IGSL

[ivconic]

To support my attitude explained in previous post;
ok, DSP is superior! Is it?
I have seen, tested and worked with majority of high end md's that are actual today.
If i name anyone of those - i can risk to provoke bedlam of opposite opinions, that's why i will not name neither one of those.
However, fact is i seen, tested and worked a while with majority of those.
Many of those are under "DSP" label (whatever that may imply! )
A pile of top electronics inside those machines! Latest technology! DSP! Blah, blah!
...
...
At the end of a day... results... average, not that superior over "analogues" !
In many cases much less than superior - "worse" would be the shortest description..
So? What we are talking about here?

[WM6]

Call me "old school", call me "backward", call me whatever you like - but i still prefer more analogue approach than "digital" (whatever that may imply).
Why?
Analogue signal is ALWAYS carrying more informations than any kind of digital signal! (talking about md's)
You may sample it at any rate - still it will stay "sampled", if you understand what i mean...
Problem in processing analogue signal is not in how much informations it can care - but in accuracy and complexity of processing it.
That's where uPC's appear as more logical choice.
But still main fact will remain unbeatable: analogue signal is carrying much more informations!
Always, ever!
That's why i don't like much to think about "DSP".....at least not yet.
Another reason is because i am amateur and can not afford (nor have enough knowledge) to "play" more with it.
So, naturally, i am turned to think only about "analogue" solutions...
(i am sure some people will laugh much on this... but what can i do?)

Quite close to (our) reality.
Is not the same in field of audio? No one digital approach can give us audio quality as an old tube audio device.
By analog detecting you need your brain to processing signal, by digital detecting you do not need your brain any-more all is still processed - you need only to believe in displayed data or not to believe.
But digital is not saying its last word, and combined solutions are possible too.

[ivconic]

That's another way to do it. But there is no true need because you only have one RX coil, so all the information is in the RX signal, whether you have 1 uPC or 10 uPCs. But practically, it might be easier with more uPCs ("channels"). If you use 1 uPC, you can still keep separate "channels" in software.

I'm playing with the MPS430 u-controller from TI. It is not really powerful enough to directly sample the RX signal. But it could easily sample the output of the LM308s and perform the discrimination and detection logic. That is one possibility I am considering. It might be possible to sample an earlier stage also, to reduce the parts.

We have plenty of work to do reinventing the wheel. Maybe invent a new spoke or two along the way...

Regards,

-SB

Call me "old school", call me "backward", call me whatever you like - but i still prefer more analogue approach than "digital" (whatever that may imply).
Why?
Analogue signal is ALWAYS carrying more informations than any kind of digital signal! (talking about md's)
You may sample it at any rate - still it will stay "sampled", if you understand what i mean...
Problem in processing analogue signal is not in how much informations it can care - but in accuracy and complexity of processing it.
That's where uPC's appear as more logical choice.
But still main fact will remain unbeatable: analogue signal is carrying much more informations! (md's...don't take this generally)
That's why i don't like much to think about "DSP".....at least not yet.
Another reason is because i am amateur and can not afford (nor have enough knowledge) to "play" more with it.
So, naturally, i am turned to think only about "analogue" solutions...
(i am sure some people will laugh much on this... but what can i do?)

[simonbaker]

"..With uPC, you will not need two separate analog channels -- just sample the RX signal and everything else is done in software..."

Ha! That's the part i don't like.
Because it's been already done that way, many times before.
Nothing new in such method.
No..what i meant is literally 2 blocks (2 channels each) separated as on IGSL.
Audio generation and appearance is a bit unsolved (yet) at IGSL.
I would try to implement uPC more in that direction... if i knew how!

That's another way to do it. But there is no true need because you only have one RX coil, so all the information is in the RX signal, whether you have 1 uPC or 10 uPCs. But practically, it might be easier with more uPCs ("channels"). If you use 1 uPC, you can still keep separate "channels" in software.

I'm playing with the MPS430 u-controller from TI. It is not really powerful enough to directly sample the RX signal. But it could easily sample the output of the LM308s and perform the discrimination and detection logic. That is one possibility I am considering. It might be possible to sample an earlier stage also, to reduce the parts.

We have plenty of work to do reinventing the wheel. Maybe invent a new spoke or two along the way...

Regards,

-SB

[ivconic]

"..With uPC, you will not need two separate analog channels -- just sample the RX signal and everything else is done in software..."

Ha! That's the part i don't like.
Because it's been already done that way, many times before.
Nothing new in such method.
No..what i meant is literally 2 blocks (2 channels each) separated as on IGSL.
Audio generation and appearance is a bit unsolved (yet) at IGSL.
I would try to implement uPC more in that direction... if i knew how!

[simonbaker]

In addition;
now.... i am plain amateur with very limited capabilities, so analogue approach is all i can do.
Would be nice to see IGSL idea transfered into "digital" area (two simple detectors in one, controlled by uPC).
Main goals would be accuracy, small sizes, smt and very low consumption (which is now the main issue at present IGSL; some 65-70mA) ....
and possible more benefits than i can name right now.
..
Referring the power drain; i was thinking to ask here about LP versions of some IC's.
LM339, LM883, LF347..and others.. are those having LP versions?
Can we calculate power savings if use LP versions?
I bet we can... but i would rather let somebody more conversant than me to do that...

Would be nice to see IGSL idea transfered into "digital" area (two simple detectors in one, controlled by uPC).

I have already started a project along those lines. However, I'm using the wrong uPC to do it right, but it will help me get started. Given my speed, I'm sure someone else will get there first. I bet Aziz has all the tools to do it in a week or two.

With uPC, you will not need two separate analog channels -- just sample the RX signal and everything else is done in software. Interpreting the signal can be done same as IGSL or any other logic you want. By adding some analog hardware, you can use a cheaper, less powerful uPC also, but less flexibility.

But is it fun? I think less, but has to be done anyway.

Cheers,

-SB

[ivconic]

Because large rusty targets can give both a ferrous and a non-ferrous response, you will often see (or have experienced yourself) the case where you dig a large hole and find an old rusty horseshoe. This seems to happen with all detectors. I was wondering if it is possible with the IGSL to really distinguish between a coin and nail lying close together, and an old horseshoe?

Perhaps a new video is needed??

Yes that is the main question!
Most of the time operator (already skilled with IGSL) will easy hear difference in such possible audio sequences.
Yet there maybe rare situation where audio sequences from both cases will sound pretty same...
But hey.. nothing is perfect!
That's why i am in the middle of "heat" to upgrading it more...

[ivconic]

Yes there is LP339....! One less to worry about!

[Qiaozhi]

IGSL is performing excellent on such large irons in most of the cases.
It is producing both tones at the same time, same lasting, same decay.
...
Significant detail from that day, would be several times repeated situation; where Vaquero's clearly indicated "colored" item while IGSL indicated large iron (both tones at the same time).

Because large rusty targets can give both a ferrous and a non-ferrous response, you will often see (or have experienced yourself) the case where you dig a large hole and find an old rusty horseshoe. This seems to happen with all detectors. I was wondering if it is possible with the IGSL to really distinguish between a coin and nail lying close together, and an old horseshoe?

Perhaps a new video is needed??

[ivconic]

In addition;
now.... i am plain amateur with very limited capabilities, so analogue approach is all i can do.
Would be nice to see IGSL idea transfered into "digital" area (two simple detectors in one, controlled by uPC).
Main goals would be accuracy, small sizes, smt and very low consumption (which is now the main issue at present IGSL; some 65-70mA) ....
and possible more benefits than i can name right now.
..
Referring the power drain; i was thinking to ask here about LP versions of some IC's.
LM339, LM883, LF347..and others.. are those having LP versions?
Can we calculate power savings if use LP versions?
I bet we can... but i would rather let somebody more conversant than me to do that...

[ivconic]

How does it react to a large rusty chunk of iron, such as a horseshoe?

Seems Silverdog missed this?
I will try to answer instead.
Good question.
IGSL is performing excellent on such large irons in most of the cases.
It is producing both tones at the same time, same lasting, same decay.
But as i noticed; shape of large iron sometimes can cause a slice different sequence of tones.
If object is simple shaped than both tones appears equal.
If object is long and "slim" than audio sequence will depend on exact position of that object, related to coil position.
For sure; in both cases, both pitches will occur.
...
The other day, when i discovered that nice silver coin, there were two original Tesoro Vaquero's in "company" too. One "Germania" and one "USA".
So we performed few mutual checks.
Significant detail from that day, would be several times repeated situation; where Vaquero's clearly indicated "colored" item while IGSL indicated large iron (both tones at the same time).
So after digging .... IGSL was right!
Tough to claim anything, it is to early now, need more real tests to do.
But for sure; IGSL idea may turn as good step forward to solve damn "horseshoe" issue!

[Jerry]

Very Nice! It took me several tries to get one looking that neat. Like the idea on the guide posts. I did something similar except it was just two walls which formed a corner for the frame to fit into .

What did you use for a frame to hold the plastic?

Nice looking shells.

Jerry

[golfnut]

Thanks,

Steve

[silverdog]

Hi Silverdog.

If I ordered one of the IGSL kits today, would you include an up to date schematic and placement.

Plus a link to the exact coil to make. That others have done to go with my kit..

Im not being fussy beleive me , I am a dyslexic - and the thought of reading 4700 posts whould put me out.



Steve

Hi Golfnut

There is no placement as such - use the silkscreen on the pcb

I have added the relevant documents to the library box on the right hand side of my shop (silverdog.co.uk)

regards

Silverdog

[simonbaker]

This has already been discussed many many times, ad infinitum.
The deep null will not produce the best results for two fairly obvious reasons:

1. The initial phase-shift between the TX and RX coils must be correct for the discrimination circuitry to work properly, otherwise it will be out of range of the sample pulse timing. With a deep null, the initial phase-shift is completely wrong. You can clearly see this in the graphs posted earlier in this thread, and a simple experiment involving an oscilloscope will demonstrate this fact. You can even see it in a SPICE simulation.
2. With a deep null, the residual amplitude is very small (close to zero), which again means the discrimination circuitry is unable to do its job. Commercial Tesoro coils I have measured usually have a residual voltage of approximately 15mV. The discrimination circuits need to sample the RX waveform at the in-phase and quadrature positions. If the RX signal is buried in noise, the detector will fail to function correctly. However, this second point is somewhat academic, in the case of the TGSL (or IGSL), as point one has already been violated.

Respectfully I have not finished considering this subject, and may have some more thoughts about it that do not concur exactly with those statements. I know it seems beaten to death and that Qiaozhi and I spent a lot of time on it, but some new angles have come to mind that I need to look into before signing off on the "conventional wisdom". However, I agree with the nulling method described.

Regards,

-SB