GEB on IGSL/IDX

[golfnut]

It is easy to get off topic.


One of the things I was trying to say is - integrating out Iron does loose some target signal - as in the diag.

But the follow on is - all of the machines like these do this. The synch detector has ability to output plus+ and minus- signals -


If the detector was unipolar - only positve going signals would "get out from it" along into the filter/ amp and comparator chain.


The advantage would be that the -ve going Iron signals would not erode the +ve signal from a target alloy - rather leave it alone to be clean large amplitude real targets unaffected by -ve going part of the Iron signal. Very Sharp Disc would result with more Target signal sens .

As in the drwg, Steel clad coins often do not detect exactly for the above reason - Steel and nickel signals approx equal in magnitude and sign resulting in almost nothing.

[golfnut]

Like this

[simonbaker]

OK, slow down a little bit. To remain accurate we must establish some nomenclature here, otherwise this is going to be a mess.
Simonbaker noticed correctly that sampling a target signal with two phases produces two different signals at lower amplitude (if I understood it correctly), and that is mostly correct. When you multiply (sample) some signal by two sampling signals that are at 90° e.g. in perfect quadrature, you'll get products that also have orthogonal properties. The funniest property of all is that noise is reduced by the same rate as the signal and noises from different channels are mutually orthogonal, and that is a good thing. So in fact you do have conversion losses, but your noise also has "conversion losses" so your S/N is not ruined. Ability to discern small signals comes from S/N so that's good news.

Regarding integration ... it depends mostly on the build. TGSL/IGSL have severe filtration here, but IDX doesn't. Thing to remember is that target response happens simultaneously in all channels, but may reverse phase and amplitude in a process. So copper coin will have positive response simultaneously in both Disc and GEB channels. In case we have an incompetent design ... which are practically all of them unfortunately, you have level dependent durations of channel response, and these lead to utter confusion. It is very easy to solve: just remove capacitors in feedbacks of the last channel amplification stages and you are saved. Severe integration in subsequent stages just helps sorting out the dominant signal, and reducing the "chatter" that is a direct consequence of design problems.

So first and the most important - target feedback signal happens simultaneously in all channels, and variations in duration are merely a problem in design that is easily solved. Ideally these would happen simultaneously, but different levels of saturation take care that they don't. Excessive integration to avoid these problems :duh: is yet another design problem introduced to fix the initial one. As a consequence of these two design problems -- you have masking.

Yes, I believe you understood my basic points Davor, which is related to the well-known "quadrature" effect, and is in fact how discrimination actually works -- adjusting the sync pulse phase to diminish (or reverse) the demodulated signal from some metals while boosting the signal from the metals we want to detect. The key point I was making is that only one target phase will be "optimal", which is the one that matches the sync pulse phase. Practically speaking, there is a reasonable range of phases that are near-optimal. And for a two channel design, we can pretty much guarantee that one or both channels will not be optimal, and therefore detection sensitivity is never optimal (except possibly in ALL METAL mode, where both channels could be the same).

I think Qiaozhi was interpreting my statements as a demodulated signal timing concern, where we might worry that possibly the target envelope responses could be offset somehow and not act in unison, and I agree with him that the phase of the sync pulse does not directly affect that, and I did not intend to suggest that. Davor states it as "target feedback signal happens simultaneously in all channels". If I understand Davor correctly, he then notes that the dynamics of the amp/filter stages can exaggerate target signal timing and pulse width differences in the two channels due to different amplitudes or additive noise, and so he has made mods (feedback diodes, filter order) that he believes helps reduce that problem without losing other benefits.

One statement by Davor does not seem correct to me. He says:

The funniest property of all is that noise is reduced by the same rate as the signal and noises from different channels are mutually orthogonal, and that is a good thing.

Perhaps I misunderstand his point. I don't see any reason to believe that noise (EMI or circuit) is reduced similarly to a target signal when the sync pulse is "off center" relative to the target signal. I assume that noise is uncorrelated to sync pulse phase, so as you change the sync pulse phase (by turning the DISC or GEB pot), the demodulated noise signal should stay the same RMS value, and the RMS noise level should be the same in both quadrature channels, although the actual signals are not identical. Only the target signal should be affected, so we do lose S/N when the sync pulse phase does not match a particular target metal phase.

-SB

[simonbaker]

It is easy to get off topic.


One of the things I was trying to say is - integrating out Iron does loose some target signal - as in the diag.

But the follow on is - all of the machines like these do this. The synch detector has ability to output plus+ and minus- signals -


If the detector was unipolar - only positve going signals would "get out from it" along into the filter/ amp and comparator chain.


The advantage would be that the -ve going Iron signals would not erode the +ve signal from a target alloy - rather leave it alone to be clean large amplitude real targets unaffected by -ve going part of the Iron signal. Very Sharp Disc would result with more Target signal sens .

As in the drwg, Steel clad coins often do not detect exactly for the above reason - Steel and nickel signals approx equal in magnitude and sign resulting in almost nothing.

Golfnut -- I think that might be an interesting idea, I apologize if I didn't read your previous posts carefully enough. I will have to think about that. You are saying with a "rectifier" scheme, we can cause the demodulated response all metals below the DISC threshold to be exactly zero and not go negative and therefore avoid some "masking" effects of two targets very close to each other.

First, I don't think your picture is accurate as to what is going on though. You are showing the raw target signals as 180 deg opposite, but that is not how the target phases appear or how the target "signals" cancel. Your example would indeed produce masking, but actual metals are not 180 deg phase difference.

The masking problem actually occurs when you position the DISC "phase gate" exactly half-way between the phases of any two metals -- it doesn't matter how far apart they are. The combined phase of the two targets will be a signal that exactly has zero output from the Synchronous Detector. (I should be more accurate and say that masking will occur when the DISC "phase gate" is positioned so the combined phase of the targets causes a zero or negative signal.)

What I just said assumes a theory that is probably not quite accurate. I assumed that two target signals [of equal amplitude] combine to create a signal with the average of their two phases. That may not be accurate, in fact I tried to prove that with an experiment and got different results. But assuming that theory, that is how the target signal would be "masked" by another target -- it depends on where you set your DISC control.

So even if you modified your Synchronous Detector to never output a "negative" signal, the combining of the phases of two targets as they are received would still create a "masking" effect, depending on where the DISC control is set, I think. But your ideas deserves careful consideration, that is just my first impression.

To be fair, I changed the interpretation of your picture to be the output of the SD rather than the raw signals. You perhaps are talking about rectifying the actual raw signals. If you actually did that, I think the entire operation of the SD is defeated, but it requires a more complicated explanation to see the results.

Anyway, very interesting idea to think about.

-SB

[simonbaker]

One statement by Davor does not seem correct to me. He says:

Perhaps I misunderstand his point. I don't see any reason to believe that noise (EMI or circuit) is reduced similarly to a target signal when the sync pulse is "off center" relative to the target signal. I assume that noise is uncorrelated to sync pulse phase, so as you change the sync pulse phase (by turning the DISC or GEB pot), the demodulated noise signal should stay the same RMS value, and the RMS noise level should be the same in both quadrature channels, although the actual signals are not identical. Only the target signal should be affected, so we do lose S/N when the sync pulse phase does not match a particular target metal phase.

-SB

To add to my own message - I think now Davor was referring to the difference between correlated and uncorrelated noise, which may help some when two channels are 90 deg in quadrature as opposed to same phase. But still does not address fact that there exists an "optimal" sync pulse phase for any metal that makes the best S/N in a particular channel, and that for two channels in quadrature, any metals that are near 90 deg phase with either channel sync pulse will have a poor S/N for overall detection.

-SB

[Davor]

Exactly. You should pay attention to the demodulated signals in different channels - they are in 0° /180° relationship. Targets appear, and when they do all channels react simultaneously. Some with zero response, some at 0°, some 180°, but everything in between is just a consequence of unfortunate disregard to the saturation problems ... quite common with all nowadays VLFs.

[golfnut]

better pic

Like this, Yes Simon I mean a precision rectifier or just a det which has no ability to ouput signals going thru midpoint in the -ve section. So the large bounce back from Iron does not reduce gold sens for example.
I know a guy who got iron and dug it and walked on, the guy behind hi pulled a stater out of the same hole 30 seconds later - the iron masked the gold. Only iron sounded.

With a goldmaxx power actually.

See sketch..
S

[Davor]

Much easier way of achieving this is the way I did that in my IGSL. You are absolutely right about problems with negative going large response. The positive one is clipped by the limiting diode(s), but the negative is allowed to rock bottom the op amp, which struggles for some time to recover. Right there lays your masked gold.
The easiest way to fix this is just placing an antiparallel diode with the existing one(s) so that negative going signal receives the same spa treatment as the positive going one. Because of the changing impedance due to the different level of saturation, the capacitors in the compression stage makes no sense because it causes falses and chattering, AND promotes masking.

[Aziz]

Hi guys,

you can't prevent the iron masking effect of the non-iron targets with a single frequency VLF system. To separate X (reactive) and R (resistive) response from response signal S, you need at least two different frequencies.

Even though, an iron target is producing a resistive R response too and you can't distinguish it from a non-iron target on hot mineralized soil, which is producing a large X signal.

Why don't you guys go for the dual frequency VLF system? The math is quite simple to get X and R from S and to look for the relation of R to X, which delivers you a good discrimination capability.

Cheers,
Aziz

[simonbaker]

Hah, half the time we're all talking about different things -- but all interesting points.

Davor's latest point is on the dynamics of the filter stage, where one large response can extend over another smaller one, when the amplitude extends the duration of the pulse (I think). This becomes a problem when two targets are fairly close together but not virtually the same location -- so the signals are not combining as one. I believe he feels faster response with amplitude clipping can help recovery time and provide better "resolution" for separating targets spatially.

But I think golfnut wants to detect gold that is virtually on top of or touching a piece of iron by somehow ignoring the iron. He suggests a "rectifier" on the received waveform(s) might do that.

I still cannot make sense of his his picture, because it seems he is drawing the actual target and trash "signals" that are at the same frequency as the TX frequency. There are two points here: 1) you cannot think of them as two signals -- they combine into a single sine wave signal, which you could then rectify, but to what purpose? -- you cannot just rectify one of them; 2) Fe (iron) and gold, for example, are not 180 deg apart in phase -- that is not how "masking" occurs.

Masking occurs mainly because the two signals "combine" into a single signal with a different phase -- you can look in a math book to see how Asin(t) + Bsin(t + phi) combine into a single sine wave Csin(t + omega). That is because magnetic/electic fields superimpose and the RX coil can only see the resultant combination. And then masking occurs if your DISC control is set so that combined phase causes a zero or negative signal at the Synchronous Detector output.

But there is a rare case where you can succeed:

If you know for sure that your trash always has a single phase (not realistic -- different size objects have different phases), then you can avoid masking by setting your DISC control to make the trash (Fe) have exactly zero signal at the output of the SD (not at the RX coil). If you do this, then any valuable target like gold will always make the combined Fe + gold signal have a positive signal at the output of the SD (synchronous detector), and so your target is not masked.

So does that solve golfnut's problem? Not quite, because 1) maybe he wants to discriminate out nickel, or foil, etc -- too bad, the DISC pot is dedicated to canceling Fe (iron), so all other metals will make a beep; and 2) iron trash does not all have the same phase, so some of the pieces with less phase shift will make the SD output signal go negative and can indeed mask the little piece of gold.

So unless all your trash consists of the same item (non-overlapping iron poker chips), some masking will possibly occur.

But it is still worth checking what happens if we rectify the combined signal of an iron and gold target virtually at same location. My intuition says it won't provide any better way to "unmask" the gold target than what I mentioned previously, but the math would be interesting. I think basically rectifying the signal creates a new signal with a fundamental frequency of twice the TX frequency and the synchronous detector would... need to draw some pictures.

I think Aziz is correct that you can't unmask trash in general with a single frequency, because two touching targets can truly look like a totally different target to a single frequency. To do it with multiple frequencies requires that no single metal can give responses to multiple frequencies that look the same as the combined responses of two different metals. Sounds complicated too.

-SB

[golfnut]

Yes, Simon - drawing the Tx freq the same was daft - it served as a phase reference for the other signals and their phase relation to the Tx.




I still belive If the Det only had the ability to output positive going mix results we would not suffer with Iron so much - we would be able to detect smaller valid alloys in the presence of Fe


The negative going sections do not contribute to Target signal, even in absense of Fe - They subtract from target Amplitude - as, the 2 'bounce backs are -Ve the nett positive accumulated to trip the comparator is less due to these. We dont ned them.

In the Fe plus Target case the second bounce back from the large iron often is larger than the small wanted alloy response and the nett result is a negative going signal which as we all know hold off the comp - No tone Target is missed.



Nuff said, S

[Aziz]

Hi guys,

don't forget the following superimposing effects:
1. Induction balance gets out of balance due to magnetic susceptible materials (TX -> RX induction)
2. Magnetic relaxation decay of magnetic susceptible materials, which is seen by the RX coil (TX magnetising susceptible matter -> relaxation -> RX induction)
3. Target eddy current induction -> RX coil induction
All effects happening at the same time.

- An iron target may cause all of the three effects in different amount of magnitude depending on the size, shape, orientation, distance...and so on..

- Mineralized ground may heavily/significant affect 1. + 2.
- Conducting ground may additionally affect 3.

Fortunately, the magnetic susceptible matter is behaving different to the eddy current induction and one can identify the reactive part (X) of the signal (S) and one can focus to the resistive signal (R). Once you have separated the X and R, it is still difficult to discriminate.
Is the X part caused by the varying ground or by the target?
That's the crucial point.

Single frequency discrimination works only, if your ground condition does not change and you have a mild ground (less X). But reactive signal causing targets can mask out the non-iron targets when they are close together.

If you want to come close to the "World's best ground balancing and discriminination technology", you should focus to a wide band or at least dual frequency detector. The dual frequency detector has an advantage over a wide band: lower noise due to narrow band demodulation (high Q).

Cheers,
Aziz

[Davor]

I guess the most of the "theoretical" effects happen somewhere else, but not on red clay salt soaked thrash filled beaches
I'll have to seek some more difficult terrain

Just joking.

IMHO your ears and the rest of your senses are your best friends in detecting ... just about anything. Complex systems tend to synthesize their responses in a non-natural way, and in such cases your ears are just an accessory, not friends. My kinda goal is developing and becoming friends with rigs that provide me with so called "immersion". Anyway, I'm quite happy with my rig as it is now. It provides.

[simonbaker]

Yes, Simon - drawing the Tx freq the same was daft - it served as a phase reference for the other signals and their phase relation to the Tx.




I still belive If the Det only had the ability to output positive going mix results we would not suffer with Iron so much - we would be able to detect smaller valid alloys in the presence of Fe


The negative going sections do not contribute to Target signal, even in absense of Fe - They subtract from target Amplitude - as, the 2 'bounce backs are -Ve the nett positive accumulated to trip the comparator is less due to these. We dont ned them.

In the Fe plus Target case the second bounce back from the large iron often is larger than the small wanted alloy response and the nett result is a negative going signal which as we all know hold off the comp - No tone Target is missed.



Nuff said, S

It's an interesting subject to me, that's why I went on too long about it. And your idea is interesting too, so I'm going to think about what happens if we rectify the output of the Synchronous Detector and keep it in mind.

Cheers,

-SB

[simonbaker]

I guess the most of the "theoretical" effects happen somewhere else, but not on red clay salt soaked thrash filled beaches
I'll have to seek some more difficult terrain

Just joking.

IMHO your ears and the rest of your senses are your best friends in detecting ... just about anything. Complex systems tend to synthesize their responses in a non-natural way, and in such cases your ears are just an accessory, not friends. My kinda goal is developing and becoming friends with rigs that provide me with so called "immersion". Anyway, I'm quite happy with my rig as it is now. It provides.