IGSL

[sini]

Hi Davor,

I think in case of IGSL you are talking about D5, D9, are not you ?
Sini

[Davor]

Yes, those are problematic for VDI, but to keep the VDI numbers constant you'd have to keep the discrimination setting fixed. D5 and D9 can be removed, or even better moved in anti-parallel with D6 and D8. That's one of my mods.

[Davor]

Today I was able to visit my "lab" and play a little. Time was short so I picked only one easy thing, a Tx. I wanted to apply the AB->C mod, and it was a success. Unfortunately I was not able to test it deeper in class C so I did not go all the way, but even this far is good.

This is the output of a regular IGSL Musketeer Tx, very similar to IDX:



The most interesting point to observe is the output transistor's emitter because it shows how deeply it goes to saturation, and how much it lags. It is OK for this transistor to go briefly to saturation, but not beyond the supply rail. A good saturation looks like this:



I (ab)used the negative voltage rail to pull the driver transistor into C class, which is a clever thing to do. When oscillator starts there is no negative rail, so it starts as AB class oscillator. As soon as it starts, a negative rail appears and pulls it down to class C. As a consequence a Tx uses less current, and has improved signal purity. Simulations say over 10dB improvement, but a picture speaks volumes:



Unfortunately I did not take a CRO shot of the emitter for this one, and I did not go all the way with it. I tested the concept with a very conservative 1M ohm, while somewhat better results would be achieved with 680K as a pull down resistor. 470K is critical, and I didn't want to be without a working IGSL.

IGSL is a bit less nervous this way. Detection distance is roughly the same. I'd say the most of the chatters are a direct consequence of the Tx misbehaviour.

[sini]

Good job, Davor.
I am looking forward to schematic. When IGSL is less nervous with your TX oscillator than you can increase the gain and get more depth ?
One other question. Is here on the forum some convertor from phase to frequency for IGSL ? I mean something like VDI - higher number of VDI = higher phase shift = higher frequency.
Thanks

[Davor]

Not that I know of. It would require analogue division if you deal with demodulated baseband, like you do with most of the designs around here. Analogue dividers are limited in dynamic range. VDI basically does that in digital world.
There is another way of achieving accurate phase detection that requires re-modulation, limiting amplification and phase detection. Because this approach is fairly close to building a whole new detector from the scratch, I'm about to do just that for my LF project ... well, kind of. I realised that some steps are easily skipped, so it does not have to be a monster-project after all. It does require some serious re-thinking of the regularly trodden paths.

[Davor]

The AB->C mod goes like this:



I did not test the 680k for this purpose, but simulation expects some very nice results because it pulls the oscillator deeper into C class, and conduction angle is such that it reduces both 2nd and 3rd harmonics. I tested only 1Meg.
For this mod you must put 33k in parallel with the existing one or replace it with 15k.

[sini]

This week I will try C class TX oscillator. I am looking forward.
I miss acoustic ( VCO frequency ) indication of VDI number.
Despite the fact, that I removed diodes (D5 and D9) from my IGSL, VDI number is not constant. VDI number depends on distance from searching coil. I can see video on youtube, in case of IDX + VDI it is much more better.
Thanks

[Davor]

That's because IDX has a Fisher-like discrimination phase manipulation. It means that first gain blocks stages are at constant phases, and ideally orthogonal. You can have that as well in IGSL, but only in case you keep the discrimination at constant position, and preferably orthogonal to GB.

A VDI could be programmed so that it becomes an acoustic frequency source. I'm not into micro's so you are on your own there.

[parun14]

Hi to all
I like to build igst-tgst but don't know about electric.I can make a pcb and can put all part on board.You think i can make it work?
I like to know what the last rev for pcb and part list.
please.

[Davor]

Are you by any chance a Croat?

Perhaps you'll need to re-formulate your request to cover more options. What exact function you wish your detector to have? If you want a decent single tone discriminating device, maybe you should include also IDX in your choice. It is far more popular and has a very good PCB to start with.

[golfnut]

"VDI number depends on distance from searching coil." The vdi number is a representation of the signal amplitude difference in the x and Y channels. To keep the decode of phase angle correct for close up large targets you need to reduce the gain of the system to maintain linearity. To do this you require gain control either on the Rx or by reducing the Tx signal level for the same effect - or a blend of the two to max the dynamic range. The other advantage with this regime is that you keep Rx signals centred over the full dynamic range of the AtoDs for best S/N. S

[parun14]

I like to build a detector for small gold nuggets .
Is IGST can do the job?

[Davor]

Seek for Lobo. It is much better for small nuggets.
IGSL is superb for artefacts as it discriminates both iron and coloured metals, and in the overlapping area it gives indication of foil/small gold. However, a device with "all metal" indication, and a somewhat higher frequency will work better for small gold.

[parun14]

And
What the higher frequency one that have a plan pcb to build?

[Davor]

Again, seek Lobo.