TGSL Experiments

[simonbaker]

parallel reverse diodes - better???

I tried adding another diode in reverse direction in the TGSL amplifier op amp section. The idea is to limit the negative going pulse in addition to limiting the positive pulse. The purpose was to perhaps improve the response a little if possible.

A simulation suggests that this might slightly improve the response, but not by much.

Just another tweak.

-SB

[simonbaker]

This sort of filter would help - followed by Amp stages.

Transient response is lke 10ms rather than 0.5second for some tight filters.

Could be worth a bash if someone has the time. Or the will.

S

This might make a good pin-pointer since it passes DC (but you'd have to deal with the offset somehow). However, as I mentioned, you have increased your noise power by extending the filter band, so you may lose depth because of that.

What are you using for you target pulse? I have been approximating it as a trapezoid, sort of compromise between triangle and square. Or sometimes a half a sine wave.

-SB

[simonbaker]

So if we could get an filter/amp with a good time response at 1Hz and good gain and Stability we would have a tangible increase in detection of smaller targets.

S

I'm not sure it's possible to get good gain and S/N at 1 Hz and good time reponse - maybe mutually exclusive? That's why I'm thinking about a control for selecting the response as needed.

But maybe there is some improvement possible. If you just move the lower cutoff lower (to increase gain at 1 Hz), you'll increase the noise power overall, so it is probably a real design problem to pick that "perfect" filter.

Certainly worth continuing to play with.

-SB

[golfnut]

Faster amp with more gain

This sort of filter would help - followed by Amp stages.

Transient response is lke 10ms rather than 0.5second for some tight filters.

Could be worth a bash if someone has the time. Or the will.

S

[golfnut]

So if we could get an filter/amp with a good time response at 1Hz and good gain and Stability we would have a tangible increase in detection of smaller targets.

S

[simonbaker]

All, I was looking at what we can do for more bang for buck on IGSL.

A while ago I scoped the detector/mixer output and on a normal sweep of a coin the signal was a single blip sinusoid at approx 1Hz frequency.

I Modelled the post detector Amp - using similar 20MHz BW op amp from LT.


I would expect lots of gain at 1Hz so we boost the signals we want.


I may have the model wrong but, I see the peak gain at 10Hz +25dB gain...


At 1Hz the gain is 12dB lower?

probe the output of R1

Any takers.

attached

Steve

You are correct, it peaks near 10 Hz. I think the reason is for fast response and shaping the pulse a little better.

The problem with lowering the passband of the amp section is that you make a very delayed (slow) response to targets. I actually built a PCB with the center freq somewhere near 1 Hz because I was also intrigued with it. I believe it was in fact a little more sensitive (I'd have to really test in a noise-free environment), but the response was crazy delayed and a little out of control. But I didn't spend a lot of time trying to perfect it.

I still think it has a possible use however. I am slowly working on the idea of a "response" pot where we can dial in the response. If we think we might have detected a deep target, choosing a slower response and slow sweep of our coils might help confirm it. Shallow targets can create a much sharper pulse, but since the signal should be stronger, we don't worry about it being filtered down a little.

I think digital processing is a perfect platform for choosing the amp filter characteristics any way you want. Because we are looking at such low frequencies, I'm wondering if even feeble microcontrollers (like the MPS430, etc.) might be able to handle it (with a little analog pre-filter help). I'm slowly exploring that.

-SB

[simonbaker]

That's a good question, I don't think the original TGS had a non-motion mode.

One way to approximate it would be to change certain capacitors in the amplifier section to make the bandwith go very close to zero, without changing the upper cutoff. -SB

I said that a little wrong -- I meant make the lower cutoff go near zero Hz without changing the upper cutoff (which is somewhere between 10 and 20 Hz I think).

-SB

[golfnut]

100uF rather than 4u7 gives gain at 1Hz

Would need 2x 220uF's back to back or 100uF non polarised

S

[golfnut]

Amp gain response

All, I was looking at what we can do for more bang for buck on IGSL.

A while ago I scoped the detector/mixer output and on a normal sweep of a coin the signal was a single blip sinusoid at approx 1Hz frequency.

I Modelled the post detector Amp - using similar 20MHz BW op amp from LT.


I would expect lots of gain at 1Hz so we boost the signals we want.


I may have the model wrong but, I see the peak gain at 10Hz +25dB gain...


At 1Hz the gain is 12dB lower?

probe the output of R1

Any takers.

attached

Steve

[simonbaker]

Can any of you here tell me if theres a easy quick mode to be able to switch to non motion on this project please.


Thanks in advance

That's a good question, I don't think the original TGS had a non-motion mode.

One way to approximate it would be to change certain capacitors in the amplifier section to make the bandwith go very close to zero, without changing the upper cutoff. I'll check with a spice sim to see if making C15/C12 and C18/C21 much larger would do that.

Another way involves one of those feedback loops that zeroes out the offset due to the null combined with a true DC amplifier section.

I'm not sure how this would distort the response, have to check. Maybe wouldn't work.

What application would you have for that mode?

-SB

[satdaveuk]

non motion/pinpoint mode

Can any of you here tell me if theres a easy quick mode to be able to switch to non motion on this project please.


Thanks in advance

[satdaveuk]

You will probably find that using metal pots in an metal enclosure (and grounded to 0V) will make the falsing go away.

Yes maybe but maybe not, I personally have never had falsing problems of any form since changing the oscillator TRS back to plastic package types which often are marked PN instead of 2N ie PN2907 instead of 2N2907.
All I advise is change them to transistors for the types ive mentioned and im pretty convinced you wont have anymore problems.
If I can run these circuits in a all plastic housing trouble free in workshop conditions its certainly proved a point.
As i said previously before I bonded the coil socket and was using metal package oscillator TRs the detector was near on useless.

The only other difference between are detectors is maybe the circuit layout im using Ivonics PCB layout which I etch myself but the actual circuit is the same as Silverdogs, done all the mods incl the preset in the TX tank circuit, and these mods only make the circuit more sensitive so if anything it would agravate falsing etc.

Actually building another TGSL now so will be interesting to see if performance can be repeated or maybe improved, the quality of the PCB on this one is top notch with a heavy grade copper print and I shall measure every component before installing, all CXs will be external so I can play easer.
Just replace the two TRs and you will have a nice usable machine.

All the best

[Qiaozhi]

Im not using a metal housing I made it out of styrene, by using all plastic carbon pots the only thing needed to be bonded to 0v was the coil socket.
The only other difference to some builds on here is the speaker is in a independent housing.

You will probably find that using metal pots in an metal enclosure (and grounded to 0V) will make the falsing go away.

[satdaveuk]

Did you connect the metal enclosure to 0V?

Im not using a metal housing I made it out of styrene, by using all plastic carbon pots the only thing needed to be bonded to 0v was the coil socket.
The only other difference to some builds on here is the speaker is in a independent housing.

[Qiaozhi]

Hi Dennis
The symptoms with metal package type TX oscillator transistors were when touching knobs, housing wires within the detector and external coil lead was getting signal break through ie falsing.
When I first made the project had used plastic types, only changed to metal to see if it would give me a little more gain on depth, so basically copied the original transistors from tesro.
It was at that point I had created the problems related to on here, so once changing back to plastic all those problems went away.
You will find that just touching or moving wires anywhere near those metal type transistors will cause the machine to go unstable but again when plastic type are fitted no problem.
Did mention this many moons ago but has got buried amongst hundreds of posts in the original forum.

Regards

Did you connect the metal enclosure to 0V?