true" dip frequency lies exactly between F0 and F1.
In other words, with a dip meter (or at least, with my dip meter), we can approach the dip from either side and observe its presence. But, due to the mysteries of inductively coupled LC tanks, it seems that we cannot exactly reach the dip -- the oscillator is frequency is always pulled forcibly to one or the other side of the dip as we try to tune across it. The best we can do is to observe the frequencies F0 and F1 that lie closest to the dip, and take their average.
I was actually able to verify this hypothesis in hardware by using a separate, more advanced dip meter that I built some time ago. (If there is interest, I can write it up.) My advanced dip meter is tuned with a varactor and has a frequency sweep function that automatically sweeps the oscillator frequency up and down about a center frequency by about 100 kHz (by modulating the varactor control voltage with an audio-frequency sine wave), with the sweeping occurring at a frequency of about 1000 times per second.
When I coupled this rapidly-swept dip oscillator to the DUT, the rapid sweeping of the oscillator frequency seemed to successfully prevent the DUT from suddenly snapping the oscillator frequency away from the dip (because the oscillator was changing frequency so rapidly). Thanks to this snapping-resistant rapid frequency sweep mechanism, the frequency-swept dip meter is able to be tuned (without frequency snapping) over a continuous range of frequencies below, exactly on, and above the dip frequency. We can thus observe the energy extracted from the tank over this continuous frequency range.
As before, I observed the radiated output of the frequency-swept dip meter on my RTL-SDR using a small pickup coil. This shows a clear dip in the amplitude of the radiated signal of the frequency-swept dip meter, which corresponds to the "true" dip frequency at which the maximum power is being coupled out of the oscillator's tank.
In the below image, I used the gqrx software and my RTL-SDR to detect the radiated output from the frequency-swept dip meter. The dip meter is being swept from about 10.05 MHz to 10.65 MHz. A clear dip is visible at 10.394 MHz. Checking the DUT with my NanoVNA (with a link winding through the DUT's toroidal core) yielded the same resonant frequency, within about 50 kHz. The NanoVNA, with the link winding, reported a higher frequency. Some frequency difference is to be expected since the NanoVNA's test case did not couple the DUT to the dip meter's LC tank.
Again, the point is that even with a simple dip meter, we can estimate the true frequency of the dip (as shown above) by observing the F0 and F1 frequencies at which the frequency snapping occurs, and taking their average.
I've been checking YouTube and found this video of a commercial dip meter (Tesla) that shows exactly what I'm talking about. See the following video at 35 seconds, and also at 57 seconds, and watch his scope display and watch the meter as he tunes the oscillator. (His scope seems to be coupled to the DUT, not to the dip meter, and hence shows a peak in amplitude, not a dip, when the energy transfer is maximum due to the dip meter's being tuned to resonance with the DUT.)
It appears that his oscillator amplitude (as it is sampled at the DUT, and fed into the scope) is snapping abruptly as he attempts to tune slowly across the dip. His scope appears to show a constant frequency, but this is misleading -- his scope is connected to the DUT, so of course the scope will show the resonant energy in the DUT. I expect that at the oscillator, things look much different, and I expect that his oscillator is indeed jumping up and down in frequency as he tunes across the dip, and -- I think - it is the sudden shift on oscillator frequency that then causes the sudden snap upwards of the oscillator amplitude (because the coupling, due to the mistuning, has finally become so weak that the DUT can no longer influence the oscillator's frequency or amplitude).
https://youtu.be/VB3aAxgMIUo?t=35
The schematic for this dip meter can be seen here: https://www.radiomuseum.org/images/schematic-medium/tesla_praha/resonance_meter_bm342_1591805.png . Importantly, it uses a tapped coil in a Hartley oscillator arrangement.
This proves that the "snapping" behavior is not peculiar to my dip meter's cross-coupled, no-tap oscillator circuit.
Still, I would be interested in hearing from other dip meter users about how much frequency pulling and frequency snapping you observe.