TGSL Tuning & troubleshooting

[simonbaker]

On the contrary ... you must get the initial phase-shift set correctly, otherwise your coil will be out of the range of the ground balance adjustment, and your detector will be unusable in the field.

Ok, we disagree on that so far (at least for TGSL). I believe the null phase is completely independent of the target phase, so null phase doesn't matter. The GB and DISC circuits are basically the same -- they work against the target signal phase, not the null phase. The GB detector is just the same as the DISC detector, only approx 90 deg shift. The phase of both the DISC and GB detectors is determined by the TX signal and not the null (of course). It is true that both the DISC and GB detectors pick up the null signal, but they convert it to a constant voltage, which is essentially ignored.

I want to actually try to test and confirm the effects of the null settings for TGSL. We're trying to troubleshoot our detectors and we need to know for sure about this null.

Regards,

-SB

[Stefano]

Your negative phase can also be considered as a positive phase if you add it to 360 degrees.

Yes, of course but I think it's easier and more obvious say -20° instead of +340.

I have a 9x8 web coil, 12x10 web coil, brown polo, widescan, Laser polo, and a 4" concentric. They all have a phase-shift of around 20 degrees, except for the widescan, which has about 200 degrees (same as a DD).

Ok. I believe you, but since I know you have not realized the TGSL, I suppose that you have measured the phase with a scope directly on the coil connector when the coil is connected to a factory made detector. If it's like that this could lead thinking that some value shown in TGSL project is not correct.

If you look at Molzar's post, there is something odd going on there. Figure F2 - see below - shows signal 1 (TX) and signal 2 (output of preamp). On the right it states that the phase difference is -20 degrees (or +340 degrees), but the scope picture shows something completely different. Look at points A and B, they are not separated by -20 degrees. The same goes for F1.

I think you're misunterstanding Molzar's post, as I did at the beginning.

If only he is listening I hope he pops up to better explain.
As far as I have understood he presented in the pdf file 2 versions:

- the one on the left column (namely the one with traditional scope images) is Max's TGSL circuit equipped with his own DD coil.
I think Max posted those picture well before Molzar's post.

- the version on the right colomn (namely the one with digital scope b&w images) come from a standard Tesoro Golden Sabre equipped with a concentric coil. As Molzar said the 2 version were posted for comparison, because there were obvious differences between the two.

There is also clearly something wrong with the ground balance in F7, F8, F9 and F10. The sample pulse is not located over the zero-crossing of the RX wave in any position.

Yes I agree, but you're referring to Max's, nevertheless Max is one of those who has a "30 cm" TGSL version.

[Qiaozhi]

Ok. I believe you, but since I know you have not realized the TGSL, I suppose that you have measured the phase with a scope directly on the coil connector when the coil is connected to a factory made detector. If it's like that this could lead thinking that some value shown in TGSL project is not correct.

I have my own detector design that can use standard Tesoro coils.

Until you guys understand how the GEB and DISC channels actually work, you will continue thrashing around in the dark. Here's a clue:

Monitor the output of the synchronous demodulator in the GEB channel with an oscilloscope, with the sample pulse situated over the zero-crossing of the RX signal. Then wave a ferrite slug in front of the coil and watch what happens on the scope. Next try moving the GEB pulse either side of the zero-crossing and repeat the process. Once you understand this you'll understand how ground balance works.

[Stefano]

Yes, Stefano is correct. The equivalent circuit shows that RX coil is part of Second Order Low Pass Filter. That means we can change the phase lag of receved signal V from 0 to 180 deg if we change the equivalent capacitance C=C1+C2. If you don't believe me, put the following equivalent circuit in AC SPICE analysis. The voltage E induced in RX coil is in phase lead 90 deg relative to TX current. However, the TX coil voltage is not in quarature with TX current because TX coil has resistance and we measure voltage on its capacitance.
CONCLUSION: You can adjust phase lag in very wide range by changing capacitance of C2.

Unfortunately the only simulator which I'm able to use is CircuitMaker. I'll try with it a.s.a.p., however I'm not saying that the +20° phase is not achievable, but that to get it I'd have to use cap values much too far from the prescribed one and with that the circuit dos not behave well anymore in all its aspects.

[simonbaker]

I have my own detector design that can use standard Tesoro coils.

Until you guys understand how the GEB and DISC channels actually work, you will continue thrashing around in the dark. Here's a clue:

Monitor the output of the synchronous demodulator in the GEB channel with an oscilloscope, with the sample pulse situated over the zero-crossing of the RX signal. Then wave a ferrite slug in front of the coil and watch what happens on the scope. Next try moving the GEB pulse either side of the zero-crossing and repeat the process. Once you understand this you'll understand how ground balance works.

Hi Qiaozhi:

I think it doesn't matter what you see on scope. That is misleading. Target signal is added to null signal (superposition), so it looks different with different null points, but target signal is actually the same.

To truly test, see where the DISC knob knocks out a nickel when you change the null. Or, watch the output of LM308 ground balance channel and see where (on the GB pot setting) the direction of the ferrite signal changes as you change the GB pot. Then pick different null phase and try again. I think it will not change at all as you change the null phase.

Regards,

-SB

[Stefano]

I have my own detector design that can use standard Tesoro coils.

Until you guys understand how the GEB and DISC channels actually work, you will continue thrashing around in the dark. Here's a clue:

Monitor the output of the synchronous demodulator in the GEB channel with an oscilloscope, with the sample pulse situated over the zero-crossing of the RX signal. Then wave a ferrite slug in front of the coil and watch what happens on the scope. Next try moving the GEB pulse either side of the zero-crossing and repeat the process. Once you understand this you'll understand how ground balance works.

.

I'll try.

[simonbaker]

Yes, Stefano is correct. The equivalent circuit shows that RX coil is part of Second Order Low Pass Filter. That means we can change the phase lag of receved signal V from 0 to 180 deg if we change the equivalent capacitance C=C1+C2. If you don't believe me, put the following equivalent circuit in AC SPICE analysis. The voltage E induced in RX coil is in phase lead 90 deg relative to TX current. However, the TX coil voltage is not in quarature with TX current because TX coil has resistance and we measure voltage on its capacitance.
CONCLUSION: You can adjust phase lag in very wide range by changing capacitance of C2.

Hi mikebg:

You're right, changing resonant frequency of RX coil | C6 tank will change the phase of received signals, particularly if you move resonant freq closer to TX frequency -- but actually we are talking about the phase of the null signal, which is easily changed by slight adjustment of the double-D coils. My point is that you can change null phase a lot without changing the received target signal phase when you nudge the coils. We aren't messing with capacitor C6 and really shouldn't, because if too far off it will affect the target signal phase.

Hope I'm not adding confusion. But I'm going to hammer this point because I think it is particularly important for this thread (tuning and troubleshooting), and I think it is possible to have weird null phase with no problem and trying to fix it is waste of time. But null phase may also be a warning that something else is wrong, like TX frequency is too close to RX resonant freq, etc.

If I'm wrong, I'm wrong. But I need to hammer it thoroughly first.

-SB

[simonbaker]

.

I'll try.

As mentioned, that won't prove it. Check disc of nickel.

-SB

[Qiaozhi]

Hi Qiaozhi:

I think it doesn't matter what you see on scope. That is misleading. Target signal is added to null signal (superposition), so it looks different with different null points, but target signal is actually the same.

To truly test, see where the DISC knob knocks out a nickel when you change the null. Or, watch the output of LM308 ground balance channel and see where (on the GB pot setting) the direction of the ferrite signal changes as you change the GB pot. Then pick different null phase and try again. I think it will not change at all as you change the null phase.

Regards,

-SB

Sorry, but you've missed the point.

Second clue:
Why do you think that detectors with an external Ground Balance control require you to bob the coil up and down while you adjust the balance?

[simonbaker]

Sorry, but you've missed the point.

Second clue:
Why do you think that detectors with an external Ground Balance control require you to bob the coil up and down while you adjust the balance?

I may have missed point. Let's see.

You bob coil up and down because it's easier than moving planet up and down...

Seriously, the ground is just another target, totally different from null signal. Like a piece of ferrite. The phase signal from the ground is similar to a piece of ferrite. You adjust phase of GB detector until the GB channel drive signal is centered on the zero crossing of the ferrite signal, not the null signal. You can't see the ferrite signal well on a scope (buried under the null signal), but it is there and independent of the null signal. But the output of the LM308 will clearly show the ferrite signal only -- it ignores the null signal completely.

And, as you turn the GB pot, when the output of the GB channel LM308 stays flat when you wave the ferrite (or pump the detector over the ground), you have achieved ground balance.

To be more specific, you have probably "minimally ground-balanced", because if you turn the GB knob further, the ferrite signal will return but this time it will move opposite to the signal in the DISC channel and still won't make a beep. I believe other detectors use a special mode to find the "minimal ground balance" point where the signal disappears at the minimal point, but you hear it on either side. With the TGSL, I believe you will only hear it on one side if you pump the coil.

Regards,

-SB

[Qiaozhi]

My point is this:
Forget about the value of the residual voltage and the initial phase-shift at the coil for the moment. Simply set the DISC control to minimum, and the GEB control to mid-position. Then monitor the preamp output and the sample pulse into the sync demod of the GEB channel. Adjust the coil overlap to roughly position the sample pulse over the zero-crossing of the RX signal. At this point you have the correct setup, because this now gives sufficient range for the GEB to be adjusted. Now you can properly ground balance by either using a ferrite slug, or taking the detector outside and bobbing the coil up and down. If you're interested, you can then measure both the residual voltage and the phase-shift. Then when you make another identical coil you can adjust the overlap to achieve the same phase-shift. If you read Ivconic's notes, you will see that he does a similar thing by monitoring the residual voltage.
Of course, you could start from a different null point, but you could find your GEB control needs to be at one end instead of near the middle. Also, the DISC control may not be calibrated correctly with iron rejection at the minimum setting.

[dfbowers]

Wet grass problem fixed!

Over the last 2 weeks, I built 2 more coils, trying to independently determine what the best shielding and grounding configuration might be to deal with wet grass. Having 3 coils that worked very well, they all suffered when the sun went down and the lawn was dew covered.

I put all theory aside and went with some trial and error.. Plus what I thought would make sense. A little costly I think but did come to a solution which works best for me.. (your milage may vary).

Never mind what I tried and did not work.. it doesn't matter. What finally DID work is this:

A USB-2 cable, using all 5 conductors (2 power, 2 signal + ground).

* RED and BLACK wires to run the Tx coil (since they are a heavier gauge).

* GREEN and WHITE for the receive circuit.

* Rx and Tx shields tied together in the coil and wired to cable shield.
Cable shield is then tied to ground on PCB.

* Rx and Tx coils NOT connected to ground inside the coils at all.
In other words, all active parts of the coils do not touch the grounding
system in any way.. Only the cold end of the Tx circuit is connected to
ground on the PCB.

Null was less than 100 mV at pin 7 of U101a but climbed to 300mV when epoxy cured.

Best airtest without chatter on 1e is 31cm with latest coil.

Best airtest without chatter on 1e with coils wired per TGSL coil making by Ivconic is 35cm (indoors).

Depth test in hard packed clay test garden. 16cm for .58 Cal minie ball. Same for all 3 coils .. about the same as my Shadow X5. ($935.00 machine).

I know that some post will agree with my wiring configuration, some will not.. These are just my findings.

I also vacuum formed a new shell and managed to shave 3 ounces off of my previous coil (the one closest to the bottom of the picture..
Don

[dfbowers]

Over the last 2 weeks, I built 2 more coils, trying to independently determine what the best shielding and grounding configuration might be to deal with wet grass. Having 3 coils that worked very well, they all suffered when the sun went down and the lawn was dew covered.

I put all theory aside and went with some trial and error.. Plus what I thought would make sense. A little costly I think but did come to a solution which works best for me.. (your milage may vary).

Never mind what I tried and did not work.. it doesn't matter. What finally DID work is this:

A USB-2 cable, using all 5 conductors (2 power, 2 signal + ground).

* RED and BLACK wires to run the Tx coil (since they are a heavier gauge).

* GREEN and WHITE for the receive circuit.

* Rx and Tx shields tied together in the coil and wired to cable shield.
Cable shield is then tied to ground on PCB.

* Rx and Tx coils NOT connected to ground inside the coils at all.
In other words, all active parts of the coils do not touch the grounding
system in any way.. Only the cold end of the Tx circuit is connected to
ground on the PCB.

Null was less than 100 mV at pin 7 of U101a but climbed to 300mV when epoxy cured.

Best airtest without chatter on 1e is 31cm with latest coil.

Best airtest without chatter on 1e with coils wired per TGSL coil making by Ivconic is 35cm (indoors).

Depth test in hard packed clay test garden. 16cm for .58 Cal minie ball. Same for all 3 coils .. about the same as my Shadow X5. ($935.00 machine).

I know that some post will agree with my wiring configuration, some will not.. These are just my findings.

I also vacuum formed a new shell and managed to shave 3 ounces off of my previous coil (the one closest to the bottom of the picture..
Don

Attached cable diagram..

[simonbaker]

My point is this:
Forget about the value of the residual voltage and the initial phase-shift at the coil for the moment. Simply set the DISC control to minimum, and the GEB control to mid-position. Then monitor the preamp output and the sample pulse into the sync demod of the GEB channel. Adjust the coil overlap to roughly position the sample pulse over the zero-crossing of the RX signal. At this point you have the correct setup, because this now gives sufficient range for the GEB to be adjusted. Now you can properly ground balance by either using a ferrite slug, or taking the detector outside and bobbing the coil up and down. If you're interested, you can then measure both the residual voltage and the phase-shift. Then when you make another identical coil you can adjust the overlap to achieve the same phase-shift. If you read Ivconic's notes, you will see that he does a similar thing by monitoring the residual voltage.
Of course, you could start from a different null point, but you could find your GEB control needs to be at one end instead of near the middle. Also, the DISC control may not be calibrated correctly with iron rejection at the minimum setting.

Hi Qiaozhi:

Yes and no.

Your choice for the null is good because it gives you a reasonable voltage range on the GB JFet due to capacitor C12 while you adjust the GB pot. I would go so far as to recommend a null that keeps the voltage on capacitor C15 fairly non-negative over the Disc pot range.

But you have not addressed the actual ground balancing of the target signal. The null signal phase does not affect the phase range at all (significantly) of the GB or DISC.

I have tested this many times and I just repeated it. I set the null on one side of the minimum, then on the other, so the RX null signal is about 180 deg different in each case. And the ground balance for ferrite is the same in each case. Same for the discrimination of a US Nickel. The null does not affect it. Not for my TGSL, maybe for other detectors.

---------

It's what the math says also. When a target is present, the Rx signal is the sum of the null signal (A) and the target signal (B), two independent signals.

Vrx(t) = Asin(wt + phi1) + Bsin(wt + phi2).

If you put them into a synchronous detector, the phi1 of the first signal does not affect the detection of the second signal. Only the amplitude and phase of the second signal determines its response.

When we ground balance, we adjust the pot to null out the signal of phase phi2 -- it doesn't matter what phi1 is.

The null signal [Asin(wt + phi1)] makes a constant voltage on C12 and C15 which never appears at the output of LM308. It simply does not affect the detection of the ferrite; only the relative phase of the ferrite signal and the ground balance sync pulse make the audio beep that we hear during ground balancing.

For me, theory and tests show the same thing. The null phase does not matter, at least not for the reasons that have been often stated. Instead I think it is useful for keeping the JFets forward biased or for achieving maximum coil overlap, etc. I'm still open to other evidence, but I need to see it.

Regards,

-SB

[simonbaker]

Over the last 2 weeks, I built 2 more coils, trying to independently determine what the best shielding and grounding configuration might be to deal with wet grass. Having 3 coils that worked very well, they all suffered when the sun went down and the lawn was dew covered.

I put all theory aside and went with some trial and error.. Plus what I thought would make sense. A little costly I think but did come to a solution which works best for me.. (your milage may vary).

Never mind what I tried and did not work.. it doesn't matter. What finally DID work is this:

A USB-2 cable, using all 5 conductors (2 power, 2 signal + ground).

* RED and BLACK wires to run the Tx coil (since they are a heavier gauge).

* GREEN and WHITE for the receive circuit.

* Rx and Tx shields tied together in the coil and wired to cable shield.
Cable shield is then tied to ground on PCB.

* Rx and Tx coils NOT connected to ground inside the coils at all.
In other words, all active parts of the coils do not touch the grounding
system in any way.. Only the cold end of the Tx circuit is connected to
ground on the PCB.

Null was less than 100 mV at pin 7 of U101a but climbed to 300mV when epoxy cured.

Best airtest without chatter on 1e is 31cm with latest coil.

Best airtest without chatter on 1e with coils wired per TGSL coil making by Ivconic is 35cm (indoors).

Depth test in hard packed clay test garden. 16cm for .58 Cal minie ball. Same for all 3 coils .. about the same as my Shadow X5. ($935.00 machine).

I know that some post will agree with my wiring configuration, some will not.. These are just my findings.

I also vacuum formed a new shell and managed to shave 3 ounces off of my previous coil (the one closest to the bottom of the picture..
Don

That's very interesting. Maybe worth having two different kinds of coils depending on whether grass or not. Would you prefer original (Ivconic) wiring for beach for example?

I like idea of red and black for TX - also because white and green often twisted pair in good USB cables, good for low noise for RX.

-SB