Coil nulling - again..

[dfbowers]

We absolutely beat this topic to death a few years ago, and to be honest, I don't have a complete understanding of the fine details yet. What I would like to do is throw out there is a discussion on why manufacturers null their coils the way that they do.

The specific coil I am working with now is my 4th concentric - this one is a 10" for the IDX project. To this point, they all have worked wonderfully.

The way I make my coils, I basically "copycat" one of my coils for the White’s Spectrum - frequency, phasing, nulling, etc. My commercially made 9.5" coil is nulled with the Rx coil about -90 degrees (leading the Tx signal) to around 100mV P-P. (Performance gets copied but not the cost ; )

So we have several ways to null a coil- either Rx leading or Rx lagging the Tx signal or the deepest null... and heck if I can find any difference at all !! I can prove it to myself over and over again. GB, DISC and sensitivity all seem be exactly the same no matter. The reason that there seems to be no difference between nulling either way is that the "sample" into the demodulators gets flipped on the opposite side of a null and so does the phase shift direction, (relative to the Tx signal) hence the overall response through the demodulators is the same.
Case in point - Some Nautilus detectors have an adjustment to manually feedback phase and/or amplitude from the Tx signal to the Rx signal to achieve a good null. No difference there either.
The only odd thing that happens while nulling to one of the sides is that large metal objects cause the Rx phasing to "roll" from one side to the other though the null point. (coil rollover?)
So the question becomes this - If no difference can be found while bench testing, why did White's choose -90 degrees to null?
Don

[Davor]

heck if I can find any difference at all !!

That' because the so called "air signal" or better say unbalanced Tx portion is regarded to as static offset in Rx and completely eliminated by motion filters. It may be phased front, back, or sideways, but in all cases it is eliminated further on and only the dynamic components are processed.

Your best bet is to null it in it's working conditions, in a tank or otherwise, with a preamp and it's input impedance, and go for the dip. As lower as better. Less "air signal" - more sensitivity.

Why companies do whatever they do to tune the balance I can't tell, but I'm sure they'll not spend a whole afternoon tuning a single coil to make it perfect, and instead they'll go for some key parameter found by someone in R&D that makes a coil balanced to their specification. As you can't make them perfect that's kinda good enough.

[dfbowers]

Thanks Davor!

Don

[golfnut]

HI Don, on your whites coil ..

Did whites tune for an amplitude ?? - and you just get the phase you get when operative gets the volts below a level..??




My guess... long shot..
On a lot of our deigns here the Tx and Rx are resonant - this is special case, the voltage across the tank is in phase with the current flow in the tank - as tanks does not look like an L or C at res. So measuring the Rx volts rel to Tx volts is good measure of line up of Rx signal volts vs Tx signal current / mag field phase.

If the original whites unit Tx was not resonant - but rather a pulsed coil from a divided clock, then voltage across it and the current running thru it would have been 90 deg split - the current lagging the excitation volts by 90.

If true, in this case the Rx coil phasing would need to have a 90 deg lag from Tx signal volts - in order for the Tx signal Mag component to align with the Rx signal volts.

Steve

[golfnut]

A down side to a 'bad null' is the mixer / product det is then partially compressed with the Tx signal. But, tens of mV is ok as we all observe, whereas 1/2V or so, is a real problem.

At the mix / det output the coupled/leaked tx signal will be ~20dB down on real signals, (plus there is an RC low pass at about 80Hz corner on most det o/p) there will however be a small dc offset from this mix onto channel - in PP mode this will reduce sense a little.

In Disc mode the DC does'nt get thru beyond post mix filter, because of the caps - so, little degradation.

There could tiny advantage having dc block on the output of the mix / det to stop DC on input the post mix filter.

S

[dfbowers]

I think may of us have different projects in our heads as we ponder this concept, but in general, if I construct a coil and signals never saturate, phase and amplitude change in the correct directions and the GB sample can be centered in the ground signal, I have a perfectly usable coil. It's all the other fine details that make some coils better than others that make me think about nulling.

[Skippy]

I agree that a WELL-nulled coil should work fine, the details of residual phase don't matter. However for commercial mass-manufacture, they want the slackest limits defined, for cost reasons. I think you will find the quadrature demodulated signal will have more gain than the Zero degrees channel, so it is more prone to saturation/compression. This puts different limits on the null - the allowable amplitude is phase dependant. Also, the quad component can only go in one direction when ground/iron/non-iron are near the coil, so in practice it may be acceptable to bias the null in a negative sense, knowing that real world conditions will only increase the signal, hope that makes sense.

[Davor]

A down side to a 'bad null' is the mixer / product det is then partially compressed with the Tx signal. But, tens of mV is ok as we all observe, whereas 1/2V or so, is a real problem.

At the mix / det output the coupled/leaked tx signal will be ~20dB down on real signals, (plus there is an RC low pass at about 80Hz corner on most det o/p) there will however be a small dc offset from this mix onto channel - in PP mode this will reduce sense a little.

In Disc mode the DC does'nt get thru beyond post mix filter, because of the caps - so, little degradation.

There could tiny advantage having dc block on the output of the mix / det to stop DC on input the post mix filter.

S

You are mostly right, and I'll add only a few corrections...

Product detectors are the least of your problems because they can cope with signal swing to near the rails without much problems. The most vulnerable to saturation is the preamp because it starts intermodulating at much lower levels. The crucial parameter is a THD ... you are supposed to amplify signals at below 1uV with "air signal" at few mV. Anything more than that and THD kicks in.

The DC offset is not small, unless it is seriously compensated. Even then you have TX AM noise that introduces a dynamic component that can't be compensated, but it is at the ferrites' response phase and goes away with GB ... provided you are dealing with easy grounds. TX's 2nd harmonic is a cheerful li'l bugger that may assume whatever phase it wants and it also rides on an "air signal". It affects the unbalanced product detectors, and it is produced by unbalanced TXes.

The best case scenario would include:
- the least possible "air signal" as a better coil balance also reduce the AM noise and 2nd harmonic content
- balanced Tx with well stabilised supply
- balanced product detectors to reduce 2nd harmonic

I'm attaching a picture indicating the importance of a good coil balance. You'll notice that poor balance is also responsible for fuzzy discrimination as it shifts the targets' phase response left and right.

[Gwil]

Hi Steve

I'm not sure if I understand what you said in post #4. Do you mean that at resonance the current in the coil is in phase with the voltage across it? Because I'm pretty sure that if its Q isn't too low the current must lag the voltage by about 90 degrees. That's what happens with inductors, regardless of what other components happen to be connected across them. The tank circuit as a whole looks like a resistance, but that's another matter.

Gwil

[golfnut]

Hi Gwill, sure if you consider an L the volt and I are 90 split similarly for C.


The point I was after was this.. "the voltage across the tank is in phase with the current flow into the tank" .
So applied driving voltage from transistor (which we scope as a phase reference) is in phase with the current flow into the tank.

Reactances of the L and C are the same magnitude at resonance - they differ only in sign. So at resonance XL and XC are not there -when looking across the tank . Only residual R remains - current flow through this residual R is in phase with the applied voltage.


with a parallel tank,
much below res, the L is dominant in the tank and the I into the tank lags the applied volts by 90
much above res, the C is dominant in the tank and the I into the tank C leads the applied volts by 90
at res is above case.

Steve

[Gwil]

Can't argue with that. And the magnetic field is in phase with the L current.

G