Minelab GPZ 7000

[deemon]

So, it is as I suspected, and this technique is not designed to work with short decay constant targets.

Hm ... why not ? The key parameter here is a flyback pulse duration . So if you need , for example , to feel a target with 1uS TC - just make the flyback pulse about 1 uS and enjoy . Less turns in the coil ( made of a good Litz wire ) , less capacitance in parallel , pump the current and frequency ( up to 100 kHz , for instance , or even more ) , use a very fast transistors - and we'll get a good device for very thin targets .... but of course , if the target TC is equal to the salty water TC - we cannot separate them using TC as a criterion . And another tip , comparing to the classic PI technology - increasing the main sample delay ( in classic PI ) is equal to increasing the flyback pulse duration ( in square wave PI ) .

[Davor]

Thanks Davor all book marked except this one, its no longer found.
A Fast 4-D TEM System for UXO Characterization
http://www.estcp.org/documents/techdocs/UX_0105.pdf

You may find it here: http://www.dtic.mil/get-tr-doc/pdf?AD=ADA442592

[Qiaozhi]

This question is interesting and must be observed more thoroughly ....

Thanks for the detailed explanation. So, it is as I suspected, and this technique is not designed to work with short decay constant targets.

Of course, salty water can also be rejected by simply increasing the main sample delay. In that case, how much improvement in depth are you seeing in practice using "tilt feedback", as opposed to the standard method?

[deemon]

Does this not desensitize the detector to small targets?
If the "tilt feedback" can reduce the response to salty water, then the same will happen for small gold nuggets.

This question is interesting and must be observed more thoroughly .... The main point here is the flyback pulse duration . For simplicity we can consider that any target response starts in the middle of the flyback pulse , and now we can observe the situation with a different targets TCs . Imagine for example that our flyback pulse duration is 5 uS , and a measuring interval after it is 500 uS long . Now imagine 2 targets , TC1=1 uS , and TC2=25 uS . You see , signal of the first target , being started in the middle of the flyback pulse - almost totally decayes up to the end of the flyback , and then only a negligible part of this signal ( its "tail" ) continues to decay further in the measuring interval . So we can say that it releases about 99% of its energy during the flyback , and about 1 % on the measuring interval .... and now remember what this "tilt feedback" does . This feedback EQUALIZE the currents before flyback and after ( restoring the flyback energy drop ) , so it completely compensates all this 99% of the target signal , and leaves only 1% for the further measurement - and we can say that this feedback mechanism really desensitizes our detector for such kind of a ultra-short targets .

But for another target , with 25uS TC - the situation is completely different . Now we can see that this signal spends only a 10% of its life inside the flyback pulse , and other 90% in the measuring interval .... and now - this "tilt feedback" does compensate only this 10% , leaving 90% for the measurement - and we can feel the target very good . But if we try the very big target , with TC3=1000 uS , for example - then we'll notice that it hasn't enough "room" for its decay , because now all the measuring interval ( 500 uS ) is too short fot its tail to decay enough , and now the sensitivity starts to faint again . So we can understand now that this circuit , although being very simple - does operate like a "time filter" , selecting the target by their TCs . Any target signal being mush shorter than a flyback pulse is being suppressed completely due to this tilt feedback , and every target that longer than measuring interval ( square wave half-period ) - is being suppressed too . So we can "program" this kind of detector to feel a desired targets , setting both the flyback duration and pulse duration ( working frequency ) as we need , and it really works as I told here in my experiments . For example , my flyback pulse is long enough to suppress the salty water signal , but the coin signal it feels OK .

Of course , all that I say here is true for the "all metal" detector based on the coil current measurement ( VERSION 1.0 in my topic ) . And only for this type of detector this tilt feedback is an important part of this "time selection" mechanism , helping to suppress all "too short" targets ( like salty water and so on ) . But for other variants of my processing algorithms , especially for the full-balanced VERSION 3.0 ( that I called an ultimate metal detector in the topic ) - tilt feedback can be replaced to the simple manual control ( variable resistor ) , because all the discrimination functions there being performed by more advanced correlation technique , and we don't need to maintain our CC wave so precisely .

[6666]

Thanks Davor all book marked except this one, its no longer found.
A Fast 4-D TEM System for UXO Characterization

http://www.estcp.org/documents/techdocs/UX_0105.pdf

[Davor]

I've gone through the forum and most probably I missed most of the stuff. I'm here from 2012, so many things I missed by then.

The first mentioning of constant current drive that I observed was here: complete project
I did not grasp many things like viscous soil and ways of the PI by then, but others did.

You can find deemon's constant current project here PI metal detector with energy recuperation.

Beside the Triangular wave topic linked before, PiTec posted several solutions employing constant current here: SIMULATIONS
You can check his constant current source Tx at Findmall forum Here: Constant Current Test Circuit and it is still available at the linked address.

Funny, but it seem that some technologies start glittering in their full glory only after they become marketed by some large manufacturers. The CCPI approach is one and the same as TEM (Time Domain Electromagnetic Method, or Transient Electromagnetic Method) and it was about time it descends from the geophysics academic circles to the metal detecting arena. A 2003 paper "A Fast 4-D TEM System for UXO Characterization" that was referenced in this forum in 2006 that beautifully describes this technology, and the best of all - it is free for everyone to play with. I have that paper, but it is too large to be posted here in one piece. It may be found on Internet.

Aziz plays a lot with TEM, and it is a shame he is no longer here.

[Qiaozhi]

So I need to use the "tilt feedback" circuit in my CC transmitter only to correct a minor errors in some special conditions ( salty water , for example ) , where the flyback energy loss can change significantly .

Does this not desensitize the detector to small targets?
If the "tilt feedback" can reduce the response to salty water, then the same will happen for small gold nuggets.

[6666]

I'll post links when I find them

Thanks Davor.

[deemon]

I agree that changes in coil inductance (due to the ground) don't affect the peak current, which is determined by ohmic resistances, but it does affect the turn-on & turn-off slews, which can affect ground balance and target response. I'd guess that Minelab is attempting to keep the slews constant in the presence of ground.

In my circuit this slew influence is negligible , in reality . I use a quasi-resonant current reverse , the principle of this circuit is an energy exchange from L to C and return with an opposite polarity of the current ( just like a half-period of a pendulum oscillation ) . So if we increase the L value ( due to a ground influence ) , what we can see ? We only can see that the flyback pulse became slightly higher , and the flyback pulse duration became slightly longer ( both values change is proportional to the square root of the L change ). But the current in the coil after flyback doesn't change , and the target response has only a negligible change - this is why this circuit has good "inherent ground tolerance" . So I need to use the "tilt feedback" circuit in my CC transmitter only to correct a minor errors in some special conditions ( salty water , for example ) , where the flyback energy loss can change significantly . And in many cases - manual adjustment of CC compensation is enough ...

But in Minelab CC circuit the situation is different . As I can understand ( from their patents and promo video ) , they use a "direct" method of current reverse , so they are applying the voltage from the hi-volt supply to the coil ( via the fast mosfet switch ) and waiting till the current swings from +I value to -I , for example . And when we know the current I , inductance L and voltage U , we can easily calculate the time that we need to completely reverse the current , using the simple equation T = L*I/U . Of course , this method needs a very fast and precise switch control , and another issue is that every deviation of the coil inductance must cause an error ( directly proportional to L change ) in the current value just after flyback pulse - the next current pulse will start with the wrong current . So their circuit cannot work without a very precise feedback control loop , in order to maintain the stable coil current with any disturbing factors , and the most important among these factors is a ground influence , as I think . By the way , another factor is a mechanical stability of the coil .

[Davor]

I think they simply actively filter (by means of feedback) the Tx coil supply during CC phases to reduce noise injection to Rx, and the rest of it is marketing.
Such filtering makes sense because even with proper balance you'll have at most some 60dB isolation between Rx and Tx, and we are talking about some serious currents here.

Say a Dude invents shoelaces. Marketing guy asks him where you may wear those, and Dude says "everywhere". And the advert says "moonwalking shoe-fastening devices" or MSFD.

Similarly with Tampax tampons, you're all set for summer. You can swim. And ride, dance, cycle, play golf or tennis in complete comfort with nothing to hamper you ... even if you are a man and you never learned to do any of those things.

[greylourie]

Sorry Carl,

I lifted the term from your post, below :

So far I've seen feedback loops to compensate for the TX drive tilt. But mineralized ground alters the loop inductance and changes the overall drive level. Minelab uses a feedback loop to detect this effect and maintain a constant drive level. That's what their patent is about.

Being challenged, electronically speaking (and in many other things), I thought having a constant current type of drive - implied that any outside influence affecting inductance, would naturally and automatically be compensated and adjusted for ? Basically that constant current and constant current drive are mutually inclusive ? When I hear the word "constant", it conjures up this notion of a device which adjusts to compensate and ensure the correct result.

Thats why I linked to Tinkers thread. I guess my biggest shortcoming is I don't think electronic, unlike like others here on Geotech. Sometimes ignorance is bliss. Its frustrating, both trying to grasp what is being discussed, and trying to put an idea or notion across.

[Carl-NC]

In my circuits ( unipolar and bipolar ) inductance change doesn't have any influence on the coil current . So I use feedback ( servo loop ) to maintain a proper current pulse shape ...

I agree that changes in coil inductance (due to the ground) don't affect the peak current, which is determined by ohmic resistances, but it does affect the turn-on & turn-off slews, which can affect ground balance and target response. I'd guess that Minelab is attempting to keep the slews constant in the presence of ground.

[Carl-NC]

Hello Carl,

What is the difference between constant current and constant drive level ?
There was a thread over here, that fizzled out : http://www.geotech1.com/forums/showt...urrent-control
I get confused with everything.

I refer to 'constant current' as the on-time behavior; when you turn the coil on, the coil current quickly jumps up to a certain level and stays there until it's turned off. This requires that the turn-on drive include a high voltage supply in order to achieve the slew rate.

In a traditional PI, the turn-on behavior is an obvious (slow) exponential. In CCPI, it's also likely exponential but the high voltage supply makes it look more instantaneous. There are ways to get other-than-exponential CC drive.

Finally, with CCPI you can either continuously cycle a bipolar coil current (a current square wave) as Minelab does it, or pulse a bipolar coil current as in the White's patent.

Not sure what you mean by constant drive level.

[deemon]

.

[deemon]

Deemon, does your system automatically compensate for changes in inductance caused by ground changes ?

In my circuits ( unipolar and bipolar ) inductance change doesn't have any influence on the coil current . So I use feedback ( servo loop ) to maintain a proper current pulse shape ...