TGSL Tuning & troubleshooting

[simonbaker]

Cool! Thanks Simon.. Now you might not think that I'm nuts anymore. . Divide and conquor!! Now we need to come up with a noise reduction scheme.. Isn't that strange that detection range just sucks sometimes for no apparent reason? We all need to build Faraday cages around our workshops!! I was in a lab once in Harrisburg that had one.. It looked like a chicken coop.

Don

Ha! Never doubted it for a minute... relieved also that I didn't put a whammy on it somehow...

I seriously have thought about a Faraday cage... and maybe raising chickens too. I wonder how big it would have to be to not interact with a coil...

Of course Qiaozhi's advice to not take air tests too seriously is a good one and something to be tested if possible. The real test is in the ground. But I look at it this way:

First, getting a successful air test of around 30 cm means you have at least reproduced the design as others have and gives you confidence your parts, wiring, and coil building are normal.

Second, although a "noisy" ground may obliterate the very small signals that make the difference between 20 and 30 cm air test, there may be occasions where the ground is very favorable and an optimally built TGSL will pay off.

Third, I would ask, is it possible for a TGSL that air tests at 20 cm to actually hunt deeper in soil than a TGSL that air tests at 30 cm? Perhaps yes, and if so, we should look into why and how. But our instinct is, a 30 cm TGSL should do as well as better than a 20 cm air-tester. Still, if they are virtually the same in all ground, then I agree, don't sweat it. My own feeling is that the beach is a special type of ground well-suited to deep detecting and I would want to know I can get 30 cm with my TGSL.

Since eventually I would like to experiment with even deeper and more sensitive designs, I'm a little stymied about how to efficiently do that, given the noise limitations in my work area. Right now it looks like it would involve a complete wire-up and trucking to a remote area for testing. Probably not too feasible in pre-retirement days. I'll do what I can. Maybe I'll end up in a cage one way or another.

-SB

[dfbowers]

First time I went to test dfbowers' MD in a noise free mountain area, I couldn't find the road.

Second time, I forgot my Euro.

Third time, I tried hard not to forget anything (is that why I have headache afterwards...?).

Purpose was to air-test dfbowers TGSL in noise-free environment, and also my PCB with dfbowers coil.

Took:

- dfbowers MD
- my PCB, battery pack
- Euro, ferrite, other targets
- ruler to measure air depth
- screwdrivers in case need to adjust something
- meter in case need to check my PCB
- headphones
- camera, tripod
- board to put it all on
- swiss army knife in case of bears or other need

Found a remote location off side of road and set it all up. dfbowers MD horizontal with coil vertical on board, ruler lying flat.

I planned to make a video of test. I realized that I should have taken a long extension cord for headphones to get them near the camera to record the sound... next time.

Results:

Bingo! dfbowers MD just gets 30 cm with sensitivity control in mid position. If I turn sensitivity pot a little further, just to edge of chatter, it appeared to get well over 30 cm, maybe 33 or 34. My ruler stopped at 30 (note: get longer ruler...).

Now it wouldn't be right if everything went as planned. When I went to hook my PCB to dfbowers coil, I discovered battery pack was hot as a pistol -- it had been shorting out the whole time due to temporary wiring touching something. I guess I should be grateful it hadn't started a fire in the trunk of my car, but my words were not those of gratitude at the time. Guess I'll be seeing those mountains again....

---------------
Conclusions:

This is a major result as far as I'm concerned. dfbowers TGSL gets around 20 cm in my workshop, but at least 30 cm away from noise.

That tells me how important it is to test away from noise. I could spend the rest of my life trying to get 30 cm in my workshop, nulling, tweaking, changing chips, changing cables, messing with coils... but it wouldn't mean squat because of noise. (Of course I need to keep testing this.)

The amazing thing is that you don't hear that much noise from the MD in my workshop with dfbowers TGSL set to mid sensitivity, but it apparently can kill the weak target signals quite effectively. (possible theory is it breaks signal into small pieces that can't get through the next low pass filter). Of course it also limits the sensitivity knob from going much past mid position, but it is quite dramatic how it masks the weak target signals for all positions.

So when people report 20 cm, etc. it is important they redo the test outdoors away from noise. If you have trouble finding a good place, there is always dfbowers basement...

Due to cars going by I only have very short video snippets. I will try to put on youtube, but maybe can get better next time.

-SB

Cool! Thanks Simon.. Now you might not think that I'm nuts anymore. . Divide and conquor!! Now we need to come up with a noise reduction scheme.. Isn't that strange that detection range just sucks sometimes for no apparent reason? We all need to build Faraday cages around our workshops!! I was in a lab once in Harrisburg that had one.. It looked like a chicken coop.

Don

[simonbaker]

Mountain Test

First time I went to test dfbowers' MD in a noise free mountain area, I couldn't find the road.

Second time, I forgot my Euro.

Third time, I tried hard not to forget anything (is that why I have headache afterwards...?).

Purpose was to air-test dfbowers TGSL in noise-free environment, and also my PCB with dfbowers coil.

Took:

- dfbowers MD
- my PCB, battery pack
- Euro, ferrite, other targets
- ruler to measure air depth
- screwdrivers in case need to adjust something
- meter in case need to check my PCB
- headphones
- camera, tripod
- board to put it all on
- swiss army knife in case of bears or other need

Found a remote location off side of road and set it all up. dfbowers MD horizontal with coil vertical on board, ruler lying flat.

I planned to make a video of test. I realized that I should have taken a long extension cord for headphones to get them near the camera to record the sound... next time.

Results:

Bingo! dfbowers MD just gets 30 cm with sensitivity control in mid position. If I turn sensitivity pot a little further, just to edge of chatter, it appeared to get well over 30 cm, maybe 33 or 34. My ruler stopped at 30 (note: get longer ruler...).

Now it wouldn't be right if everything went as planned. When I went to hook my PCB to dfbowers coil, I discovered battery pack was hot as a pistol -- it had been shorting out the whole time due to temporary wiring touching something. I guess I should be grateful it hadn't started a fire in the trunk of my car, but my words were not those of gratitude at the time. Guess I'll be seeing those mountains again....

---------------
Conclusions:

This is a major result as far as I'm concerned. dfbowers TGSL gets around 20 cm in my workshop, but at least 30 cm away from noise.

That tells me how important it is to test away from noise. I could spend the rest of my life trying to get 30 cm in my workshop, nulling, tweaking, changing chips, changing cables, messing with coils... but it wouldn't mean squat because of noise. (Of course I need to keep testing this.)

The amazing thing is that you don't hear that much noise from the MD in my workshop with dfbowers TGSL set to mid sensitivity, but it apparently can kill the weak target signals quite effectively. (possible theory is it breaks signal into small pieces that can't get through the next low pass filter). Of course it also limits the sensitivity knob from going much past mid position, but it is quite dramatic how it masks the weak target signals for all positions.

So when people report 20 cm, etc. it is important they redo the test outdoors away from noise. If you have trouble finding a good place, there is always dfbowers basement...

Due to cars going by I only have very short video snippets. I will try to put on youtube, but maybe can get better next time.

-SB

[lunamay]

thank you simon you helped me a lot..yesturday i was trying to repeir broken chanel and all broke was in lf353..i replaced it and its work..now i wil try to balance a coil because i made a beautifull coil with great parameters 6.000mH and 6.4998mH so my coils seem to be good..thery are also shealded ..i must ake some mesurements..i will post them as soon as posile

best regards

[simonbaker]

Hi lunamay, glad you're still working for Ivconic.

The ALL METAL mode uses same JFet as Disc, so I don't think your problem is there.

Some questions to help us:

1. How well does you TGSL work in ALL METAL mode? What air depth?

2. I would check your .015u capacitor C6 (the one in parallel with the RX coil). Have you been playing with that? Have you changed to a different value? Have the solder joints broken?

3. Unhook the RX coil and measure the capacitance of capacitor C6 and let us know.

4. Measure the RX coil inductance and let us know.

5. Measure the oscillator frequency and let us know.

6. Do your coils have shields? Check the ground wire from the cable to make sure it really contacts the shields. Use an ohmmeter.

7. If your coils have shields, do they have correct gaps so they are not continuous loops? Are the two coil shields insulated from each other so they do not form a continuous loop?

8. To balance/null the coils, I refer to Qiaozhi's advice. I would measure null voltage at pin 8 (relative to ground) of LF353 (the output) because easier. Pins 5 and 6 I'm not sure, probably OK, but much smaller voltage expected.

Since you don't have scope, this is how I would proceed for Qiaozhi's method:

A) Adjust your coils overlap until AC voltage at LF353 pin 8 (or pins 5, 6) is approximately minimum.

B) Set the GB pot in mid position.

C) Put your DC voltmeter probes across capacitor C12 (connected to source of JFet TR5, GB channel), use millivolt scale, put ground probe to ground, and make final small adjustment to your coil overlap so the DC voltage (relative to ground) on C12 is zero millivolts or as close as you can get it. If not exact, I would choose slightly positive.

D) I don't know if this step will help or not. You can take a piece of ferrite and wave it around 7 cm in front of coils, and adjust the GB pot to remove any clear beep you hear. Some residual beep may be OK.

If you change the GB pot much, go back to step C and repeat.

Let us know how it goes!

Regards,

-SB

Hi lunamay:

There is be a little more to the nulling and I don't know how practical for you. When you null your coil, on either side of the minimum the null phase is a little more "stable", so I think you want to pick one of those positions. Qiaozhi once said the correct side is the side with more overlap. I have no opinion at the moment, need to understand it better. Let others advise you on that.

You might try finding the minimum with your AC voltmeter and then push the coils a tiny bit more together until the voltage rises 30%. Just a guess.

Regards,

-SB

[simonbaker]

Simon,

I was primarily interested in how the Spice simulation depicts the basic operation of the TGSL with an ungrounded Rx circuit in contrast to a grounded one.

Don

A stab at it here...

http://www.geotech1.com/forums/showthread.php?p=114710#post114710

-SB

[simonbaker]

Simon,

I was primarily interested in how the Spice simulation depicts the basic operation of the TGSL with an ungrounded Rx circuit in contrast to a grounded one.

Don

The short answer is I don't know the answer, I would want there to be not much difference. I'll check if I have any sims that cover that area.

I tend to simulate the TGSL in pieces so I may have nothing useful yet on that. I usually leave out the RX coil and just inject a signal for the things I was interested in.

As soon as the questions involve physics outside of electronics circuitry, Spice has dubious value in my opinion.

I haven't even seen a good simulation of the null signal, for example, that represents what we see when nulling our coils (Monk seems to be trying to make one). We can probably model one, but I wouldn't trust its predictive value. Ultimately we could keep tweaking to make it match reality, but when do we trust we have modeled the correct physics and trust it predict useful things? We're not really modeling magnetic fields and such with Spice.

But it's worth a try just to see what happens. I'll try to whip up a front end with an oscillator and TX coil and the RX coil. But how do I model the null? With a small mutual inductance? Well, I could start there and just see what happens. Then look at the output of LF353 pin 8 with the RX coil grounded vs. ungrounded.

Beats working... but feels like it sometimes...

-SB

[simonbaker]

the problem is that i do not have a scope..only a good multimeter

when i turn on my md there's a signal only on"one chanel" that means it works good when it's switched to all metal position..when i switch to other position there's no response..firstly i thought that maybe a half of 393 was broken but i don't think that this may happen..so i was looking on schematic and maybe i'm wrong but i thing that one of j107 is responsible for all metal function and secound is responsible for disc function..so there may be my problem..but like i said i can be wrong..i have replace all ic's so its very posible that my transistor can be broken

another question about ballancing coils: to balance a coil i should check the voltage on rx coil on beetwen pin 6 and 5 right

maybe my questions are stupid but i'm trying to lern how much i can to know this project as good as posible

anyway i'm greatfull for any help

when i finish and repair my md i will post mesurements to help other people

thanks a lot and regards to all

Hi lunamay, glad you're still working for Ivconic.

The ALL METAL mode uses same JFet as Disc, so I don't think your problem is there.

Some questions to help us:

1. How well does you TGSL work in ALL METAL mode? What air depth?

2. I would check your .015u capacitor C6 (the one in parallel with the RX coil). Have you been playing with that? Have you changed to a different value? Have the solder joints broken?

3. Unhook the RX coil and measure the capacitance of capacitor C6 and let us know.

4. Measure the RX coil inductance and let us know.

5. Measure the oscillator frequency and let us know.

6. Do your coils have shields? Check the ground wire from the cable to make sure it really contacts the shields. Use an ohmmeter.

7. If your coils have shields, do they have correct gaps so they are not continuous loops? Are the two coil shields insulated from each other so they do not form a continuous loop?

8. To balance/null the coils, I refer to Qiaozhi's advice. I would measure null voltage at pin 8 (relative to ground) of LF353 (the output) because easier. Pins 5 and 6 I'm not sure, probably OK, but much smaller voltage expected.

Since you don't have scope, this is how I would proceed for Qiaozhi's method:

A) Adjust your coils overlap until AC voltage at LF353 pin 8 (or pins 5, 6) is approximately minimum.

B) Set the GB pot in mid position.

C) Put your DC voltmeter probes across capacitor C12 (connected to source of JFet TR5, GB channel), use millivolt scale, put ground probe to ground, and make final small adjustment to your coil overlap so the DC voltage (relative to ground) on C12 is zero millivolts or as close as you can get it. If not exact, I would choose slightly positive.

D) I don't know if this step will help or not. You can take a piece of ferrite and wave it around 7 cm in front of coils, and adjust the GB pot to remove any clear beep you hear. Some residual beep may be OK.

If you change the GB pot much, go back to step C and repeat.

Let us know how it goes!

Regards,

-SB

[Monk]

I think it's interesting and would like to explore it, but in a different thread so we don't get confused. Would you start a thread and repost your message there?

Regards,

-SB

Done....

[dfbowers]

My simulations so far aren't sophisticated enough to include EMI noise or static charge, but that would be an interesting thing to study. The question is, can we make a realistic model in spice for that? We could try.

How about a sudden voltage pulse on the RX coil (but which side? both?)? That could be simulated.

I could believe that if static charge hits both sides of the RX coil simultaneously, but one side is grounded, a fat differential hits the op amp. On the other hand, if you don't ground, could we blow out our chip with a huge common mode pulse? Depends on current? Maybe protection diodes would fix that? (hey, that's a thought -- is your ungrounded RX coil endangering your PCB?)




SB

Simon,

I was primarily interested in how the Spice simulation depicts the basic operation of the TGSL with an ungrounded Rx circuit in contrast to a grounded one.

Don

[simonbaker]

The following is the approach that I've taken with regard to simulating the coupling between Tx-Rx coils.

Much of this (text) is from Wikipedia.

Starting with inductance:

The quantitative definition of the (self-) inductance of a wire loop in SI units (webers per ampere, known as henries) is

L= N\Phi*i

where L is the inductance, Φ denotes the magnetic flux through the area spanned by the loop, N is the number of wire turns, and i is the current in amperes. The flux linkage thus is

N\Phi = Li

N d(Phi)/dt=v.

A 2 coil search head can be modeled as a transformer with the following equations which describe transformer action:

V1 = L1*di1/dt - M*di2/dt

V2 = L2*di2/dt - M*di1/dt

The mutual inductance also has a relationship with the coupling coefficient. The coupling coefficient is always between 1 and 0, and is a convenient way to specify the relationship between a certain orientation of inductor with arbitrary inductance:

M = k sqrt{L1 L2}

where

k is the coupling coefficient and 0 ≤ k ≤ 1,
L1 is the inductance of the first coil, and
L2 is the inductance of the second coil.

Once the mutual inductance, M, is determined from this factor, it can be used to predict the behavior of a circuit.

The magnetic flux linked to both the primary winding and the secondary winding is said to be the main flux, (φ12 or φ21). The magnetic flux which interlinks only with the primary winding, and does not interlink with the secondary winding, is said to be the primary leakage flux, φσ1. The magnetic flux which interlinks with the secondary winding, and does not interlink with the primary winding is said to be the secondary leakage flux, φσ2. The primary side leakage flux becomes the primary side leakage inductance, and the secondary side leakage flux becomes the secondary side leakage inductance. Defining k to be the coupling coefficient, and denoting the leakage inductances of the primary side and the secondary side as Le1 and Le2 respectively, it follows that:

Le1 = (1-k)L1,

Le2 = (1-k)L2

In the case of a IB search head such as the TGSL, we want essentially zero mutual inductance, N*d(Phi)/dt=0, i.e. k=0.

Which then leads to:

Le1 = (1-0)L1 = L1

Le2 = (1-0)L2 = L2


So instead of a search head being modeled schematically as:



We have:





To simulate the presence of ground, metal or other objects I use the following schematic/configuration.






L3 is driven such that it is synchronized to Tx but contains driving circuitry for wave shaping (phase and amplitude) to simulate effects of ground and or metal.

K adjsts the coupling between the simulated response to Tx and the Rx coil.

I have used this configuration in recent simulations and it appears to work quite well. Of course other lumped parameters such as distributed capacitance, resistance, etc. can be added to the model.

Hope this is of help if not for anything else to stimulate conversation.

I think it's interesting and would like to explore it, but in a different thread so we don't get confused. Would you start a thread and repost your message there?

Regards,

-SB

[Monk]

Moved to http://www.geotech1.com/forums/showthread.php?t=16866

[simonbaker]

Hey Simon,

Disregard my question from earlier.. it was a total brain fart and I think I'm just getting old. (I even passed for a senior citizen at a place yesterday)..

Ha! Yes, I find a little scrooge-like frown can really net some nice theater ticket discounts...

I just need to double check some of my observations from last week when I get home, on what test point it was I was monitoring when nulling my coils.. it may have been U103 pin 7 now that I think about it.. All DC levels get filtered out downstream. Probably the same as monitoring at C12.

As far as Rx zero crossing goes, I have not found that it matters but Qiaozhi's argument makes sense, particularly from a quality control perspective. I may just pay particular attention to that when making subsequent coils just to help eliminate variables between coils and allowing for a full range of adjustment.

Yes, I'm remembering there are some other bonuses you get by choosing a null phase to cancel the RX signal in the GB channel. One is that it makes that channel insensitive to oscillator amplitude modulation.

However, we should remember that just because we "ground-balanced" the RX signal in the GB channel, it still may have a rip-roaring effect in the DISC channel. So I think the "ground effect" can still "make waves" by interfering with the DISC channel, even to the extent of masking targets in some cases.

I also am thinking more about whether there really are two ground phenomena we have to watch out for: a) Qiaozhi's ground effect, which amplitude modulates the RX signal, and b) "hot rocks" which create a signal with a distinct phase that can also be discriminated out in the GB channel. My ferrite targets might be in that last category.

I think it might be worth starting a new thread to kick around ground balance subjects.

I will let you know about balancing with and without the null signal zero crossing under the GB sync pulse center next week when I have my scope to set it up.

I have not gotten to Spice simulations yet, but that's something I want to play with soon. I was wondering.. how would your simulations behave NOT grounding the Rx coil at all? Based on all my field test, the ONLY coil I have made that is suitable to be used in wet grass is the last one - Shields only connect to ground on the PCB and nowhere else.. Both leads from the Rx coil float free. Performance is not noticeably different is actual use and the benefits of a noise free coil on wet ground seems to be the way to go..

My simulations so far aren't sophisticated enough to include EMI noise or static charge, but that would be an interesting thing to study. The question is, can we make a realistic model in spice for that? We could try.

How about a sudden voltage pulse on the RX coil (but which side? both?)? That could be simulated.

I could believe that if static charge hits both sides of the RX coil simultaneously, but one side is grounded, a fat differential hits the op amp. On the other hand, if you don't ground, could we blow out our chip with a huge common mode pulse? Depends on current? Maybe protection diodes would fix that? (hey, that's a thought -- is your ungrounded RX coil endangering your PCB?)

BTW.. I found enough change on the beach in a few hours to buy my next set of batteries. I even found a euro! Finding coins in loose sand has it challenges as coins are more than likely on edge.. No gold yet, but I have a few days to go..

Don

That's better than I do in Las Vegas.

I had miserable luck the one time I took my Compadre to the beach. Nothing turned up but bottle caps. However, they groom the beach with a dragger, maybe part of the problem. But I was surprised. I'll try another beach with your TGSL and I expect some serious treasure.

Coins on edge... brings up a thought... a need for coils that lay down a more horizontal field in the ground (figure-eight coils?). I'm sure already looked into, but maybe that's the way to hunt the beach - once with regular coil, then again with horizontal field coil. It never ends!... fortunately!

Regards,

SB

[pebe]

Are you sure it's not tuned? Have you checked inside the search coil to see if there's a tuning capacitor there?

No, I cannot get into the search coil without wrecking it and I don't want to do that (that's the reason for modifying the fromt end).

I assumed it was not tuned because the signal induced into the rx coil is exactly 90º from the Tx coil. But I'll add capacity to see the effect.

[Qiaozhi]

the problem is that i do not have a scope..only a good multimeter

when i turn on my md there's a signal only on"one chanel" that means it works good when it's switched to all metal position..when i switch to other position there's no response..firstly i thought that maybe a half of 393 was broken but i don't think that this may happen..so i was looking on schematic and maybe i'm wrong but i thing that one of j107 is responsible for all metal function and secound is responsible for disc function..so there may be my problem..but like i said i can be wrong..i have replace all ic's so its very posible that my transistor can be broken

another question about ballancing coils: to balance a coil i should check the voltage on rx coil on beetwen pin 6 and 5 right

maybe my questions are stupid but i'm trying to lern how much i can to know this project as good as posible

anyway i'm greatfull for any help

when i finish and repair my md i will post mesurements to help other people

thanks a lot and regards to all

I'm afraid you're going to find it very difficult to debug without a scope. As you said, you've already replaced all the ICs, and you're not using a factory coil. It could be as simple as a short on the PCB or as complicated as ... well, who knows what? Since the TGSL is a motion detector, you really need to be able to see the waveforms to understand what's going on. If you can get access to a scope, perhaps we will have half a chance of helping you to fix this remotely.