Correct Phases?

[golfnut]

"Wel I don't know. I haven't fully nulled a coil set yet."

You may be jumping the gun a little here.. Ive done this a few tmes now and you can get caught out easily.



1) Id Fix one coil with a few blobs of adhesive - leave the other floating for adjustment.

2) Get your null (go for 200mV or less) and THEN see what the phase is.


If its within the 40degree window for TGSL / IGSL then it will already be able to Disc out Iron and tune out ferrite..

3) Air test - if range is not what you hoped for - probe the preamp out put and note the pp level.. While live hold a 2n2 on the Rx cap.

If the preamp output gets bigger - you need more C. Plus opposite is true.

If when you get the value that really peaks it like crazy - add a 1nF to pull out of resonance. This will be close to the best you can get.


You need to re test initial phase now, if its off or the null is more than 200mV - adjust position of coil again for good null and phase.


4) So you have sensitvity - good air test - try tuning out ferrite, try rejecting clean unplated iron. Accepting bronze / silver/ gold.

If all good pot while observing null as the heat of some resins shifts position of null placement.

If when potted it drifted and you were not available to keep it tuned - you can try a small peice of ferrite placed to give final phase tweak.

The above kind of sums it up really for us amatuers.

If the phase drifts too much during pot and ferrite puck does not give you what you need, then you made a bad coil and options are buy one or make another.
Steve

[golfnut]

Im ashamed to say but all of this took about a year for me to partially understand..

"So as far as I can tell we need the GEB to start at the zero crossing going positive and
move say 20 degrees to ignore the ground."



The ground signal phase is slightly early WRT the TX signal phase. Iron is even earlier. Bronze is later WRT TX.

So this is why there is so much emphasis on getting the Rx coil and its phase timing/coupling to the Tx system good.


The Tx drive and and the Rx preamp output if they are "approx the same phase" then this allows things to happen properly.


The Geb Gate time, set by the simple phase advance RC circuit hanging off the Tx cct will open the GEB gate at a time where the received ground signal phase lives.

The trick in use is to finely arrange the start of the GEB gate to align with say the 90degree time of the geb signal and the geb gate to close on the 270degree angle of the geb signal.

You will have spotted that this gate time and width - and its sampling of the Geb signal, will reduce the geb signal to almost zero.


It gulps a positve chunk and a negative chunk of ground signal and the net GEB ch o/p is nothing. This is good for us as this is a phase notch at one phase, all other phases come thru!!


If the initial Rx phase was off the ground signal would come thru at the wrong time. The Geb timing would be incorrect, as would the Disc timing. - You may not be able to reject iron also..

S

[Silver Dollar]

Hey thanks for the nulling info. That will be very usefull to me! I just finished my Fat TX (Minelab style with .5mm wire and 45 turns)
and have shielded the RX with some cool RF mesh shielding from work. Once I shield the TX I can try nulling permanently.

I have some wire to make the differential RX as I'd prefer to not use shielding if I can get away with it!

So I'm guessing we are trying to get a sample of the waveform that is equal parts above zero and below to balance to 0v?

So the GEB starts at 90 and goes to 270? OK but where is the Disc? It has to overlay the GEB to produce a sound...

Maybe the problem is the discriminate chart shows ferite at 0 and salt at 90 and most coins from 90 to 180, I imagine the
targets are also 180 from those locations, so ferrite is at 0 and 180 and salt is at 90 and 270 so the disc can run around the other
side?

[Silver Dollar]

Like this?

[Silver Dollar]

Well I wasn't sure my signals were correct but these pics come from my board. I was mistakenly thinking
the GEB and Disc had to be in the same position for the MD to output a tone but actually the signal is converted
to a DC level then the cap holds the sample for a bit so I guess these are correct?

[Silver Dollar]

"I made a phase shifting solution that I think will be just right for your TGSL upgrade. It would be nice if you zip your .sch file so that I can put in my suggested solution. I'll be off for the weekend, but I'll be able to fix it on Monday. It will have GEB span of over 90°to accommodate for beaches, including coarse and fine tune to avoid multiturn pots and speed up tuning for salty beaches. Disc tuning span can go 180°, but for practical reasons it can be reduced to a smaller span. Tuning is almost linear."

This sounds good. I zipped the file and put it in the other topic. I can see the benefits to that kind of phase span! I'll get the AB/C H bridge going this weekend. Maybe I won't need that PI unit after all!

[Davor]

... I imagine the targets are also 180 from those locations, so ferrite is at 0 and 180 and salt is at 90 and 270 so the disc can run around the other side?

That's precisely what happens with target signals going to counter-phase, and that's the reason I'm after 4-quadrant discrimination solution. For 2-quadrant you need AND function, while in 4-quadrant you need XNOR function, or some other clever approach.

[golfnut]

So I'm guessing we are trying to get a sample of the waveform that is equal parts above zero and below to balance to 0v? As I understand it.. This way you eradicate a single signal phase - that of the ground ( the soil it has to be uniform consistency to work properly )

So the GEB starts at 90 and goes to 270? That the general rule... OK but where is the Disc? It has to overlay the GEB to produce a sound... The Discc gate is later - It needs to sample phase ****ed signals that are correspondingly late.. You need the two Rx channels to be able to Rx target energy at the same time to trip the common reference on the comparator stage so there is some overlap or they wouldn't work. You would need a short duration sample and hold on the Geb CH if they were split up too much.. So the early result was still available for the Disc CH to tally with. This is taken from my IDX - shows the time split between the 2 channels. This picture is from a 250 I have.. again shows the split - this maybe a hint different as although similar architecture it does not have the comparator AND function it has A/Ds AND sw backend S

[Davor]

Works the same. In a 360° circle, GEB channel will have two zeros on the opposite sides, splitting the circle in two halves. Targets appearing in one half will have a positive response, while the other half negative. Same goes with Disc. The area where both GEB and Disc are positive is passed through (AND), while in case of 4-quadrant discrimination also the area where both are negative passes through (XNOR). Analogue or digital - works the same.