Golden Sabre PLUS

[Sean_Goddard]

Hi All, Please have a look at these proposed mods. If you try them let me know how they work out please .

[ivconic]

"Picking" arround treshold and sense stage can give some better results.
I agree with you.
But i never tried anything arround TX. Will try at next handmade.
Other thing i would like to hear from you; how about rising a bit gain in all stages in chain? Starting from LF353 up to sense stage...
TGS offers very descent depths, i am pretty satisfied, but sometimes i have feeling it can act much "shortly". Sometimes seems like signal faint..How can i explain? It is detecting deep coins very accuratelly but audio is faint. I modded audio many times and gained much stronger response on target but i lost deepest coins than!? Now is easy to miss some deep or very small coin if sweeping coil to fast or a bit detached from soil surface...So i was thinking to rise gain a bit at every stage, from input to treshold stage.
How about that? Bad idea...or not?

[simonbaker]

"Picking" arround treshold and sense stage can give some better results.
I agree with you.
But i never tried anything arround TX. Will try at next handmade.
Other thing i would like to hear from you; how about rising a bit gain in all stages in chain? Starting from LF353 up to sense stage...
TGS offers very descent depths, i am pretty satisfied, but sometimes i have feeling it can act much "shortly". Sometimes seems like signal faint..How can i explain? It is detecting deep coins very accuratelly but audio is faint. I modded audio many times and gained much stronger response on target but i lost deepest coins than!? Now is easy to miss some deep or very small coin if sweeping coil to fast or a bit detached from soil surface...So i was thinking to rise gain a bit at every stage, from input to treshold stage.
How about that? Bad idea...or not?

I noticed in TGSL that threshold voltage for comparators does not go all the way to zero - meaning if your noise is very low, you could modify it lower. I would check that first, because if your noise is just below threshold, raising gain would only give you many false signals. But if noise very low, then raising gain sounds good, or lowering threshold might work similar. Lowering threshold is much easier.

Increasing gain trickier because you have to change resistors and capacitors just right or you will change filter shape, although maybe not too critical.

Cheers!

-SB

[ivconic]

I noticed in TGSL that threshold voltage for comparators does not go all the way to zero - meaning if your noise is very low, you could modify it lower. I would check that first, because if your noise is just below threshold, raising gain would only give you many false signals. But if noise very low, then raising gain sounds good, or lowering threshold might work similar. Lowering threshold is much easier.

Increasing gain trickier because you have to change resistors and capacitors just right or you will change filter shape, although maybe not too critical.

Cheers!

-SB

Sheesh!? What a coincidence!? Few days ago i "played" with what you just explained. You are very correct; resistors and capacitors can not be changed with more different values - IT DOES change filter shape and affects overall behavior aka working much worse. Detector acts like there is major phase shift in front end!? I was very confused.
So, yes; you are right; those values are very critical. I just experienced that.

[ivconic]

I tried to raise gain. Here is schematic of what i done; results were bad. This acts like phase is shifted to much. Iron detects very average but coloured metals does not detect at all, in both modes. There is some "spitting" audio reaction on coloured metal, but not clear audio. So it abhor and spit on coloured metals !!! Just what i always needed (like hell) !

[Sean_Goddard]

Ivconic, you missed off the negative supply at the bottom of the sensitivity chain. This is the BIGGEST single mod possible. Also you added and extra diode on the final LF442 stages. Why, the existing Tesoro design only has one. Might I suggest ONE change at a time then you can eliminate whatever is causeing the problem i.e. go back to a standard TGS then change the filters one component at a time (in BOTH filters to keep the paths balanced obviously) then you can report on changing what does what.

I take it the fundamental design works OK with no mods?

You can change the values of the capacitors / resistors in the "filter sections" to allow a slower sweep, but as for changing the gain of the LF353 ! I tried this and for some reason, it screws the whole machines balance up. What I think happens is that there is increased "ground signal" seen by the machine and this causes problems. I've not done a full investigation into why thiat is s problem, but it is so I left that stage alone.

Not much else can be done, you COULD take the demodulators out of the Magnum by Andy Flind and replace the TGS ones as these offer a MUCH better S/N ratio. Also replacing the oscillator with a digital one and having better zero Xing detectors that Tesoro have would help a LOT. The problem with Tesoro Xing detectors is that if you look at the signal on the demodulator output, there is a HUGE spike on the end of every cycle and this is in opposition to the incoming signal. What this appears to do is cancel out quite a bit of the wanted signal and thus reduce the overall sensitivity. This problem also causes the psitting and popping you hear whne using high gains. Eliminate this and you can run the machine to almost the noise floor and get AWESOME gains in sensitivity and depth.

Also some MUCH heavier filtering on the supplies would help. have you looked at the ripple on the negative supply of a TGS ?

[ivconic]

I will return original components first, you are right.
I havent changed anything arround front end. There, all components are critical. I noticed different phase shift there, even when replaced LF353 with some simillar dual opamp. TL082 works fine there, but NE5532 not, for example..

[simonbaker]

Can you say which components you are changing each time? My eyes too lazy to look for all changes...

I think Spice is best way to experiment with gain so you can see effect on filter frequency/phase - equations too thick!

-SB

[porkluvr]

... you COULD take the demodulators out of the Magnum by Andy Flind and replace the TGS ones as these offer a MUCH better S/N ratio.
.... and having better zero Xing detectors that Tesoro have would help a LOT.

By replacing the demodulators, do you mean full wave synchrounous demodulation with a SPDT analog switch?
Or, Bandito uses SPST analog switches for half-wave demodulation - would that help eliminate the spike that you mentioned?

Tesoro doesn't seem to be very intent on making the ZCD outputs have 50/50 symmetry in their detectors, so, full wave demodulation seem like it would require lowering distortion in the tx reference (ZCD inputs) for greatest benefit.

Which "better" ZCDs are you referring to?

[Sean_Goddard]

Use full wave rectification, like Whites do. As for the ZCD, try this;

http://sound.westhost.com/appnotes/an005.htm

or have a go at modifying the existing ones. One thing you could do is replace the direct connection to ground with a 1K resistor then put a 100nF cap across it do decouple it. This would allow the comparator input to swing a little at each transition, but being on the OPPISITE polarity input will help reduce the spikes. As to the 50/50% ratio, that is quite critical if you are using a full balanced Demod like Andy's.

You may also have to "tweak" the Tx waveform for maximum purity which, funnily enough, can be done with that FET and diode limiter circuit that Ivconic removed from his design and changing the value of the resistor I mention by replacing it with a potentiometer .

I know this is going to sound weird, but Simon, emulating a detector in spice almost never works. Beepers are just another example of "I know what works, and what don't. Don't ask me why it don't, it just don't" type electronics where experience counts for 95% of understanding what is going on.

[simonbaker]

Use full wave rectification, like Whites do. As for the ZCD, try this;

http://sound.westhost.com/appnotes/an005.htm

or have a go at modifying the existing ones. One thing you could do is replace the direct connection to ground with a 1K resistor then put a 100nF cap across it do decouple it. This would allow the comparator input to swing a little at each transition, but being on the OPPISITE polarity input will help reduce the spikes. As to the 50/50% ratio, that is quite critical if you are using a full balanced Demod like Andy's.

You may also have to "tweak" the Tx waveform for maximum purity which, funnily enough, can be done with that FET and diode limiter circuit that Ivconic removed from his design and changing the value of the resistor I mention by replacing it with a potentiometer .

I know this is going to sound weird, but Simon, emulating a detector in spice almost never works. Beepers are just another example of "I know what works, and what don't. Don't ask me why it don't, it just don't" type electronics where experience counts for 95% of understanding what is going on.

I agree, the "detector" part is not spice-friendly -- I was thinking of the filter amps later on.

I'm not clear on usage of ZCD term and Golden Sabre -- GS uses synchronous detector, not same as ZCD, right?

Now it's true that GS synchronous detector only uses half of waveform, throws out other half. I'm interested in ways to use the other half without too much circuit complication. I have looked at some possibilities and want to try. Theoretically should increase signal-to-noise a little (good), but extra circuit adds some noise too. Probably not too exciting improvement, but I want to try anyway.

I'm not sure what question is about 50/50 symmetry. Are we talking about pulses driving the synchronous detector? I notice ground channel has almost triangle pulse. I tried shaping the pulses for symmetry, making square, but it didn't improve anything for me. TGS is very much a motion detector, and small absolute phase levels caused by nulling or certain drive signal shapes are filtered out anyway.

I also got interested in pure zero crossing ideas for simplified MD, but it seemed that the discrimination was harder to make immune from depth, at least in my designs. I'm interested in other ZCD design ideas here.

Cheers,

-SB

[Sean_Goddard]

Simon,

ZCD = Zero Crossing Detector. This is what te LF353's do in the GS. They are used to "square" up the Tx sinewave but on the TGS the ouput is not 50:50 square wave (YES these are the pulses driving the demod). By using a full wave demod (simply rip out the FETS and make up a small PCB instead) I'm betting you could run the TGS a lot "hotter".

Changing the filters will only change the response to the speed the detector is swept at although it will give some perceived depth gain if you increase the time constant. Decreasing the TC will make the detector "whippy" like the first motion machines (remember how fast you had to sweep them)?

To get better sensitivity you will have to make your own search head and balance it almost PERFECTLY. Troy proved this with his 7" search head on the X2 and X3. The depth and sensitivity that machine got was way above the normal Silver Sabre uMax, but he did it by making really good coils.

Tesoro coils are rubbish really. OK they work, but the balance is far from perfect and I admit Tesoros are excellent machines, but they could be a LOT better. Funny thing is, and most people don't seem to realise this, the SEARCH COIL is the metal detector, the better the coil, the better the detector. That box at the top of the stem is just a thing that goes beep by analysing the sugnal returned from the coil and the better the balance, the more information is returned etc, etc.

I used to sell the "Turbo Coil", 11" dia spider made to my specification in Turkey. I got some reports of silly depths with those and excellent finds but the build quality sucked. If I could find a manufacturer who would make them again I would happily have a few 1000 made as I KNOW I could sell them.

[ivconic]

I was thinking to adapt TGS/TGSL handmade to work with newer Tesoro coils. Already noticed certain simillarity between Vaquero coils and Musketeer coils ( 1mH and 15mh approx.) Since i already have several Excellerator coils (most of those seems to be for Minelab and White's ), made very good, i am looking a way to adapt TGSL to some of those coils.
Kind a tired of making my own coils for TGSL. Lot of trivial job and much expenses. Why not adapting some already available (and made so good - hard to compete) coils?
So...i am wandering; would it be easy to do that?
Any ideas?

[simonbaker]

Simon,

ZCD = Zero Crossing Detector. This is what te LF353's do in the GS. They are used to "square" up the Tx sinewave but on the TGS the ouput is not 50:50 square wave (YES these are the pulses driving the demod). By using a full wave demod (simply rip out the FETS and make up a small PCB instead) I'm betting you could run the TGS a lot "hotter".

Changing the filters will only change the response to the speed the detector is swept at although it will give some perceived depth gain if you increase the time constant. Decreasing the TC will make the detector "whippy" like the first motion machines (remember how fast you had to sweep them)?

To get better sensitivity you will have to make your own search head and balance it almost PERFECTLY. Troy proved this with his 7" search head on the X2 and X3. The depth and sensitivity that machine got was way above the normal Silver Sabre uMax, but he did it by making really good coils.

Tesoro coils are rubbish really. OK they work, but the balance is far from perfect and I admit Tesoros are excellent machines, but they could be a LOT better. Funny thing is, and most people don't seem to realise this, the SEARCH COIL is the metal detector, the better the coil, the better the detector. That box at the top of the stem is just a thing that goes beep by analysing the sugnal returned from the coil and the better the balance, the more information is returned etc, etc.

I used to sell the "Turbo Coil", 11" dia spider made to my specification in Turkey. I got some reports of silly depths with those and excellent finds but the build quality sucked. If I could find a manufacturer who would make them again I would happily have a few 1000 made as I KNOW I could sell them.

I agree about the filters - I'm experimenting a little with slower designs to see how they feel. I'm pretty sure Tesoro arrived at a very good compromise, hard to improve on.

In the current design, slower filters will create a delay so the beep will be off target -- hard to work with. But as you say, could detect a little deeper. I want to take the original TGS design and convert the pinpoint circuit to another motion circuit with a slower filter and different tone. This would allow you to slow your sweep to try to identify a marginal detection you hear on the faster channel. Just an idea.

For squaring up waveform -- do you mean LF353s, or comparators?

I see the JFETs and LF353's a little differently. The JFETs look like a synchronous detector to me, integrating the half-signal from the RX coil using capacitors C15 and C12 (ref TGSL Ivconic message #5 above).

The LF353 connected to RX coil looks like gain amp. The LF353 connected to TX coil looks like isolation buffer as part of phase shift circuit to me.

If I remember, it is the "All Metal" drive signal that has weird triangle shape since phase shifting comes after comparator. Could we move phase shifting before comparator as possible mod?

Maybe improving shape would help, but my crude experiments didn't show any advantage. However, I need to test more, simple theories often inadequate when detecting tiny signals buried in noise.

I'm very interested in your ideas about improved coils and how the improvements work.

Best regards,

-SB

[ivconic]

This debate turned to very interesting and educating. Thanks to both of you guys.
I agree, TGS is not easy to improve. Generally i am very satisfied with what i gained so far. TGSL preforms excellent on small coins in ground.
But...main goal here is not to make device and stay satisfied with it! Right?
Main goal for me is to constantly experiment and try to improve it at least a bit. So i did "million" small experiments and mods. Most of those gave no improvements. Very few gave some results. I can name a list of all what i done so far, but no need for that. I focused latelly only on audio stage cose that showed to be weakest part of design. BD140 version gives good compromise and i am using it now.
But.. something else is not clear to me. TGSL is great for coins...small items..perfectly reject unwanted targets etc...etc... But does not have good "depths" on very large objects...like...Fisher 1265 for example.
Let me explain; TGSL is better and more powerfull than 1265 on small coins in soil. Discrimination is also much more accurate than at 1265. But...1265 reacts on large metal plate (20x20) at much more distance in air and soil than TGSl??? How come?
With Sense adjusted to max you can run 1265 to be very sensitive and reactive on larger items. On some objects it can react at more than 1m in air. Other hand TGSL can not?? How come? If we know previously that TGSL successfully detects 1e coin at ...let's say 25cm in soil, while 1265 bairly detect same coin at 15cm in soil (just an example).
Something is missing here? Something i still dont understand and would like to learn. It is almost like TGSL has "limited field" of detection but all items in that field are detected very well. And 1265 has more amplification and can react on larger distances but it is generally weaker at small items in soil, than TGSL... (don't guess me wrong; 1265 is also very good for coins...but not that good as TGSL).
What relations are those? Scoping TX output at both machines i cant see any significant differences (except frequency).
It is like TGSL need more amplification,gain...or what?
Need to act more "wild". At very small and distant targets it tends to produce faint signals. And this is not only the case with my handmades. I had chance to see and test few original Tesoro machines - same case; good but "limited"...
So... any explanations?