Non motion detector

[ArchibaldSTM]

I have such a photos... There is a board with elements inside. Unfortunately no other information

[ivconic]

I have such a photo... There is a board with elements inside. Unfortunately no other information

That's what I am talking about. Unfortunately the "mass" is so strong and hard; I am almost sure that tiny wires will be broken if I try anything.
I have bad experiences and lot of bad results in numerous attempts to "decipher" inner composition and wirements in coils.
Usually I don't mind. But this time it is about very old and (in my case) rare and hard to obtain coil.

[ArchibaldSTM]

And such a scheme clear things up?Although it's a little different... Hmm

[ivconic]

Wouldn't such a scheme clear things up?

Thanks!
I had it.
Most probably this coil is alright and I wouldn't need to do anything about it.
But I wanted to know inductances... maybe I would want to try to make DD coil for it.
Or larger one.
Never mind. First to fix the machine and later i will see.
Thanks gain; previous 3 pics are valuable for me.

[ArchibaldSTM]

Well, I helped as much as I could.

[ivconic]

Well, I helped as much as I could.

[Davor]

I'll be lurking here.
I have some concepts in my head for some time now, only I was too much engaged otherwise to put them on paper.
In short, you can't expect to have a truly static response if you don't compensate for "air signal" variation. There are 2 possible paths. One is using a Tx envelope voltage as AD converter reference, and effectively get a 1/(air signal) multiplication. In effect it would act as if your Tx is rock solid, and there is no "air signal" variation at all. In digital processing the Dc component of such rock solid offset is simply deducted, and all that remains is a target response.
The other approach would compensate the air signal by a scaled down sample of Tx envelope. So instead of a compensation using a reference voltage, which would effectively remove only the unchanging bulk of the "air signal", you'd compensate using a voltage that follows the very "air signal" change, and in effect lose the majority of chatters caused by it. The compensation would be adjusted with preferably digital potentiometer. The resulting target signal should be amplified with aDC amplifier, say CDS or chopper. Sampling demodulator is in itself a CDS of a kind.
The first approach would work fine for rigs with signals processed digitally, while the other may be more universal, yet also more complex. Both approaches would improve sensitivity over motion detectors because the chatters caused by "air signal" variation can't be removed by motion filtering.
There is perhaps a 3rd approach, using a rock solid Tx, say square wave, supplied with superior power supply. I don't think a free running Tx without proper compensation of "air signal" in Rx would ever cut it.

[ivconic]

I'll be lurking here.
I have some concepts in my head for some time now, only I was too much engaged otherwise to put them on paper.
In short, you can't expect to have a truly static response if you don't compensate for "air signal" variation. There are 2 possible paths. One is using a Tx envelope voltage as AD converter reference, and effectively get a 1/(air signal) multiplication. In effect it would act as if your Tx is rock solid, and there is no "air signal" variation at all. In digital processing the Dc component of such rock solid offset is simply deducted, and all that remains is a target response.
The other approach would compensate the air signal by a scaled down sample of Tx envelope. So instead of a compensation using a reference voltage, which would effectively remove only the unchanging bulk of the "air signal", you'd compensate using a voltage that follows the very "air signal" change, and in effect lose the majority of chatters caused by it. The compensation would be adjusted with preferably digital potentiometer. The resulting target signal should be amplified with aDC amplifier, say CDS or chopper. Sampling demodulator is in itself a CDS of a kind.
The first approach would work fine for rigs with signals processed digitally, while the other may be more universal, yet also more complex. Both approaches would improve sensitivity over motion detectors because the chatters caused by "air signal" variation can't be removed by motion filtering.
There is perhaps a 3rd approach, using a rock solid Tx, say square wave, supplied with superior power supply. I don't think a free running Tx without proper compensation of "air signal" in Rx would ever cut it.

Come again!
I really enjoy watching you trash-talking!


P.S.
Welcome back!
Welcome here!
You know my intentions and plans... this will go step by step, in time...

[Davor]

Let me explain the 1st variant.
"One is using a Tx envelope voltage as AD converter reference, and effectively get a 1/(air signal) multiplication. In effect it would act as if your Tx is rock solid, and there is no "air signal" variation at all."

Air signal is a portion of Tx signal that is received in a Rx coil, either in phase or shifted in phase, and for ease of explanation let's assume it is 10mV. On top of this 10mV we'll have some amplitude variation in a whereabout of 1uV (in the case of a well-behaved Tx). This 1uV follows the 1/f laws, and is responsible for most of the chatters. After demodulation both 10mV cyclostationary, and 1uV variable AC voltages become const. and baseband variable respectively. Target response is superimposed on the air signal, and minimum response may be as low as 5nV, but it isn't. We'll see why.

In a motion detector both constant and variable components of the air signal hit a high-pass filter, after which the constant portion of 10mV is eliminated, while the variable portion is reduced. A true static operation is nearly impossible due to this variable portion of air signal, and even a motion filter struggles with it. Because it may have any phase at the coil, this component is not easily removed with a GEB.

The first solution:
If a Tx voltage is rectified in a way that the result is a Tx envelope voltage, of say 10V, and if an ADC receives a reference voltage input of about 1V, the Tx envelope voltage is reduced to 1/10 and fed to a reference voltage input of ADC, causing digital values to be multiplied by 1/reference. What happens is as follows:
Tx = 10+variable
Air signal = (10 + variable)/1000
Reference = Tx/10 = (10+variable)/10
Digital stream = (target + (10+variable)/1000)/((10+variable)/10) =target*(10/10+variable/10)+10/1000*(10+variable)/(10+variable)
where (10+variable)/(10+variable)=1, so
Digital stream = target*1 + target*variable/10 +1/100,
and target*variable/10 is negligibly small, so
Digital stream = target + const.

Obviously the variable component of Tx and the resulting air signal is lost. With it the 1/f chatters are gone, and you are one step closer to a working static response VLF.
I'll explain the variant No.2 tomorrow.

[Davor]

Now the 3rd option, a rock solid Tx.
In Rx front end there are a target response and also superimposed to it is "air signal" mix due to the imperfect balancing in a coil. This Air signal with a free running oscillator as Tx contains a cyclostationary component of a portion of Tx carrier that may be assumed to have a constant amplitude. On top of it there are anomalies: AM random walk noise at about -80dBc following 1/f law, and volatile harmonics at about -40dBc (at best).
I mentioned that a minimum target response that might be detected would be about 5nV due to the bandwidth of ~10Hz and input noise of a good preamp. But because of the -80dBc 1/f noise and intermodulation with the volatile harmonics, the noise floor is dominated by chatters caused by Air signal's imperfections, and not so much by thermal noise. Therefore in a classic detector the linearity of input stage is more important than its low noise. You may forget about the theoretical ~5nV sensitivity.
If a free running oscillator is replaced by a square wave one, excited by a symmetrical digital source, and supplied from a good low noise PSU, the picture above changes drastically. First, the amplitude of 1/f component drops to the value of a PSU, or even better - a battery. Next, the harmonics stop being volatile, and may be forgotten about.
So if a slow motion response is sufficient, a square wave Tx would do, with exception that phase shifters/comparators won't work well.

[Davor]

And the 2nd option for the end:
"The other approach would compensate the air signal by a scaled down sample of Tx envelope. So instead of a compensation using a reference voltage, which would effectively remove only the unchanging bulk of the "air signal", you'd compensate using a voltage that follows the very "air signal" change, and in effect lose the majority of chatters caused by it. The compensation would be adjusted with preferably digital potentiometer. The resulting target signal should be amplified with aDC amplifier, say CDS or chopper. Sampling demodulator is in itself a CDS of a kind."

No math this time. This option would be in effect a sort of an electronic balancing of a coil in a demodulated baseband. It works by subtracting the unwanted signal from the mix. Assuming there are I and Q channels after demodulation, and also that Air signal may produce different response in I and Q channels due to different coils behaviour, each channel has to be compensated independently using the portion of Tx envelope voltage. After proper compensation all offending components are diminished, just as if you have an ideal coil. To achieve that you'd need 2 independent compensation devices that are not prone to drift, say digital potentiometers. Once adjusted, and with hope you don't damage your coil too much after that, the setup is good for a given coil. Complete removal of a DC component for static operation will still be necessary, but the random walk 1/f noise is significantly reduced. Disadvantage is that you need 2 sets of compensating pots that must be adjusted independently.
For a given coil it may be feasible to have this compensation done manually using multi-turn pots, and a whole system may be analogue. Mind you, digital pots may be tuned by up/down action with no need for a MCU.

[Carl-NC]

In short, you can't expect to have a truly static response if you don't compensate for "air signal" variation. There are 2 possible paths. One is using a Tx envelope voltage as AD converter reference, and effectively get a 1/(air signal) multiplication. In effect it would act as if your Tx is rock solid, and there is no "air signal" variation at all. In digital processing the Dc component of such rock solid offset is simply deducted, and all that remains is a target response.

I'm not quite sure what you are calling the "air signal." There is the null error in the loop which creates DC offsets in the demods and is normally removed by the HPFs. Next, there is ground-induced TX amplitude variation due to changing permeability which generally creates small variations in the null-induced DC offsets. There have been designs that attempt to compensate for this in the way you mentioned: use the TX amplitude as a reference for the ADC. I'm not sure how effective this has been; my limited testing showed it didn't do much at all. Finally, there is 1/f noise which creates random phase jitter and shows up as just a chatty detector. The only cure for 1/f noise is to make a better transmitter. Did I miss anything, and do all these make up what you call the air signal?

In a motion detector both constant and variable components of the air signal hit a high-pass filter, after which the constant portion of 10mV is eliminated, while the variable portion is reduced. A true static operation is nearly impossible due to this variable portion of air signal, and even a motion filter struggles with it. Because it may have any phase at the coil, this component is not easily removed with a GEB.

It would be possible to use a GB-like mechanism that is phase-adjusted to the coil's null error, then ground-induced null variations would effectively be silenced. But raw ground phase would not. However, you then use a classic X-R subtractive null to suppress the ground response. You end up with an all-metal signal that is free of coil null noise and ground noise.

If a free running oscillator is replaced by a square wave one, excited by a symmetrical digital source, and supplied from a good low noise PSU, the picture above changes drastically. First, the amplitude of 1/f component drops to the value of a PSU, or even better - a battery. Next, the harmonics stop being volatile, and may be forgotten about.
So if a slow motion response is sufficient, a square wave Tx would do, with exception that phase shifters/comparators won't work well.

All of my current development work is with square-wave drive. Assuming time-domain sampling, you no longer need phase shifters & comparators as the demod clocks can come directly from the micro.

[Davor]

Yes, last time around the signal due to the misbalance of a coil was "air signal" and is entirely correlated to a Tx, although phase shifted.
The phase shifter remark was for the constructors that may wish to marry a square wave Tx with a classic design of Rx. It is far from optimal.

[Davor]

I also thought of a GB-like mechanism that would provide an independent all-metal response, but there's a problem. Ground response is independent from the "air signal" making the phase go all over the place, and thus there is not a simple way to get rid of it. But the "air signal" portion may be eliminated by the counter-offset using Tx envelope.
Having a perfect Tx fixes most of the problems, but there is a case for independent I and Q compensation anyway.

[ivconic]