IGSL

[Davor]

These effects are related to your sweeping speed in the same manner targets do, and while the soil response is phase shifted and reduced by properly adjusting GB, the unbalanced Tx signal modulation passes through. Point is that even due to some voltage fluctuation in the rig, this will be happily passed through to the audio, and the same thing happens with the ground proximity changing the coil Q. As a direct consequence you have varying 2nd harmonic fluctuation that directly causes offset shift PWM style, and it all passes through the motion filter as well. As a consequence you have a rig that behaves as if you can't find a proper GB anywhere. It is not severe, but I noticed it. It can be partly cured by finding the Tx oscillator bias Goldilocks point. Thanks Ivconic for a Tx trimmer - some VLFs don't have any.

In conclusion, this is not a problem for a deaf rig. Going any deeper and it catches up on you. First thing to do is going full wave, and second is running a Tx with suppressed 2nd harmonic, e.g. a symmetric one.

[simonbaker]

Just imagine how changing ground proximity would influence Tx coil Q and the 2nd harmonic content, and next just look at the picture above and see how 0V is not placed at a value that stands for equal off and on times.

There are two effects not to be confused. The soil can modulate the amplitude of the TX signal and it can also shift the average DC level due to the distortion of the waveform you described.

First, when the soil modulates the amplitude of the TX signal, we expect the SD to register an amplitude change, so any small correlated shift of the DC average value of the RX signal at the same time isn't big deal, it just gets combined into the overall change.

Second, let's assume the soil can cause amplitude change of the TX signal as you sweep for targets. The amplitude change can cause distortion which can alter the harmonics, which are multiples of the TX frequency, but I will assume as you say that it can also shift the DC level of the signal -- but typically at some frequency which is not a multiple of the TX frequency -- for example, the frequency at which the soil changes while sweeping for targets.

So if the soil is able to shift the DC average level by a change in the waveform shape, the equivalent additive signal is typically at a very low frequency and you should not see it in the RX signal; it would be attenuated going across the tuned transformer comprising the TX/RX coils and target.

Any second harmonic (I assume that means twice the frequency) of the TX frequency should act the same as the fundamental, only smaller, in a half-wave synchronous detector, should it not? In other words, if the TX signal is integrated to zero by the SD, then the second harmonic would integrate to zero. If the TX fundamental integrates to some voltage (due to phase), then the second harmonic would also integrate to fraction of that value -- it would behave like a small percentage addition to the fundamental.

So while I see the point of your picture, you haven't shown how that causes a problem when you follow it through from TX to target to RX to SD. I'm not saying it doesn't cause a problem, I'd just like to see more clearly how.

-SB

[Davor]

Just imagine how changing ground proximity would influence Tx coil Q and the 2nd harmonic content, and next just look at the picture above and see how 0V is not placed at a value that stands for equal off and on times.

There is a solution that uses a PLL to fix unequal timing, but I think it will require somewhat faster loop filter than it is usually set to account for the additional phase noise. I gave it a further thought and it makes sense.

Anyway, full wave is the way to go because it fixes many problems.

Imperfect timing is a different ballgame but improving it may be considered very desirable. I think I may have a trick up my sleeve, but I won't voice it until it is ripe for publishing here.

[simonbaker]

2nd harmonic is a source of PWM here. You'll see it in a picture below. Problem rises with offset, and there are many sources of offset, including various modulations. The way full wave switcher fixes things is by successive reversing the offset contribution, and thus suppressing it. In fact full wave switchers also suppress the carrier as well - they are balanced mixers.

All of that, of course, in case switchers are supplied with symmetrical timing. Full wave will suppress most of the artifacts even with not-so-perfect timing, but the half wave switchers will not.

Our rigs are usually made with single ended oscillators (which are fine sources of 2nd harmonic), and phase shifted AC component is compared with ground to obtain the switchers timing. Now, consider waving a coil against the rough surface and think of it as a variable loading - guess how it reflects on offset in PWM terms with changing levels of 2nd harmonic in amplitude and phase (due to the phase shifting).

We may argue that these effects are of second order, but there is a large unbalanced signal that is supposed to stop at motion compensation filter, and it's modulation will pass through. Microvolts of target signal on one side against milivolts of unbalanced Tx on other.

Thanks -- I understand it to be an "offset" problem. In other words, we want to detect amplitude changes in the RX signal, but not "offset" changes. "Half-wave" SD designs are more sensitive to "offset" modulation.

So the next question is: where do these offsets come from and how big a problem are they?

I believe "offets" arise when there is an "additive" component to the TX signal that shifts the oscillator signal up or down without changing its amplitude.

It seems we would be most worried about offsets with a frequency near the sweep frequency of our target signal (10 Hz). I would not think additive signals to the TX oscillator voltage at 10Hz would make much impact on the RX signal due to the second order RX tank (RX coil & RX capacitor) which peaks around 17 KHz and should greatly suppress components down near the 10Hz region.

So I'm still not sure it's a huge problem here theoretically, but what happens in reality is what matters. In any case, I'm all in favor of full-wave SD for designs where some extra circuit complexity is acceptable.

-SB

[Davor]

2nd harmonic is a source of PWM here. You'll see it in a picture below. Problem rises with offset, and there are many sources of offset, including various modulations. The way full wave switcher fixes things is by successive reversing the offset contribution, and thus suppressing it. In fact full wave switchers also suppress the carrier as well - they are balanced mixers.

All of that, of course, in case switchers are supplied with symmetrical timing. Full wave will suppress most of the artifacts even with not-so-perfect timing, but the half wave switchers will not.

Our rigs are usually made with single ended oscillators (which are fine sources of 2nd harmonic), and phase shifted AC component is compared with ground to obtain the switchers timing. Now, consider waving a coil against the rough surface and think of it as a variable loading - guess how it reflects on offset in PWM terms with changing levels of 2nd harmonic in amplitude and phase (due to the phase shifting).

We may argue that these effects are of second order, but there is a large unbalanced signal that is supposed to stop at motion compensation filter, and it's modulation will pass through. Microvolts of target signal on one side against milivolts of unbalanced Tx on other.

[simonbaker]

Oh yes, it does. There is always some residual unbalanced signal from the coil on top of which you'll find the targets' response. Any floating there will be passed forward as an amplitude change in case of half wave, but not with full wave. By floating I mean EF, 50Hz, internal voltage change ... you name it. Most of these are cancelled out by the full wave switching, and THAT is the main benefit.

I don't see how full wave detecting fixes that, because it just acts as a full-wave rectifier instead of a half-wave rectifier. Amplitude modulation of the TX signal will still come through whether it is full or half wave detected.

Even harmonics are always canceled and don't contribute to the SD output. Odd harmonics contribute to the signal, regardless of half-wave or full-wave detection.

So I don't get the idea yet, maybe a detailed example would help.

-SB

[satdaveuk]

So some machines only see half the RX signal' In a ideal situation by seeing all the RX signal theres lots more to be gained but not 50% worth, nearer 10% if your lucky but by getting that your putting another 100% worth of guts into the machine together with the full quad demodulator which gives you these prices of £1500.
Its all horse for courses, lots put in little gained, but of course that extra 10% can be wether you get the hit or not so worth it to some.
Very interesting.

[golfnut]

I guess there are the three mixer styles


in approx order of performance...



Full Quad demod - proper good machines. Tier one GMP etc £800 - £1500
Balanced type with alternate 1/2 cycle sampling - eg Cibola. £350 - £800
The single ended type like a single cmos switch used in IGSL/TGSL/IDX £200 - £350



S

[Davor]

Oh yes, it does. There is always some residual unbalanced signal from the coil on top of which you'll find the targets' response. Any floating there will be passed forward as an amplitude change in case of half wave, but not with full wave. By floating I mean EF, 50Hz, internal voltage change ... you name it. Most of these are cancelled out by the full wave switching, and THAT is the main benefit.

[simonbaker]

There are better reasons for full wave sampling than twice the samples, like 2nd harmonic suppression, common mode suppression, and offset suppression. Please note that PWM induced fluctuations resulting from 2nd harmonic content are largely ignored in most of the designs, including my beloved IGSL.

If the PWM does not shift the phase of the signal, does it really matter?

Regards,

-SB

[Davor]

There are better reasons for full wave sampling than twice the samples, like 2nd harmonic suppression, common mode suppression, and offset suppression. Please note that PWM induced fluctuations resulting from 2nd harmonic content are largely ignored in most of the designs, including my beloved IGSL.

[golfnut]

Talking of a better sych det to give better depth - like a modern mid price machine. Pls read attached.. S

[Davor]

I almost forgot, yesterday I bought a pair of rechargeable 9V cubes from Lidl. At 7.5 Eur for a pair - unbeatable! They are rated 200mAh and I filled them for the first time already. I hope to make the 9V mod this weekend or the next week.

[Davor]

I think the only sensible purpose for the other GEB channel would be an all metal channel with slower motion response. However, I strongly believe that lowering the high pass response in the rest of the channels would do the very same job but better because of the discrimination. Other than for the increased 1/f noise - there is no reason not to lower the high pass cutoff. So instead of 4u7 to go for 15u or something. That would give you the best of both worlds, and without additional knobology. In such case you may completely abandon the other GEB channel and embrace full wave switching instead.

As you can see I did not make changes to the very frontend (yet) and I intend to maintain the FKK coils as a project on it's own. I can't expect commercial coils to adopt this approach any time soon, and many people are not into building their own coils. So shielded coils frontend there is.

BTW, for shielded coils it is better noise-wise to ground the inverting input side of the coil as you suggested once. It has no common mode suppression then, but you rely on the shield and there you have it.

[ivconic]

I am buried deep in various commitments lately (autumn... what to say?) so i delayed my further work on IGSL at least till winter.
However; i am collecting all the modifications done here and intend to apply them on my IGSL's i have here.
IGSL is not finished story, it is half way there.
So i am glad that Davor is showing good will to continue improving it.
That job is not for nothing, because i have serious intentions for later, to transfer IGSL into "digital" world.
So once we get most optimized IGSL setup; next step shell be "digitalization".
Not so soon. Maybe next year.
So all you done here so far is not wasted time.
Later will be easy to apply adds on existing setup and continue with quite decent machine.
Davor don't throw away second GEB channel because we gonna need it.
Cheers!

P.S.
For Christ's sake DO apply shield on your coil!