TGSL Tuning & troubleshooting

[simonbaker]

Ivconic posted something a while back and it's starting to make a little more sense now. While experimenting with my latest TGSL circuit on my work bench, I noticed that the proximity of the speaker to the coil has much to do with the audio. If the speaker is within about two feet of the coil, the circuit could go into a slow oscillation after passing metal over the coil. Likewise, the orientation of the speaker in relation to the coil has some bearing on the quality of the audio - short vs long, broken vs solid sound! Audio can be improved by making sure that the speaker coil stays perpendicular to the search coil! (Not in the same geometric plane) In my case, it's a small 2", 8 ohm magnetic speaker. Just another variable to be aware of when testing an troubleshooting.. There HAS to be some magnetic coupling between the two and hence some feedback. It's best to keep the two separated as much as possible..

I can't find the exact post, but Ivconic mentioned that once he put his circuit in an metal enclosure, the audio improved. It seems to support my observation.

I will attempt to put together a short video demonstrating.

Don

Yes, Ivconic and others spent quite a time trying to understand why audio was flaky, unpredictable, and different with headphones, speaker, jack, etc. You may have hit on something. The crazy thing is that the feedback that can cause beeps may actually improve sensitivity and audio when just below the threshold of beeping. But it is a wildcard factor. If there is a way to improve sensitivity/response, it would be better to do it in a controlled way. But I like what you have discovered, especially orientation.

Have you tried headphones and compared to the performance with a speaker on your TGSLs?

-SB

[dfbowers]

Yes, Ivconic and others spent quite a time trying to understand why audio was flaky, unpredictable, and different with headphones, speaker, jack, etc. You may have hit on something. The crazy thing is that the feedback that can cause beeps may actually improve sensitivity and audio when just below the threshold of beeping. But it is a wildcard factor. If there is a way to improve sensitivity/response, it would be better to do it in a controlled way. But I like what you have discovered, especially orientation.

Have you tried headphones and compared to the performance with a speaker on your TGSLs?

-SB

Oh yea, I do remember the post by Ivconic on the phone jack thing.. I have been using headphones with TGSL #1 so I forgot the symptoms I had while testing it until I build TGSL #2 and started experimenting with a hardwired speaker again. Now, is the reason that Tesoro uses off resonance coils to avoid funny things like this as well? Maybe when their final products are assembled and put into enclosures the synchronous audio is not a problem but testing on a bench can be? I guess by solving one problem and creating clean audio by driving the audio from the oscillator they can cause the potential for another.. Anyway, this probably doesn't matter if we all finish our projects.

Don

[ivconic]

Substantially; what is speaker? Coil! Coil driven by low frequency power output. So you have search head on one side and speaker ("tx-ed" coil) on the other side of pcb.... 3 divisions (by 4024) from main TX frequency are passed to audio out and "TX-ed" through "coil" (speaker)....fine..Only problem is in fact that those 3 divisions are in direct (harmonical) resonance with main frequency..... See what is the problem? I faced that long time ago...maybe at my second or third tgsl. Eversince than i tried to overcome it. TGS//TGSL is such concept and nothing we can do to overcome that issue....unless apply some major changes and possible change its charactrer too. Compromise is to pack it in metal enclosure and assure 90 degreees oriented relation between speaker and the rest of "hot" electronics. Couple ferrite perles will not harm to; tied on speaker leads...etc...etc...

I am still alive and i am still following this subject....but will not take a part in discussions ...unless if those were started by members i do respect for real (like this was the case)...
Cheers!

[dfbowers]

Substantially; what is speaker? Coil! Coil driven by low frequency power output. So you have search head on one side and speaker ("tx-ed" coil) on the other side of pcb.... 3 divisions (by 4024) from main TX frequency are passed to audio out and "TX-ed" through "coil" (speaker)....fine..Only problem is in fact that those 3 divisions are in direct (harmonical) resonance with main frequency..... See what is the problem? I faced that long time ago...maybe at my second or third tgsl. Eversince than i tried to overcome it. TGS//TGSL is such concept and nothing we can do to overcome that issue....unless apply some major changes and possible change its charactrer too. Compromise is to pack it in metal enclosure and assure 90 degreees oriented relation between speaker and the rest of "hot" electronics. Couple ferrite perles will not harm to; tied on speaker leads...etc...etc...

I am still alive and i am still following this subject....but will not take a part in discussions ...unless if those were started by members i do respect for real (like this was the case)...
Cheers!

I thought maybe you were listening..

Thanks!

Don

[simonbaker]

Oh yea, I do remember the post by Ivconic on the phone jack thing.. I have been using headphones with TGSL #1 so I forgot the symptoms I had while testing it until I build TGSL #2 and started experimenting with a hardwired speaker again. Now, is the reason that Tesoro uses off resonance coils to avoid funny things like this as well? Maybe when their final products are assembled and put into enclosures the synchronous audio is not a problem but testing on a bench can be? I guess by solving one problem and creating clean audio by driving the audio from the oscillator they can cause the potential for another.. Anyway, this probably doesn't matter if we all finish our projects.

Don

If a circuit doesn't work until it is on a minimal PCB and in an enclosure, I always wonder how the first designs were developed (you really got to trust your theory) - being a putzer, I like stuff that works on the breadboard, but MD circuits seem too sensitive to go too far that way, although I do have a mediocre version of the TGSL working on a solderless breadboard -- still very useful for some tests.

Yes driving audio from oscillator turns out to be really smart because most noise, spikes, etc from audio/speaker gets turned into a constant voltage by synchronous detector. Yet there is obviously some kind of feedback that causes those beeps - I'm thinking the "constant voltage" I referred to (from emi when speaker sounds) appears initially as a pulse that travels nicely through the very narrow low-pass amplifier section (approx 15 Hz low pass filter), and causes hiccuping if the gain is high enough. Sort of a whole separate sub-band feedback path.

By "off-resonance" coils, if you mean RX coil resonance being quite higher than TX oscillator -- I think reason for that is phase stability. If you get near RX resonance, the phase shifts dramatically with small frequency changes. But boy do you get a lot more gain! Ultimately I'm trying to work out some way to take advantage of that. It potentially could kick @$$. The discrim circuit would need to be redesigned because any target signal near resonance is offset quite a bit phase-wise from where we see it now, I believe.

Your observations as usual are quite helpful to anyone trying to get a good build of one of these beasts. Thanks.

-SB

[nakky]

Substantially; what is speaker? Coil! Coil driven by low frequency power output. So you have search head on one side and speaker ("tx-ed" coil) on the other side of pcb.... 3 divisions (by 4024) from main TX frequency are passed to audio out and "TX-ed" through "coil" (speaker)....fine..Only problem is in fact that those 3 divisions are in direct (harmonical) resonance with main frequency..... See what is the problem? I faced that long time ago...maybe at my second or third tgsl. Eversince than i tried to overcome it. TGS//TGSL is such concept and nothing we can do to overcome that issue....unless apply some major changes and possible change its charactrer too. Compromise is to pack it in metal enclosure and assure 90 degreees oriented relation between speaker and the rest of "hot" electronics. Couple ferrite perles will not harm to; tied on speaker leads...etc...etc...

I am still alive and i am still following this subject....but will not take a part in discussions ...unless if those were started by members i do respect for real (like this was the case)...
Cheers!

Hi Ivconic,happy to see you.

Grt Nakky

[erolunall]

Hi İvconic

I think everybody miss you & your excellent works over MD.

Geotech forums are going very slowly without you.

Best Wishes
Erol

[Stefano]

If you get near RX resonance, the phase shifts dramatically with small frequency changes. The discrim. circuit would need to be redesigned because any target signal near resonance is offset quite a bit

I had little time for testing my TGSL last night and I also noticed the same thing.
My homebuilt coils are 5.7 and 6.3 unshielded at the moment. I played with the Rx tuning cap in a wide range trying to find the value wich gave me the +20° offset at minimal nulling. I found that what I wanted was just the minimum residual voltage, not the correct phase. In my case I was able to get a good and stable working even with -20°/-40° if coils adjusted for right nulling!
In neither case I was able to get more than 16-18cm for a 10 euro cent coin (it's a small coin), but I could reject ferrite in all metal with the trimmer set to about mid-range and disc mode fully rejected iron.
By changing the cap to have a closer tx-Rx tuning freq I could get some 2-3 cm more but it became extremely noisy and couldn't reject ferrite anymore and got iron always detected in disc mode, so I suppose that this means that the phase behaviour of targets changes with tx-rx tuning frequency.

[simonbaker]

I had little time for testing my TGSL last night and I also noticed the same thing.
My homebuilt coils are 5.7 and 6.3 unshielded at the moment. I played with the Rx tuning cap in a wide range trying to find the value wich gave me the +20° offset at minimal nulling. I found that what I wanted was just the minimum residual voltage, not the correct phase. In my case I was able to get a good and stable working even with -20°/-40° if coils adjusted for right nulling!
In neither case I was able to get more than 16-18cm for a 10 euro cent coin (it's a small coin), but I could reject ferrite in all metal with the trimmer set to about mid-range and disc mode fully rejected iron.
By changing the cap to have a closer tx-Rx tuning freq I could get some 2-3 cm more but it became extremely noisy and couldn't reject ferrite anymore and got iron always detected in disc mode, so I suppose that this means that the phase behaviour of targets changes with tx-rx tuning frequency.

That makes sense to me what you say.

I think many things can affect the null signal phase, such as nearby metal while you are nulling, the type of shielding, various capacitance of cable, coils, etc. So I suspect the exact phase of null signal is not critical, better to null by other criteria as you found.

For TGSL, the null phase does has some importance because if completely wrong, it will cause a negative voltage at the capacitors C15, C12 at the source side of the JFet detectors. This makes it hard for the JFets to work properly because they might be forward biased part of the time.

I hope to experiment some more with Tx, Rx frequency closeness, but I think the specs for TGSL are as good as possible without changing the design.

16 -18 cm seems pretty good for small coin. What do you get for 1 Euro?

-SB

[Stefano]

Hi SB. I tried this afternoon: with 1 euro unfortunately not much more: 18-20 cm. And I've built a 10" DD, not a 8" !!!
I also noticed that: If the Rx resonance freq. is higher than Tx (as it is supposed to be in the TGSL design) the phase at null is always negative (so not +20° but instead -20/-40°). If the Rx freq. is lower, the phase at null point is positive.
With negative phases I was able to reject ferrite and disc worked fine, with positive I had more noise and disc and geb were sort of "out of tuning" and didn't work correctly, but got more depth.
Pay attention that 1KHz less as Rx resonance (13.5 Kz) can be obtained only with 20 and more nF, so I think too much far from the given 15nF value!!!
And also: shifting from a higher to a lower rx freq than tx one should reverse coil leads.

Apart of understand how coil must be set I might have another issue in my circuit around U106, because I noticed that turning the sensibility trimmer had no effect, but I will investigate on it in future. I hope to better improve the depth there...

[simonbaker]

Hi SB. I tried this afternoon: with 1 euro unfortunately not much more: 18-20 cm. And I've built a 10" DD, not a 8" !!!
I also noticed that: If the Rx resonance freq. is higher than Tx (as it is supposed to be in the TGSL design) the phase at null is always negative (so not +20° but instead -20/-40°). If the Rx freq. is lower, the phase at null point is positive.
With negative phases I was able to reject ferrite and disc worked fine, with positive I had more noise and disc and geb were sort of "out of tuning" and didn't work correctly, but got more depth.
Pay attention that 1KHz less as Rx resonance (13.5 Kz) can be obtained only with 20 and more nF, so I think too much far from the given 15nF value!!!
And also: shifting from a higher to a lower rx freq than tx one should reverse coil leads.

Apart of understand how coil must be set I might have another issue in my circuit around U106, because I noticed that turning the sensibility trimmer had no effect, but I will investigate on it in future. I hope to better improve the depth there...

Usually null is taken slightly to side of minimum point, not exactly on it. I think Qiaohzi said that you choose the side with more coil overlap. Is that what you do?

- Is there any metal near your coil (including oscilloscope, table legs, etc.) when you null?

- Where do you measure Rx voltage and phase at?

I have also played with rx resonant freq lower than tx by raising tx freq higher (17 - 20 kHz). I noticed similar need to reverse leads. And detector worked, even though a bit different! I don't know what it means though at this point, if there is a strong reason that Rx resonance should be greater than Tx freq or not. Maybe both designs possible.

Sensitivity pot at U106 is tricky. Sometimes you just notice more noise beeps when you turn it. Also, it requires precision op amps for U104 because their quiescent output should be very near zero volts. You can check by disconnecting RX coil. Outputs of U104 should have DC voltage very near zero (hopefully within a millivolt).

In my opinion, a good sensitivity circuit should be able to make noise at some point when turning pot. If you have no noise ever, then perhaps you can improve the gain somewhere.

Note: the sensitivity pot circuit has minimum threshold of about 5 millivolts (I think) at input of LM393, so you cannot turn it more sensitive than that (unless you jumper the 1k resistor R37).

-SB

[Stefano]

Rx output was taken at preamplifier output (pin 7 IC 101a)
Thaks for hints sb, I'll try a.s.a.p.

[simonbaker]

waveforms

This is for dfbowers and anyone with good working TGSL:

I am interested in seeing the waveforms on a scope for:

1. TX oscillator
2. Pin 1 LF353 (disc driver output)
3. Pin 1 of U102 LM393 (output to JFet gate) when disc pot is turned all the way up.

4. Also measurement of DC voltage at source of JFet TR4 (disc channel) when disc pot is turned all the way up.

Regards,

-SB

[simonbaker]

TGSL reference test data

Here are some useful TGSL reference data, courtesy of dfbowers, who has good working MDs which air test 1 euro in the 29 - 30 cm range, to help troubleshoot our TGSL MDs. I believe he made the TGSL Final version.

Tx amplitude at J1-1: 17V p-p

Background noise level: about 40mV peak-to-peak. (LM308s pin 6)

DC voltage range across C15: .1 (min, ccw) to -.2 (max, cw) for disc pot.

DC voltage range across C12: .18(min, ccw) to .25(near max cw) then drops to almost 0v at full cw setting on GB pot.

Notes on nulling and setting GB pot: I don't spend too much time nulling. I just hook up a DVM to the Rx wires with a 15nF cap in series. Then adjust coils to a minimum. It usually matches what I see on the scope on LF353 pin 7.

Setting GB : IMHO, this is the setting that is critical! I look at u106b pin5 and adjust pot for minimal response to a ferrite rod. Too high a setting and it reduces detection range significantly and looses sensitivity to high conductors (like thin silver coins). Too low a setting and the TGSL may pick up the ground. Speaker should barely "crack" when a ferrite rod is passed over coil.

I think he will add some oscilloscope photos showing phase of TX/RX null signals.

Some other interesting observations he has made:

A final thought.. There is no reason that you cannot obtain a target distance of 29-30cm with these coils. I have verified that you can. Just as a comparison to my White's XLT, Nautilus DMCIIb, Shadow X5 and Tesoro Cibola.. Air tests are all different. X5 gets 15" on a Nickel. In the ground, it's probably the worst! Oddly enough, Tesoro Cibola is the deepest for the price! (Clay soils sucks..)

I have 12 .58 Cal Minie balls buried in my back yard from 1 to 12". NONE of my detectors hits anything deeper than 6". TGSL gets 5" for under 50 bucks!!

Nautilus DMCIIb has an adjustable Tx from 5 to 55V, a preamp in the coil and a coil balancing system (plus GB). Depending upon ground conditions, it goes deepest with reduced Tx power.. .So there is something to be said about low power transmitters in my opinion.

Thanks to dfbowers!

-SB

[simonbaker]

dfbowers test

Hi dfbowers -

Would you test how the sensitivity pot affects your air depth tests using setup similar to your video?

What is depth with min sensitivity.

What is depth with max sensitivity.

Would be interested in same tests by anyone with good working TGSL.

Regards,

-SB