DETPIC 03

[PhilippeM]

HI every body,
As i said you later here is some picture of our project with WILLY.
More information as soon as possible.
Regards
PhilippeM

[alex]

Re: DETPIC 03

Beau travail Philippe ,quelles sont les caractéristiques de détection , et diamètre de la bobine?

[PhilippeM]

Re: DETPIC 03

Coil's diam: 200 mm
Can detect a piece of 1 euro at 250 mm and a coca tin at ~600mm.
I need to make more tests. Be patient.
PhilippeM

[Sean Goddard]

Re: DETPIC 03

How COOL does that look?

Let me know when I can have details please, I'm VERY interested in this project.

[Willy Bayot]

Re: DETPIC 03

We tried to contact you in the contact part of your web site but this does not work.
Since you did not publish your email address in this forum, we have no way to contact you directly.
To get an overview of the features of our DETPIC03 project, search for my previous posts in this forum.
If interested for more, send me a message.
Willy

[Sean Goddard]

Re: DETPIC 03

Hi Willy,

Try my website now, I've put the OLD contact page up for a while, until I can get the new one running.

[marshall]

Re: DETPIC 03

>Coil's diam: 200 mm
>Can detect a piece of 1 euro at 250 mm and a coca tin at ~600mm.
>I need to make more tests. Be patient.
>PhilippeMDear Philippe,
I would appreciate any details ,updates on this project.Is this the much rumoured new pi detector to be released this year in Australia? Rumours abound here that such a detector has been trialed in Australias goldfields.I know very very little about it except that it claims to be able to detect large gold nuggets at greater depths than any other technology and ignore or discriminate ferrous targets correctly at depth.It is rumoured to be very expensive and be operated by a digital key so that if it is stolen nobody else will be able to use it.
Best regards,
Marshall

[Willy Bayot]

Re: DETPIC 03

The DETPIC03 is NOT the detector rumoured about in Australia.
It is a new All-Digital PI project made by a team of two European guys.
Search on this forum for posts under my name and you will find some featuring details about it. Its main features are an automatic ground control and a positive discrimination with target ID.
It will still probably take two to three months before we could actually announce more practical details like : how to get it and build it yourself.
At this time, we only have a working prototype version 1 (see pictures in the forum) which has been thoroughly tested in lab but no field tests yet. We are actively working on a second version with a re-packaging to separate the Human-Interface module from the back-processing board.

Willy and Philippe

[Willy Bayot]

Re: DETPIC 03

Sean,
Sorry, but it is now worst than yesterday as your home page does not even come to life.

Willy

[Claude]

How could you achieve discrimination on a PI??

Hello Mr Bayot,
Your project is very interesting, I'm eager to see it on this site along the other projects, as I understand you will kindly allow us to buid it ourselves. COuld you comment please on how could you discriminate with a PI? I knew this is impossible.
Thanks,
Claude

[Willy Bayot]

Re: How could you achieve discrimination on a PI??

This is not impossible since we did it. However, one must take a completely different design route compared to the traditional PI systems based on analog integration circuits. It requires to evaluate precisely the shape of the decaying pulse almost µsec by µsec and to extract a typical signature out of it. It is actually a pattern recognition principle as used to identify finger prints or written signatures. Obviously, this requires an all-digital signal processing. That's what we did using standard micro-controller devices.
A bit more patience and you will know more about it. This is not a joke nor a gimmick but we need to be ready with doc. and field test results before we publish more.

Willy

[FJ]

Re: How could you achieve discrimination on a PI??

So, can it actually tell what kind of metal it is detecting and this be independant of size? More detail, no patience here
FJ

[Willy Bayot]

Re: How could you achieve discrimination on a PI??

Yes, it can, within the limits of the usual laws of physics.
According to our many lab tests (with log recording on PC), the signatures of the same target at various distances and same targets of various sizes (same thing), are grouped like a good sharpshooter shots on a target.
On the other side, the signatures of most metal alloys are very well separated when different.
The discrimination is reliable within a large range of size/distance and we have an indicator telling the operator whether the current size/distance is sufficient for a reliable ID.
Hope that helped,

Willy

[Willy Bayot]

Re: DETPIC 03 discrimination test results




This is a XY chart showing a sub-set of the DETPIC03 Discrimination test results.
The target signature generated by our discrimination algorithm is a two-dimensional point defined by its X and Y coordinates.
The quality criteria of a good positive target identification system are :
1. Targets made of the same metallic alloy should give the same or almost the same signature whatever their shape, size and distance (and possibly the proportion of the metal mix in the alloy and the position of a target with a flat or ring shape relative to the coil)
2. Targets made of different metallic alloy should present identification signatures as different as possible.

The attached chart shows the plot of a consecutive number of XY signatures of various targets made of various metallic alloys and of various size, shape and distances. This is like a machine gun shooting short series of cartridges to a target.
The results of the curve pattern matching are obviously not responding to a simple Boolean logic (i.e.. It is sure that this target is a piece of brass with dimensions 40x40 mm) but rather, to a Fuzzy logic (i.e. Most of the signatures of this target hit a XY region generally considered to be brass or copper or aluminum).
However, it is obvious from the chart that:
1. Targets made of the same metallic alloy give well clustered XY shots whatever their size and distance. The slight variations in the grouping is mainly due to a varying distance of the target and to some noise remaining after filtering.
2. Most of the signature clusters of targets made of very different alloys are well separated from each other (e.g. see the group of ferro-magnetic alloys (iron and steel) and the large group of non-magnetic alloys (brass, alu, copper))
3. Generally rejected target types are very well identified (e.g. foil, beer caps, iron/steel, pull tabs)

However, it is also shown that some shapes of targets can give a rather different signature for the same alloy (e.g. The clusters of a gold coin and that of a Gold ring are rather separated partly for their different shape but also maybe for their alloy mix, this is to be checked in more details)
It is also easy to see that some different metal alloys group together because of their almost equal conductivity (e.g. brass, copper, alu) but, is it really important? In all those cases, you dig them up if you search for relics or coins, is not it?
It is not shown on this chart but we had also to accept the fact that a thin gold ring and an aluminum pull tab give signatures grouping near each other. This is a difficult case well known to all discrimination systems. It is due to the simple laws of physics and can not resolved (dig them both!!)

It is also a fact that the same ring presented flat or standing to the coil gives different signatures and also gives the best signal level either under the center of the coil or under its edges. This is also well known in the detector world.

Our DETPIC03 system is not pre-programmed with a frozen table of target identification.
By default, it displays the X and Y values of the target signature on the LCD together with an indication of the signal level. If the level is too low or too high, the XY signatures are not displayed. In that simple mode of operation, the user would keep a paper-based table showing the correspondences between XY pairs and target types.

As an option, it is also equipped with a 'TARGET TRAINING MODE'. In this mode of operation, the detector is connected to a PC while a user presents sets of known targets to the coil, their signatures are sent to the PC where they are displayed in real-time on an XY chart. The user tells the PC how the detector should identify them in his own terms and language. This table is stored on a file on the PC and can be downloaded to the EEPROM of the detector. This process can be later repeated to complete or correct the current table with new target types.
Willy and Philippe

[Pedro]

Re: DETPIC 03 discrimination test results

Congratulations!!!
Wonderfull work!!!
It's nice to see you guys making all this development and showing the results with no fear of other people comercial use of you work...

Best Regards
Pedro