TGSL Tuning & troubleshooting

[lunamay]

very nice work dfbowers..i'm trying to make enclosier for my coil and i hope it will be done soon

best regards

[simonbaker]

For your entertainment.. I have posted a new video of some tests from the basement on my latest coil (#7) . This one is coming out a little better than average.. but keep in mind, my basement is quiet!!

This one has the coil shields grounded only and is set up to minimize wet grass per some of my earlier posts. Also, I used conductive mylar for shielding material and paid attention to the null phase.

Enjoy!

Don

http://www.youtube.com/watch?v=Nx4Oxac0Ufg

I think that could be a weekly show -- Don's Basement! (I have a thing about basements and garages -- want to do a documentary some day).

You sure know how to build 'em, I've got to say. I'm still struggling with my latest PCB to try to get it to work well with your coil. Those darn trips to the mountains hold up my experiments. I've tried to find a nearer locale, but no luck yet. This urban sprawl is criss-crossed with power lines, cell towers, underground cables, you name it.

36 cm is phenomenal, especially since you ground-balanced and discriminated out junk. I've got those controls wide open for now just trying to get 30 cm. My DISC and GB pots can really limit the distance it seems.

Whatever you're doing, keep going and keep showing us!

If you ever get bored (like when you hit 50 cm), I'd be interested if you could try a coil with the TX coil made from 24 gauge wire (yeah, it's big and heavy). Then see if you can penetrate the clay any deeper.

I have tried such coils (I've got a 7 lb spool of 24 gauge wire... need some?), but did not have the knack of getting a stable signal at the time - I think the gain goes up, and maybe my noise environment confounded the experiment.

In theory, the lower resistance wire should raise the Q of the TX coil and increase the current. However, I did a graph -- http://www.geotech1.com/forums/showp...37&postcount=4
that showed the current did not increase nearly as much as I thought it would -- and I don't know why. But one idea would be to center-tap the TX coil and apply the driving voltage there somehow in the oscillator. Another idea I'd like to try eventually is to drive the TX coil from a multi-vibrator circuit instead of making it part of the oscillator. In any case, you can almost double the current it looks like. I'd be interested if you get more depth with it.

Enough rambling. Cool basement.

-SB

[dfbowers]

I think that could be a weekly show -- Don's Basement! (I have a thing about basements and garages -- want to do a documentary some day).

You sure know how to build 'em, I've got to say. I'm still struggling with my latest PCB to try to get it to work well with your coil. Those darn trips to the mountains hold up my experiments. I've tried to find a nearer locale, but no luck yet. This urban sprawl is criss-crossed with power lines, cell towers, underground cables, you name it.

36 cm is phenomenal, especially since you ground-balanced and discriminated out junk. I've got those controls wide open for now just trying to get 30 cm. My DISC and GB pots can really limit the distance it seems.

Whatever you're doing, keep going and keep showing us!

If you ever get bored (like when you hit 50 cm), I'd be interested if you could try a coil with the TX coil made from 24 gauge wire (yeah, it's big and heavy). Then see if you can penetrate the clay any deeper.

I have tried such coils (I've got a 7 lb spool of 24 gauge wire... need some?), but did not have the knack of getting a stable signal at the time - I think the gain goes up, and maybe my noise environment confounded the experiment.

In theory, the lower resistance wire should raise the Q of the TX coil and increase the current. However, I did a graph -- http://www.geotech1.com/forums/showp...37&postcount=4
that showed the current did not increase nearly as much as I thought it would -- and I don't know why. But one idea would be to center-tap the TX coil and apply the driving voltage there somehow in the oscillator. Another idea I'd like to try eventually is to drive the TX coil from a multi-vibrator circuit instead of making it part of the oscillator. In any case, you can almost double the current it looks like. I'd be interested if you get more depth with it.

Enough rambling. Cool basement.

-SB

I guess we need to consider the reactance of the coil as well.. at 14.5KHz and 6mH? In which case, it would be 543 ohms. (XL =2PiFL) So, a few ohms resistance might not make a huge difference since it's not DC? I would have to dig out some of my old radio engineering books to figure out how much the actual power difference would be vs. wire size.. If I could figure it out at all!!

[simonbaker]

I guess we need to consider the reactance of the coil as well.. at 14.5KHz and 6mH? In which case, it would be 543 ohms. (XL =2PiFL) So, a few ohms resistance might not make a huge difference since it's not DC? I would have to dig out some of my old radio engineering books to figure out how much the actual power difference would be vs. wire size.. If I could figure it out at all!!

The Q goes up pretty well going from 20 something ohms to 7 ohms, I thought it would be more than double the current (a more underdamped pendulum), but I think because driven by voltage the way it is it can't get as high as it might. But twice the current is still an attractive idea, if you don't run into gain/noise problems.

-SB

[simonbaker]

motorboating experiment

My PCB "motorboats" when you turn the sensitivity up high enough (I have the dfbowers mod installed for more sensitivity range).

I did some tests. To make the sensitivity range higher, I jumpered the 220K resistor R39 with a 47K resistor.

As I guessed, the motorboating occurs in a narrow sensitivity range. Below the range, it is normal chatter or quiet. Above the range, it is constant tone.

In the critical range, there is some kind of feedback. Now what is it? I don't know, but did more tests.

I shorted the RX coil where it connects to the PCB. This gives the best motorboating effect; I guess because the noise drops very low. With the RX coil, the motorboating is more broken up and not as steady.

So I'm pretty convinced that the motorboating is not magnetic coupling between the headphones and the search coil.

I suspect more it is in the power rails because the speaker puts a heavy load on the power, and I have even seen the pulse in the outputs of the LM358s and the rails themselves.

I tried a jumper from the speaker ground directly to the battery ground and lo and behold, it made a difference. The range of motorboating became very narrow and really a higher frequency screech that was not easy to stabilize.

I was a little surprised because I had tried that experiment before with no results I thought. But this time it seemed to make a clear difference. I think previously that I had the RX coil active, and perhaps there was a noise signal that was partially responsible; or maybe that case did involve some magnetic feedback to the coils.

Anyway, I am now feeling that it is desirable to use a PCB layout where the speaker ground has a very close connection to the battery ground, so you don't send all that current down a long and winding rail.

In fact I have a new PCB where I have done just that, and I'm wiring it up right now (takes days with my boneheaded method).

But I realize I didn't go far enough. Why not have the -5V supply ground similarly close to the battery ground? That supply makes notoriously wicked spikes on the rails (although there is no evidence they cause harm -- they just look alarming if you have a good fast scope).

So I'm working on a newer PCB layout trying to get the speaker, the voltage regulator, and the -5V supply circuitry close together so the grounds go directly in a more "star" configuration.

I think Eclipse (haven't heard a peep from Eclipse lately) might have had a layout somewhat like that. Yes, I just checked. It looks like a very good layout, worth looking at again. He did some interesting experiments with a big "ground plane" around all his wires that didn't work out, but interesting idea anyway. I think it's a very worthy layout though. It is I'm sure based on one of Ivconic's layouts, so he maybe has very similar one too.

-SB

(P.S. top and bottom copper layers merged in photo, jumpers not clearly distinguished)

[dfbowers]

The Q goes up pretty well going from 20 something ohms to 7 ohms, I thought it would be more than double the current (a more underdamped pendulum), but I think because driven by voltage the way it is it can't get as high as it might. But twice the current is still an attractive idea, if you don't run into gain/noise problems.

-SB

Simon.

Here is what I was thinking when I posted earlier today in regards to wire size. Please correct me if I'm wrong as I'm a little rusty. I have played with a ton of crystal sets and Tesla coils over the years, So I have a fair idea about Q. Not so much about metal detector coils..

I think that there are 3 or more parts to the equation when changing wire size and power consumed.

Pdcr
The wire loss caused by dc resistance is easy to calculate.
Pdcr (W) = I rms^ 2 × DCR
I rms = The rms value of the peak current applied to the inductor
DCR= The dc resistance of the inductor.
By increasing wire size, we can get more current through the coil..

Pacr
The wire loss caused by ac resistance.
Pacr (W) = I rms^ 2 × ACR
I rms = The rms value of the peak-peak ripple current
applied to the inductor.
ACR= The ac resistance of the inductor

This will still be 543 ohms. (XL =2PiFL) at 14.5 kHz and 6.0mH regardless of wire size.
Coincidently XC is 543 at 14.5 kHz and .020uf (for C1,C2) for resonance and cannot change.

Pcore
Pcore (mW) = K1f x B y × Ve
This should remain nearly constant.
(Core material, Freqency, flux density and core volume)

And:
PlossInductor = Pcore + Pdcr + Pacr
Our greatest limitation is still XL since this has to be contant, determined by L and F.

So. even if we change our DCR by 50 ohms, our ACR is still 10 times as much. With PI detectors XL is less of a concern (I think) because of the lower freqency, lower inductance and the duty cycle of the pulse allows large currents to flow and DCR becomes the limiting factor for power in the coil.

Now relating Q to L and R
Q = W L/R

This will affect the bandwidth of our coil (-3db rolloff) but XL is still the same.

Let me know your thoughts..
Don

[dfbowers]

My PCB "motorboats" when you turn the sensitivity up high enough (I have the dfbowers mod installed for more sensitivity range).

I did some tests. To make the sensitivity range higher, I jumpered the 220K resistor R39 with a 47K resistor.

As I guessed, the motorboating occurs in a narrow sensitivity range. Below the range, it is normal chatter or quiet. Above the range, it is constant tone.

In the critical range, there is some kind of feedback. Now what is it? I don't know, but did more tests.

I shorted the RX coil where it connects to the PCB. This gives the best motorboating effect; I guess because the noise drops very low. With the RX coil, the motorboating is more broken up and not as steady.

So I'm pretty convinced that the motorboating is not magnetic coupling between the headphones and the search coil.

I suspect more it is in the power rails because the speaker puts a heavy load on the power, and I have even seen the pulse in the outputs of the LM358s and the rails themselves.

I tried a jumper from the speaker ground directly to the battery ground and lo and behold, it made a difference. The range of motorboating became very narrow and really a higher frequency screech that was not easy to stabilize.

I was a little surprised because I had tried that experiment before with no results I thought. But this time it seemed to make a clear difference. I think previously that I had the RX coil active, and perhaps there was a noise signal that was partially responsible; or maybe that case did involve some magnetic feedback to the coils.

Anyway, I am now feeling that it is desirable to use a PCB layout where the speaker ground has a very close connection to the battery ground, so you don't send all that current down a long and winding rail.

In fact I have a new PCB where I have done just that, and I'm wiring it up right now (takes days with my boneheaded method).

But I realize I didn't go far enough. Why not have the -5V supply ground similarly close to the battery ground? That supply makes notoriously wicked spikes on the rails (although there is no evidence they cause harm -- they just look alarming if you have a good fast scope).

So I'm working on a newer PCB layout trying to get the speaker, the voltage regulator, and the -5V supply circuitry close together so the grounds go directly in a more "star" configuration.

I think Eclipse (haven't heard a peep from Eclipse lately) might have had a layout somewhat like that. Yes, I just checked. It looks like a very good layout, worth looking at again. He did some interesting experiments with a big "ground plane" around all his wires that didn't work out, but interesting idea anyway. I think it's a very worthy layout though. It is I'm sure based on one of Ivconic's layouts, so he maybe has very similar one too.

-SB

(P.S. top and bottom copper layers merged in photo, jumpers not clearly distinguished)

Nice observation! This may be worthwhile looking into. I see a lot of sloppy things going on in the circuit that don't really affect the basic functionality of things or do they!?

As far as the motorboating symptom. I have only seen this twice with mine.. When I hook up a huge speaker and get it oriented wrong to the coil, or when I detect around the corner of the house where the service entrance is..

[simonbaker]

Simon.

Here is what I was thinking when I posted earlier today in regards to wire size. Please correct me if I'm wrong as I'm a little rusty. I have played with a ton of crystal sets and Tesla coils over the years, So I have a fair idea about Q. Not so much about metal detector coils..

I think that there are 3 or more parts to the equation when changing wire size and power consumed.

Pdcr
The wire loss caused by dc resistance is easy to calculate.
Pdcr (W) = I rms^ 2 × DCR
I rms = The rms value of the peak current applied to the inductor
DCR= The dc resistance of the inductor.
By increasing wire size, we can get more current through the coil..

Pacr
The wire loss caused by ac resistance.
Pacr (W) = I rms^ 2 × ACR
I rms = The rms value of the peak-peak ripple current
applied to the inductor.
ACR= The ac resistance of the inductor

This will still be 543 ohms. (XL =2PiFL) at 14.5 kHz and 6.0mH regardless of wire size.
Coincidently XC is 543 at 14.5 kHz and .020uf (for C1,C2) for resonance and cannot change.

Pcore
Pcore (mW) = K1f x B y × Ve
This should remain nearly constant.
(Core material, Freqency, flux density and core volume)

And:
PlossInductor = Pcore + Pdcr + Pacr
Our greatest limitation is still XL since this has to be contant, determined by L and F.

So. even if we change our DCR by 50 ohms, our ACR is still 10 times as much. With PI detectors XL is less of a concern (I think) because of the lower freqency, lower inductance and the duty cycle of the pulse allows large currents to flow and DCR becomes the limiting factor for power in the coil.

Now relating Q to L and R
Q = W L/R

This will affect the bandwidth of our coil (-3db rolloff) but XL is still the same.

Let me know your thoughts..
Don

I'm not sure I understand your Pacr term and the phrase "wire loss". I think maybe you are saying that reactance can greatly limit the "current" in a simple circuit. The key is: which current?

Basically I'm looking at it as a resonant "tank" (LCR circuit) which can have very unusual properties because there is more than one current to think about. There is the current coming from the voltage source, and the current "in the tank" circuit itself.

The current "in the tank" can be much greater than the current feeding it from the voltage source. If the Q is high enough, you can have huge currents sloshing back and forth (capacitor through coil) with only a tiny dribble from the battery keeping it going -- like gently tapping a big pendulum with low damping. And the total reactance can be really high at resonance, which keeps the feeding current tiny. I'm sure this is not news, just thinking out loud.

But I'm not handy enough with equations to extend that LCR circuit to an oscillator circuit, so I simulated it with LTSpice and varied the TX coil resistance and got the graph I showed in the link.

It does show an almost doubling of current going from about 30 ohms to 7 ohms for the TX coil -- and I had hoped for more. But I'll take doubling for now. I expect though that the gains will be minimal for the extra poundage of coil weight on your arm! However, I would like to test it.

People have said that raising the current doesn't help because it's like highbeams in fog -- you don't see more. I agree, if there is sufficient ground "roughness" to mask targets below a certain depth with ground noise. But if the "fog" is very smooth, you can theoretically boost the signal to overcome EMI noise (different than ground noise), then turn up the "contrast" (something our eyes don't do too well) to see the target. So I'd like more current to play with.

I know you had a detector that worked worse in "high voltage" mode, but I wonder if it wasn't applied optimally to the situation.

What I was also bringing up is that with a simple LCR "tank" circuit, as R goes to zero, I believe the current goes to infinity. So I'm interested in looking into a different kind of oscillator that can better take advantage of that effect. Of course certain power losses that you mentioned can't be reduced and that is the limit.

All these experiments are on my wishlist -- it's been great having you/others pitch in and start getting some useful answers.

Cheers,

-SB

P.S. On Pacr, AC power loss, there is a true radiative loss that produces damping of the resonant tank, true.

[simonbaker]

Nice observation! This may be worthwhile looking into. I see a lot of sloppy things going on in the circuit that don't really affect the basic functionality of things or do they!?

As far as the motorboating symptom. I have only seen this twice with mine.. When I hook up a huge speaker and get it oriented wrong to the coil, or when I detect around the corner of the house where the service entrance is..

Yes, I suspect several ways to cause motorboating. Maybe this last effect is peculiar to my dystopian wiring techniques -- thicker buss wires could help. I'll keep playing with the layouts just in case -- although it's something I am particularly slow at.

-SB

[simonbaker]

minor worries

Another "worry" I'd like to investigate and possibly put to rest is the effect of the pot/switch wires (disc, sensitivity, mode) on making noise.

Depending on how you mount your pots and switches, there can be some fairly long leads. And those leads could be nice little antennae (bugs anyone?).

You might think any signal they pick up would be dwarfed by the huge TX oscillator voltage or sync pulse they carry, but don't forget the MD is hyper-sensitive to the slightest phase shift -- that's what it's designed for!

So injecting a small signal into the pot or mode switch wires could potentially cause "phase jitter" in the sync pulse or zero crossing detectors.

I happen to believe most noise is due to the RX coil picking up stray EMI -- that is one hell of a big antenna. You can see the effect by jumpering the RX leads at the PCB -- immediately your noise at the LM308 outputs drops way down (at least in my noisy work area it does).

To test the effect of pot and switch leads, I guess I would jumper/disconnect the RX coil, use a computerized oscilloscope that could measure RMS noise, then solder the pots/switches with the shortest leads possible and re-measure the RMS noise.

Hopefully, the leads are not an important source of noise.

I notice that on the TGSL dfbowers sent me, he mounted the sensitivity and disc pots on the front of the box where the coil cable comes out. I think this allowed shorter pot wires, maybe to some advantage????

I can't think of a good PCB design where the coil leads connect to the front and the pot leads connect to the back, so perhaps we're stuck.

Just another little item to check out if anyone has the inclination and equipment. Not the easiest test to do.

SB

P.S. Another slight concern -- those same pot/switch wires carry some big signal voltages. Do they radiate "noise" to other parts of the circuit? Somehow I think not, but who knows?

[dfbowers]

I'm not sure I understand your Pacr term and the phrase "wire loss". I think maybe you are saying that reactance can greatly limit the "current" in a simple circuit. The key is: which current?

Basically I'm looking at it as a resonant "tank" (LCR circuit) which can have very unusual properties because there is more than one current to think about. There is the current coming from the voltage source, and the current "in the tank" circuit itself.

The current "in the tank" can be much greater than the current feeding it from the voltage source. If the Q is high enough, you can have huge currents sloshing back and forth (capacitor through coil) with only a tiny dribble from the battery keeping it going -- like gently tapping a big pendulum with low damping. And the total reactance can be really high at resonance, which keeps the feeding current tiny. I'm sure this is not news, just thinking out loud.

But I'm not handy enough with equations to extend that LCR circuit to an oscillator circuit, so I simulated it with LTSpice and varied the TX coil resistance and got the graph I showed in the link.

It does show an almost doubling of current going from about 30 ohms to 7 ohms for the TX coil -- and I had hoped for more. But I'll take doubling for now. I expect though that the gains will be minimal for the extra poundage of coil weight on your arm! However, I would like to test it.

People have said that raising the current doesn't help because it's like highbeams in fog -- you don't see more. I agree, if there is sufficient ground "roughness" to mask targets below a certain depth with ground noise. But if the "fog" is very smooth, you can theoretically boost the signal to overcome EMI noise (different than ground noise), then turn up the "contrast" (something our eyes don't do too well) to see the target. So I'd like more current to play with.

I know you had a detector that worked worse in "high voltage" mode, but I wonder if it wasn't applied optimally to the situation.

What I was also bringing up is that with a simple LCR "tank" circuit, as R goes to zero, I believe the current goes to infinity. So I'm interested in looking into a different kind of oscillator that can better take advantage of that effect. Of course certain power losses that you mentioned can't be reduced and that is the limit.

All these experiments are on my wishlist -- it's been great having you/others pitch in and start getting some useful answers.

Cheers,

-SB

P.S. On Pacr, AC power loss, there is a true radiative loss that produces damping of the resonant tank, true.

Ha.. true.. I guess we need to think about the AC resistance of the coil in that is does not oppose the flow of electrons through the wire (resistance) , but the inductor opposes the changes in current through them, in relation to time and releases that energy back into the circuit at a later time (reactance). So, with an ideal inductor, the power losses would be zero. But that doesn't happen because of RF losses and resistance of the wire.

It's the exchange of this energy between L and C in our tank circuit that is the "swinging pendulum" analogy is most useful, and the dampening factor as resistance and RF losses.

But I was reasoning that by decreasing your coil resistance to something very small, current could not go to infinity because it's not DC. We still have the coil opposing the changes in current. Also, collector current is still limited by R1.

Don

[simonbaker]

Ha.. true.. I guess we need to think about the AC resistance of the coil in that is does not oppose the flow of electrons through the wire (resistance) , but the inductor opposes the changes in current through them, in relation to time and releases that energy back into the circuit at a later time (reactance). So, with an ideal inductor, the power losses would be zero. But that doesn't happen because of RF losses and resistance of the wire.

It's the exchange of this energy between L and C in our tank circuit that is the "swinging pendulum" analogy is most useful, and the dampening factor as resistance and RF losses.

But I was reasoning that by decreasing your coil resistance to something very small, current could not go to infinity because it's not DC. We still have the coil opposing the changes in current. Also, collector current is still limited by R1.

Don

Yes, I think it is something about the oscillator, like R1, that is limiting it. It's not a simple tank circuit connected to an AC source.

But if it were a simple tank circuit with AC source, then the current can approach infinity I think, except for the non-ideal losses (radiation, capacitor losses, wires) as the R goes to zero.

It's like the pendulum -- each push adds energy, and as long as it doesn't dissipate, you can keep putting more and more in.

So of course I'm attracted to the idea of this huge magnetic field, but the TGSL oscillator doesn't seem the way to get it. But if another design can, I'm interested.

The downside is lugging that heavy coil. And it would require the necessary experiments to find out if it is even useful to have all the mag field. If the "ground noise" is the limiting factor hiding targets, then forget it, we're back to the fog-highbeam problem. Given the current art of MDs, I'm sure the answer is: not worth it, go with the lighter coil. Still, I'm itching to do the experiments. And I've got the wire...

-SB

[simonbaker]

On thinking, I realize I don't know a simple electronic equivalent of nudging a pendulum, so maybe there are practical limitations on my quest for a super underdamped coil... I'll keep thinking...

-SB

[dfbowers]

On thinking, I realize I don't know a simple electronic equivalent of nudging a pendulum, so maybe there are practical limitations on my quest for a super underdamped coil... I'll keep thinking...

-SB

You have my curiosity up on using a higher Q for the Tx coil.. I'm going to wind one tonight and test tomorrow using 26 gauge wire instead of 30 gauge. Stay tuned!

I have been reading a lot on the subject the past few days and it would seem like a common conclusion is that by increasing power, we gain more distance in bench testing and lightly mineralized soil, but we also increase noise and ground saturation as well.. and doubling the power gets us little return on the investment. Guess that explains my obervation with the Nautilus.

Instead of investing in more power, another thing to do is improve the S/N ratio. One way to do that is to run at full resonance and a high Q coil. I guess when winter rolls around and I'm cooped up, that's what I'm going to experiment with next.

[simonbaker]

You have my curiosity up on using a higher Q for the Tx coil.. I'm going to wind one tonight and test tomorrow using 26 gauge wire instead of 30 gauge. Stay tuned!

I have been reading a lot on the subject the past few days and it would seem like a common conclusion is that by increasing power, we gain more distance in bench testing and lightly mineralized soil, but we also increase noise and ground saturation as well.. and doubling the power gets us little return on the investment. Guess that explains my obervation with the Nautilus.

Instead of investing in more power, another thing to do is improve the S/N ratio. One way to do that is to run at full resonance and a high Q coil. I guess when winter rolls around and I'm cooped up, that's what I'm going to experiment with next.

Yes, running at full resonance is on my list. Whole different phase shift and tricky phase stability problems, but looking forward to trying to dope it out.

For now, I'm thinking that more power should improve the S/N in terms of EMI noise. Ground variation noise won't improve of course. Jitter S/N probably not helped either (if there is such a thing).

May have to turn down the gain at the LF353 to compensate.

Looking forward to what you find. I'll be doing same eventually, just a slow poke.

SB