TGSL Experiments

[simonbaker]

I wanted to point on that subject earlier...
You noticed, in one of previous posts, an "notch" that occurs at signal when scoped. So i noticed too on few handmades. Later discovered C9 as very critical there. Must be hi quality p.p. capacitor or some very,very good ceramic. Usually i test dozen of those to find most suitable one. This turned to be very benefitial before nulling new coil on that machine (of course there is always some previously finished coil to test machine and check capacitors for C9).I do beleive that was discussed in the past. Must look at first 50-70 pages of TGS thread. Although setup at those versions is a bit different, yet i do beleive C9 is main cause of problem there.

Thanks for reminder about that C9 - now I'll pay attention to it more and check.

Regards,

-SB

[simonbaker]

I added resistor R59 (3.3k) to 180 deg disc circuit as precaution so TX coil is never connected directly to LF353 output through C9 - mainly to not drain current from TX oscillator.

May be totally unnecessary... other ways to do it also (put between LF353 output and disc pot R18?).

Note: All resistors in 180 disc circuit are subject to mods for improvement.

-SB

I tried the 180 deg disc circuit in my prototype TGSL without the extra 3.3k resistor (R59). It seems to work as expected, disc signal shifts smoothly. I didn't see any effect on TX voltage when pot all way to LF353 output. So maybe R59 not needed.

If we want smaller shift range, we can add resistors around R18 disc pot.

-SB

[simonbaker]

180 deg disc practical resistors

180 deg disc practical resistor values, tested OK.

27k, 1.5k instead of 30k, 1.4k.

-SB

[simonbaker]

Need phase offset to 180 disc circuit

I think the above 180 deg disc circuit needs an additional phase offset (90 deg?) to make it work like Tesoro models as pot turned -- All Metal first, then iron, then nickel, gold, copper, silver, etc.

To be continued.

-SB

[simonbaker]

TGSL Audio again

I am observing something about audio with my original TGSL prototype circuit. Maybe related to old problems we thought were audio jack?

My circuit (Gift Pack version) has:

1) original oscillator (with JFet)
2) original audio (Darlington)
3) threshold pot on final LM358
4) earphone or tiny speaker - jack or no jack.

I am testing indoors where there is some RX coil noise.

- When I turn threshold pot, I get some chatters as expected. But here is the observation: each time the speaker chatters, it seems to cause more noise in the detector circuit, like a little pulse. In other words, the speaker itself is part of the problem. If I disconnect the speaker, the noise has a steady range. If I reconnect the speaker, each time the speaker chatters, a noise pulse appears in the circuit.

I can see the noise pulse as far back as the output of U103 (first LM358 ), but mainly in disc channel it seems.

I also see a little of the noise pulse in the -5V supply, but not in ground or +8V rails. I'll note that my -5V rails are "star" configuration, several wires because just prototype.

I can't determine what causes the effect -- is the speaker sending EMI to the coil??? I tried moving speaker close and far, but no clear relationship.

Is it the -5V rail??? Is the rail picking up EMI from the speaker, or is the pulse from the output of U107a going back through resistors to rail? It will take some time to test that theory, I'll need to create a clean -5V rail.

Anyway, the main observations is that the chatters help cause themselves it seems! I notice that a stronger chatter makes a stronger noise pulse in the front end (which increases the chatter even more).

I am thinking if the "feedback" can be eliminated, I think the chatters will be milder and make it easier to set threshold lower, more sensitive.

So I will look further into why the speaker seems to make noise in the detector front end. Whether it can be eliminated easily I don't know. Also, it may not be a problem at all, as long as threshold set above basic noise level -- in other words, maybe it's OK to have feedback if real target makes signal. Anyway, an area I'd like to understand better.

-SB

[simonbaker]

(Correction to previous message: circuit for audio tests is not exactly the "Gift Pack" circuit. I do not have 555 timer in circuit, just Darlington as with original TGS.)

I did some more tests trying to understand how audio beep causes pulse in detected signal front stage.

1. I replaced -5V supply with separate battery pack to see if maybe supply impedance coupled the signal - no change.

2. I replaced speaker with LED and resistor to see if speaker coil is problem. When resistor high (150 ohms), the pulse goes away. If I make resistor low enough (about 10 ohms), I see pulse in the detected signal again.

So I think maybe it is just the big current draw of the audio that makes a little pulse in the first LM358 stage of the detector.

3. I tried adding a 1000 uF capacitor to the 220 uF cap across the Darlington to isolate the current draw - surprisingly, it did not make much difference to chatter - but maybe a small reduction of pulse.

Next maybe I'll try audio circuit from TGSFinal with BD140 etc.

In the end probably this is not important issue because I realize: most important thing is making background noise as small as possible to keep it below threshold, then it doesn't matter if speaker causes feedback or not; in fact, feedback helps audio for real targets.

-SB

[WM6]

Attached image is idea for 180 deg discrimination control for Tesoro Golden Sabre.

However, I have not found easy way to mod TGSL pcb. Disc switch was a very nice jumper!

.

Good job, thanks.

Maybe separate pcb can be solution over 4 wire connected to main pcb.

[simonbaker]

Good job, thanks.

Maybe separate pcb can be solution over 4 wire connected to main pcb.

Don't build yet!

I believe the circuit needs an additional +90 deg phase shift for compatibility with TGSL. I am studying a few different ways to do it, trying to keep minimum components and least noisy design. So circuit not quite ready yet! (It actually might work, but strangely).

As for layout - separate pcb of course will work, but it looks so far that most of the changes fit nicely on the PCB except for one long connection.

I am working on a PCB layout that makes the long connection with several jumpers. Maybe someone can find a better layout.

But wait for final circuit first! I need to test more.

-SB

(BTW: I have another idea that might save components -- instead of adding 90 deg to the above circuit, maybe we could subtract 90 deg from RX signal. This might be possible by changing RX coil resonant frequency to other side of TX frequency -- just a wild idea I'll check into.)

[simonbaker]

Tesoro MD shaft

For reference, building your own -- dimensions of Tesoro MD shaft. It is adjustable, these measurements set for my size (over 6').

-SB

[simonbaker]

TGS Tuning : Rx coil resistance

My general goal is to learn as many tuning tricks for TGS designs to make it easier to build a really good TGS with less trial & error and luck needed.

One question is Rx Coil resistance (Tx coil is a separate question). We can choose resistance by wire thickness/gauge. But what is best and why?

My first thought was to use as low resistance as possible to increase Q of resonant circuit with .015 uF capacitor. I figured this would increase gain. I also thought maybe better signal-to-noise ratio because narrower filter.

But now I see maybe this is wrong for basic TGS design. I forgot that Rx resonance is typically set higher than TX oscillator frequency, so target signal is quite a bit lower than resonant freq. Purpose of that I think is it makes received signal phase much more stable, because errors in Rx coil inductance or capacitor will not affect phase much. Of course you lose a lot of gain at the same time.

I thought lowering resistance of Rx coil would boost gain. But look at attached graph for a typical coil. If you lower resistance from 20 ohms to 7 ohms (by using 24 gauge wire), you do not increase gain much at the TX oscillator freq (maybe .1 dB). However, the gain at the resonant frequency is boosted a lot (9 dB). The problem is, maybe you are now increasing noise around the resonant frequency without similar gain of target signal; not what you want.

Another related problem could occur if you not only lower Rx resistance but also increase capacitor to shift Rx resonant frequency closer to Tx frequency in hopes of increasing signal gain. The problem is you may get too much front end gain which makes the background noise exceed the max sensitivity threshold and gives constant chatters. This may be easily fixed I think though by lowering gain of LF358 stage (reducing feedback resistor?).


So lowering Rx resistance does not seem like good idea after all! However, this conclusion depends on where noise comes from. If noise from Rx coil is dominant, this conclusion seems valid. However, if dominant noise is from circuit components, maybe Rx resistance not so important. My own observations are that significant noise comes from Rx coil, therefore, pay attention to Rx resistance, don't make it too low (based on this analysis)!

An additional question is: is there an optimal Rx resistance to strive for? It's easy to add resistance.

Final thought: I'm interested in exploring a different MD design where the RX resonant frequency is same as TX oscillator freq. This would be difficult to stabilize, but would get maximum target signal gain, and raising Q of RX resonant circuit should reduce off-resonant noise. I hope to experiment with that idea.

-SB

[simonbaker]

Simonbaker,
Can you make some conclusions?
No matter what I try so far this last TGSL just refuse to work like it should.
Now after so many coils I am shore that coil is problem that can't solve so far.
I use 0,25 bare wire.0,27 with varnish.
Old DD27 that were made for previous TGSL device.:
C12 .096
C15 .161
TX 6,21mH 27,9
RX 6,77mH 29,1
13kHz
nulled to .050
OO coil
C12 -.073
C15 -.325 Why both -
13,7kHz
TX 6.00 26,6
RX 6,58 27,6
nulled to .200
Both coils shielded.And sealed with epoxy.But I still have this poor performance.Low sens.1 euro coin at 22-23cm.
RX cap is 15N.Now I put 100k pot for.GEB.
My last /still not shielded and sealed coil/isFPRIVATE "TYPE=PICT;ALT="D27
TX 5,93mH 26,7
RX 6,58mH 28,0
13,9 kHz
C12 .209
C15 -1.185
I now that I miss something and just looking for solution.Because I like this project.

I have the same problem -- easy to make detector that works around 20 cm, but I have not achieved 30+.

I am trying to study each part to understand. The experienced people say "coil", so probably that is the answer -- but what makes one coil wrong, another right? I want to study and see. So I'm making various coils. It will take some time.

One difficulty I have is electrical noise inside house. Final test must be outside away from power wires, with sensitivity pot turned most sensitive.

Question: when you use outside and turn sensitivity to most sensitive, do you hear "chatter", or is it quiet?

C12 -.073
C15 -.325 Why both -

Are those voltages on the capacitors? They should not be negative. If they are negative, then you have "nulled" the coil incorrectly, you need to shift the coil slightly to other side of the null to get positive voltages.

Also, move the discrimination knob through full range, the voltages should stay positive over the whole range (at least most of it).

I have ideas I want to test.

1. Shielding -- the shielding can reduce depth. Maybe try less conductive shielding -- instead of foil, carbon paint, screen, etc. (Of course make sure shielding has gap!). Maybe try not shielding the TX coil at all.

2. RX coil resistance -- if too low, it may pick up extra noise at resonant frequency. If too high, reduces gain. Your value might be a little high, try slightly heavier wire to get around 20 ohms???? Who knows, it is something to experiment with.

3. TX frequency -- if too far from RX resonant frequency, you lose gain. You can either raise the TX frequency (by lowering inductance of TX coil, or decreasing capacitor C2 .022 uF) or you can also lower the RX resonant frequency by increasing capacitor C6 (.015 uF) or increasing inductance of RX coil. This increases gain, but watch out, if you go too far you alter phase of target signals and discrimination won't work well.

4. You can try "tuning the null" by swinging a coin above coils and adjusting the null for maximum depth. I don't know why this works exactly; theory says if you are close to null it shouldn't make much difference to motion detector. I would make sure voltages on C12, C15 are positive.

5. Gain -- if you have no "chatter" or noise sounds when sensitivity pot is maximum, then maybe you can increase gain of the RX signal. Techniques include lower resistance of RX coil; lower resistance of TX coil; move RX resonance closer to TX oscillator frequency; maybe play with gain of LF353 by increasing R15 220K, but could cause other problems;

6. Reduce sensitivity threshold -- if you have no "chatter" or noise sounds when sensitivity pot is maximum, then maybe you can decrease sensitivity threshold. This is not really recommended, but it might reveal something. This is done by shorting resister R37 (1k) connected to sensitivity pot.

7. Different diameter coil -- I am making 26-27 cm coils. I think big enough, but who knows. Maybe smaller would work better with some coin???

8. Check discrimination driver phase (LM393 output) -- wrong phase could reduce gain.

I think it is important to distinguish between two problems 1) noise, 2) gain. They require different solutions. If too noisy, must identify where noise coming from; coil or circuit? Jitter in oscillator? etc. If gain, then look at ways to increase gain as above.

I'll keep trying. Please keep us updated on your progress.

Regards,

-SB

[Aiko]

Simonbaker, the frequency was about 16 kHz and after the adjustment of 22 n capacitor (now is 23 n) the frequency is 14.8 kHz.

[simonbaker]

Simonbaker, the frequency was about 16 kHz and after the adjustment of 22 n capacitor (now is 23 n) the frequency is 14.8 kHz.

Well that's interesting. You moved in direction of reduced gain but it works deeper! My guess is that 16 kHz is too close to RX coil resonant frequency and causes phase shift in target signal, which makes discriminator and/or ground balance off center and reduces gain.... maybe!

In any case, you made your detector closer to original specs, and I guess that always helps!

Thanks, it gives me more ideas for experiments and things to try.

Regards,

-SB

[Aiko]

As well I tried to short this 1 k resistor between sens pot and the ground. In my case this not brings any advantage. I would like to ask you, did you try to add a threshold?

Regards

[simonbaker]

As well I tried to short this 1 k resistor between sens pot and the ground. In my case this not brings any advantage. I would like to ask you, did you try to add a threshold?

Regards

I'm not sure what you mean by "add a threshold"?

Some versions of the TGSL have a pot labeled "threshold" on the final LM358 (U107 whose output goes to the audio circuit). Do you mean that?

My "breadboard" version of TGSL has this threshold pot. It is not really what you expect as a threshold. It really acts as a low pass filter that stops quick pulses coming from the comparators U106.

Changing the "threshold" changes how wide a pulse it stops, so it works like a low-pass filter on the "chatter" noise coming from the comparator -- very short pulses are stopped, but long (target) pulses come through.

But I don't think the pot is too useful, because the fixed resistors in the other schematics are a good choice (and what is on the original Tesoro TGS). If you have a very low noise condition, reducing the "threshold" might give you a little better audio (longer beeps). But no improvement in depth.

Since I like to play with controls, I might add it to my next built TGSL.

The sensitivity control (R36 on some schematics) is a true threshold for the target signal. It is not a gain control.

The 1K resistor causes a minimum threshold of about 30 mV (most sensitive), and turning the pot raises it above that. Shorting the 1K makes the minimum threshold go to zero mV, so the smallest noise (in both channels) can make a sound. You can still use the pot to increase the threshold above the noise.

Shorting the 1K resistor is only useful if you have a very, very low noise condition and want maximum sensitivity. Ideally, that could increase depth a little, but realistically there is probably too much noise, and you are limited by the gain of the comparators also.

Other more experienced people say the best way to increase depth is getting the coil adjusted correctly and that is something I'm studying. But it is still interesting to see if the electronics can be tuned with various ideas.

Regards,

-SB