IGSL

[Koala]

http://www.geotech1.com/forums/showt...787#post156787


I built the Bandido version worked fine

I believe the board layout was from http://md4u.ru

[Hyena]

http://www.geotech1.com/forums/showt...787#post156787


I built the Bandido version worked fine

I believe the board layout was from http://md4u.ru

Isnt it approximately only a half of IGSL?

[Koala]

Isnt it approximately only a half of IGSL?

Err yes.

And no

And sort of in between.


Performance of the Tesoros are very much alike. A two tone mode can be added to the TGSL. Where as the Bandido has a true no motion all metal mode. Very useful in my opinion.

First you can detect much easier in streams and around bushes. Second iron sound completely different. On a slow sweep its beeeeeeeeeeep with a slow rise and fall to the sound as a good targets small beep and is a much sharper tone. Third you can size and shape a target.


Both these SMT layouts have been built and known to work. As far as I know there isn't a SMT version of the ITGSL




I could never put my finger on it but was never totally happy. Out preformed my IDX on a test garden using real discrimination levels but seemed to find less in the real field. Used it for about a year and do miss the no motion mode.

[Hyena]

Hi Davor,


please, could you place here absolutelly last schematic of IGSL Musketeer and IGSL normal with all modifications. I will try to make SMD board. Thank you very much.

Corrected schematics which I use now is in attachement and this is change list:


U12 LF351 -> LM833 + pin correction
R31, R45, R50, R64 -> 470k
C16, C24, C29, C37 -> 22n


U4c pin swap 8 <-> 9
U7c pin swap 8 <-> 9


As I said, Davor made many more mods, which you can find here:
http://www.geotech1.com/forums/showt...083#post176083

Err yes.


And no


And sort of in between.




Performance of the Tesoros are very much alike. A two tone mode can be added to the TGSL. Where as the Bandido has a true no motion all metal mode. Very useful in my opinion.


First you can detect much easier in streams and around bushes. Second iron sound completely different. On a slow sweep its beeeeeeeeeeep with a slow rise and fall to the sound as a good targets small beep and is a much sharper tone. Third you can size and shape a target.




Both these SMT layouts have been built and known to work. As far as I know there isn't a SMT version of the ITGSL








I could never put my finger on it but was never totally happy. Out preformed my IDX on a test garden using real discrimination levels but seemed to find less in the real field. Used it for about a year and do miss the no motion mode.

I get it - I also wanted to mod my TGSL to two tone, but IGSL has two independent channels - so I decided to keep TGSL as it is and make IGSL. You can set it up to produce low tone for ferrous, high tone for disc + mixed tone for higher quality trash. Its like 3 step acoustic VDI and it is quite brilliant


I thought I would use this all metal non motion channel into IGSL, but I handled it well and I do not need pinpoint channel anymore. Maybe - If i will have time in winter I would like make digital controlled IGSL with this channel, otherwise control panel will look like in power plant


For deciding what shape target has I use bar graph VDI from Daren
http://www.geotech1.com/forums/showt...763#post196763

[Davor]

I'm not up to checking all the schematics (just got back from a trip to Czechia )
I promise to see the latest. I made quite a few mods on my machine, but frankly, not all of those yield significant improvements.

@Koala, the 2 tone mod is not quite the result you get with overlapping tones you get with IGSL. The 3rd tone which is a result of overlapping is also not just the 3rd tone. The big deal is that with a response angle nearing the discrimination limit, the sensitivity drops. With overlapping tones, the border of discrimination is a bit further in an overlapped zone, so I get all the sensitivity even at a border. The overlapped tone is reserved for aluminium and coins with steel core, so it is not a big loss.

[Davor]

I used IGSL mostly as a learning platform, but the results so far are very good. I also got Whites 6000 Pro XL, and in comparison this simple IGSL does a better job in most of the terrains. It is simple, fast, and intuitive. OK, 6000 is deep but slooooooowwww.

The mods were done in a way to enhance the simple-fast-intuitive trinity:

The most important mod in "simple" direction is the pin swap mod. 2 comparators inputs are put in reverse against the original setup to allow for a common "Disc sens" control, and overall good behaviour. Even if you want to leave everything else as is, you should do the pin swap mod. Common "Disc sens" allows both channels to be tamed together in cases when, say, old vehicle passes by and spark plugs create big noise.

The other thing to consider is to apply a common GB mod. This mod is not so much towards simplicity, but more to correctness and depth on difficult terrains. Proper GB is much better than the approximate one.

The "fast" is achieved by introducing symmetry into compression, and getting rid of the excessive filtering. The first gain stages were left mostly as they are, but with diodes removed. For a small signal there is no difference, but with a large signal removal of diode ensures synchronising transitions in Disc and Geb channels. The original solution promotes shortening the beeps, with odd delays, and the effect depends on signal level. That's the classic Tesoro thing (apparently even Tesoro gave up on the old solution). The second stages have the LPF capacitors removed, and they form a soft clipper with high gain, using anti-parallel diodes. This helps levelling the 6th power signal level relationship sharpness to a much softer transition, some extra gain, and transitions from saturation are instantaneous.

Discrimination comparator stages are somewhat reconfigured to avoid deep cut-off voltages, and the associated recovery time. Some classical devices use Zener diode, but I opted for resistors.

To enhance the "intuitive" side I slightly transformed the "Disc threshold" control to perform as a "volume control" of proportional sound. The original sound is binary, which means the machine beeps loudly when discrimination criteria are met, and almost entirely regardless of the depth or any other nuances in between. Such sound is very chattery if you push it to the limit. In contrast, the proportional sound is even more chattery, but not nearly as irritating as the binary sound, and on the plus side it gives you a fine "immersion" ability that allows you to visualise fairly accurately the size and depth of a target.
I put the stereo pot there, but a few days ago Ivica suggested these controls to be separated, to enable attenuation of iron sound in heavily iron-contaminated terrains. It is optional, and I agree it makes sense.

Additional mods include:
Tx AB->C mod, where Tx goes from AB-class to C-class, as soon as the oscillation is strong enough to produce negative voltage rail. It slightly improves signal purity and power efficiency, but it is not giving a significant improvement.
I removed a voltage stabiliser (Petravka board has only one), and I use 9V cube batteries for power supply. It works for me, but it may be a problem with 12V batteries.

Refinements:
The front end as taken from TGSL is suboptimal. In original configuration it has over 10nV/sqrt(Hz) of noise, so this stage will need some extra love. I fixed it in my device by reducing resistors' values, but my coil is not a commercial one, and going that way blindly may not be a good idea.
Audio stage may also use some extra love, as it may induce instability with depleted batteries. Placing 8ohm speaker here most certainly is a bad idea.
Anti-parallel capacitors should be replaced with bipolar ones. Somewhat larger values than original are OK, to promote slow motion response.

TBD
Protective diode in power supply should be replaced with P-channel MOSFET connected as "active" diode. IGSL is a quite efficient device, and it deserves efficient supply.
GB and All-metal references are banged together as in TGSL. It would be much better to split the GB and All-metal-reference functions to separate channels, and you already have 2 channels to do that. I'll fix it in a corrected schematic.
Additionally the GB channel may be rectified in an absolute value circuit. This would extend the GB to the most exotic terrains, including beaches. Since all my mods were done with minimum changes on a PCB, I'm considering this mod to go to a separate small PCB. People that do not have exotic terrains will not benefit from absolute value circuit mod anyway.

[sini]

Hi Davor, thanks,
I will try to make SMD board. Robert

[Davor]

I (almost) finished the schematic with all the mods. It needs a careful look to fix errors that may be in there, and careful pondering if there is anything else to add. SMD is not forgiving.
I separated GB from All metal, introduced new discriminator detail, added "tune" function like in Whites 6000, fixed audio PA.
Added power switch support with wrong polarity protection.
I also replaced Tesoro-style Rx front end with 2-op-amp in-amp. While this may seem odd, this will allow for replacing the Rx preamp chip with a much better one, and get improvement as well. With Tesoro configuration you are stuck at >7nV/sqrt(Hz). With modestly priced chips like LME49720 it makes no sense to continue chasing one's own tail with old school op-amps. Besides, I can introduce a simple 20dB gain regulation with a single pot.
In Tx I added a phase correcting capacitor and a saturation-killer diode. I expect over 10dB 2nd harmonic improvement.
Unfortunately I'm not in a position to test all the new mods right now. It would be best if these are tested prior to going SMD. Better safe than sorry.

[Davor]

With a caveat that I did not test some of the mods, here is a fresh schematic.
There is a function I'd wish to include, but I'm not clear how exactly to implement it. It would be a global phase shift, a function that would facilitate coil changes. I'll add it when I decide where to put it.
This solution is still work in progress and open for debate. I made some last minute changes before posting, and because all my errors in using ExpressSCH that are hidden inside, so it is not ready for netlisting.
The vast majority of mods are doable in the existing builds even without extra pieces of PCB.
I must test the in-amp front-end, power switch, audio PA, saturating transistor switches introduced in discrimination, and splitting/joining all metal and GB channels. The rest I tested with favourable improvements.
Ditching a voltage regulator was a move in a very good direction, as batteries are quieter than any regulator by several orders of magnitude. The difference was very notable. I'm using 9V cubes. I didn't try the rechargeable ones, but those would be OK as well.
I intend to introduce an add-on PCB with wide range GB for beach hunting. Most users do not need salty beach functionality anyway. Splitting All metal functionality from GB alone will fix most of the problems related to high GB settings - no more falses on difficult terrains.

[eclipse]

Congrats Davor on the hard work!
By any change do you mean by "global phase shift" something like auto compensating module for Fortune ?

[Davor]

I don't have a Fortuna schematic with me, so I guess this device is following the ground balance phase automatically? In that case, no, the global phase shift is not what I have in mind.
Every new coil will have its unique resonance and Q factor, and will pose a slightly different phase shift of the received response. If I want to change coils easily, I'd have to compensate this unique phase shift. Without such compensation I'd have to adjust All metal reference, and both discrimination phases with every new coil.
So for example, I adjust discrimination phases and All metal reference for a first coil, and when I change it for a new coil I only have to adjust a global phase shift and keep it while I use that new coil.

[eclipse]

The phase shift you describe is incorporated (I think) in Fortune's software, haven't made it, just watched few videos and what
I've seen videos where you put a ferrite piece over the coil, and you see the balance on the screen. You can then adjust the proper phase shift.

[Davor]

Yes, that's it. You need a ferrite, and preferably small non-toroidal bunch of them to establish 0°.
Of course, I have to do it in an analogue circuit.
I considered automatic GB, but my terrains are a little specific - large stones with virtually no mineralisation scattered in highly mineralised terra rossa clay. I need to fix GB on clay, and ignore the stones by mental elimination.

[Davor]

Beware the pin numbering of gain blocks, it is wrong. I'll fix those eventually.

[CDI]

Beware the pin numbering of gain blocks, it is wrong. I'll fix those eventually.

Davor, can you confirm those pin swap's on U4 and U7?