IGSL

[simonbaker]

As I bragged already - my IGSL PCBs came today so I can start with trial and error engineering

A little bit on spice: it is a tool. And a good one. No real world component will give up smoke if you overdo something in spice, but a ton of surprises may come upon your precious little thing in a real world. That's for sure.

Now a look from completely different perspective. I used spice to explain myself, and everyone else with a shared interest who cared enough to fire up my sims - what exactly happens there. And what can be done to make it work better if possible. For example, I found that names for disc sens and threshold functions are in fact mutually exchanged: sens works by means of setting a threshold to trigger a comparator, while threshold trimmers actually control audio sensitivity or how loud a response will be.
I also found out the role of diodes in feedback paths of gain blocks, and a possible source of chatter - I'll try it first in vivo to be sure.

I also found out that the whole discrimination business is a 2 quadrant thing. I can live with that. It explained me the importance of connecting Rx loop in a proper polarisation to receive one beeeeeeep instead of two short beep ... beep which happen at phase reversal of a motion compensation filter.

Now I can start building my first MD with ease and competence. I can say I truly understand every bit of it.

I think changing direction of feedback diodes is interesting to try, look forward to how it works out, hope to try it too some time. Those pot labels can be surprising, yes. Threshold pot effect depends on how you set other params around the op amp, such as gain, etc.; in stock TGSL, audio volume effect and chatter delay effect are not independent. SD ignores half cycle for simplicity. I think sims are very useful to help demonstrate the concept at work and try out creative ideas. I'll bet there are great circuits that fail to work in sims and vice versa. Sounds like you're ready to burn resin -- go for it.

Cheers!

-SB

[Aziz]

Aziz make spice simulation of one damn good MD in you'r laptop, get the laptop outdor and start swinging and diging good finds. LoL Just joking :P

*LOL*

I am trying many many ideas & making many many errors...
Since I am simulating spice, the number of trials and errors has been rapidly increased.

Hey guys, please keep on trying & making errors. It's one of the best methods of learning and discovering interesting things.

Nothing wrong with it.

Aziz

[Davor]

Ah, the sweet smell of resin smoke ...

Some funny people made a vacuum thingie to extract the resin smokey goodness out, as if it was unhealthy. Believe me, my nose and nerves could do with a little resin smoke. It smells soooo good.

[ivconic]

"..Or do you want to realize the MD with one of the mature analog computers?
(see http://www.technikum29.de/en/computer/analog.shtm).."

Why not? Looking very promising!
Is there any of those, but battery supplied?!

Aziz don't worry, i started mockery in my last post only to break humdrum.
Of course no offense, those are childish debates, we are much more serious than that, aren't we?
But, monotony is everywhere, nothing to do, so is better to even argue than to say nothing and do nothing!
...
Of course i don't underestimate value of simulations! C'mon! I am not that illiberal!
Only i don't aprove relying on those too much.
Process must be done in parallel: simulations and material work, otherwise it will not do, at all.
...
Inversing diodes? Reverse biasing?
1N4148 will not do!
Some other type maybe.
How about using transistor as diode! (what a brilliant (and original) mind i have !!! )
...
Davor please do finish that pcb and than keep us informed.
But don't "simulate" too much, obtain proper coil and check how whole setup is behaving prior than applying any of your fresh ideas.
First be sure if that is working at all.
Without proper coil and proper behavior you will get nowhere.
Later, when all is in working condition; we can talk.
...
Cheers!

[ivconic]

Usefull as reminder:

!?!??
I lost the link!!!

Sorry!

[Aziz]

BTW, beware of LTspice models (parts) in simulations!!!
The internal semiconductors (diodes and transistors) aren't often modelled right (except their own parts of course). They just model the basics. You need the manufacturer's specific spice models to get more accurate results.

That's often the reason, why the simulation result differs from the real circuit. Always have a look at the spice model file. Some description is sometimes there (implemented features, convergence aids, etc..) .

Aziz

[Davor]

Very true. Most of the reverse operation physics is not modeled at all.
On the other hand - life is too short to do it grandpa-style with only a welding iron an an oscilloscope. Such approach often leads to order of magnitude errors.

Anyway, I have not one but two PCB-s to play with, and a few more ideas. And a well behaved Weller magnastat soldering iron. And some other things to do as well. OK, we'll see how it works out.

[ivconic]

For those who are struggling with coil for IGSL ver2, (and SMW) :

http://www.geotech1.com/forums/showt...193#post152193

[growdude]

Musketeer coils for sale

15" wot 100$ shipped in us overseas actual shipping cost apply12" excellortator coil 95$shipped in us overseas shipping actual cost apply. Good for igsl/tgsl projects they are on ebay also.

[ivconic]

15" wot 100$ shipped in us overseas actual shipping cost apply12" excellortator coil 95$shipped in us overseas shipping actual cost apply. Good for igsl/tgsl projects they are on ebay also.

Bargain prices! I payed 180e for my 12" coil.
Those coils are good for IGSL ver2, TGSL with IGSLver2 oscillator, SMW (Caesar) and i guess many other projects.
Also i am sure those coils are easy adaptable for newer Tesoro line detectors, such are Vaquero, Cibola, Tejon...
Those are untuned coils and that makes them very sweet for homebrew projects.

[UFO]

Yes simulations can be helpful, sometimes a necessity, even if they are not perfect.

On the other hand, just like a carpenter, it's all how you use your tools. A good carpenter can do more with less then a bad carpenter with all the fancy tools.

Here is a good watch, how NASA used computers to develop the space program etc.

They accomplished a tremendous amount with what is considered very little by today's standards.

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

[Davor]

I've found one mistake in "Ferrous channel" and it will not be too easy to fix on the existing board. It is related to the inverting/noninverting inputs of comparators U7d and U4c. They both need to be reversed for correct operation, and it will not be trivial

Let me explain. Comparators contained in U4 clock the 4066 switches by phases set by the preceding circuitry. Resulting signal is amplified in respective channels and fed to U7 comparators that form analog AND circuitry. For non-ferrous detection GB AND non-ferrous disc signal make for non-ferrous detection. To make them less hectic there is a "sens" potentiometer control that actually provides a threshold below which there is no detection and it works fine for non-ferrous channel ... but not for the ferrous discrimination channel.

It becomes active when there is a positive signal on ferrous GEB channel, but also for 0V on ferrous disc channel - which makes no sense. It is supposed to be active only for negative signal in ferrous disc. channel.

It can be kinda corrected by "Nonferrous accepting level" potentiometer, but it is level dependent and gives oddly overlapping results at low target response levels. I guess lots of chatter too.

I'd propose a solution with exchanging pins 8 and 9 of U4c and pins 10 and 11 of U7d. It would make "Iron sens" control behave as expected and overall behaviour more predictable. It would reverse phase of the ferrous disc. channel, and it will make discrimination comparators work for positive signals, and it will make "Nonferrous accepting level" potentiometer actually do its job.

In my builds I'll have to make it "ikebana" style by first extracting the existing sockets, doing some nasty acrobatics with wires and it should work correct then. Too bad I did not notice it before the sockets got welded Maybe I'll better cut the lines on PCB and rewire everything there - seem like a bit less trouble.

here is a picture of the proposed changes:

[ivconic]

Davor have you tried that change already?
I am looking forward to read your first impressions.
Have a luck!

[Davor]

I'll do it for sure, and I hope to have everything completed by weekend. Most probably I'll get busy balancing the coil, but anyway, It will be soon.

This correction is not the most obvious one because phase needs to be reversed somewhere, and it seem as if it is all the same if you do it at U4 or U7, but it isn't. There is a threshold "sense" pot that requires positive signal to perform correctly.

[ivconic]

I'll do it for sure, and I hope to have everything completed by weekend. Most probably I'll get busy balancing the coil, but anyway, It will be soon.

This correction is not the most obvious one because phase needs to be reversed somewhere, and it seem as if it is all the same if you do it at U4 or U7, but it isn't. There is a threshold "sense" pot that requires positive signal to perform correctly.

Treshold works perfectly well on both blocks, already.
But i agree with you that there are better ways to process Fe signal there.
Obviously many people already faced problems at Fe block. I don't know why?
All the time i was convinced that they do have problems with their coils.
At my IGSL's Fe block is acting pretty calm and smooth as well as Nfe block too.
Davor keep on going, i am interesting to see your results.
As i said; i have 2 finished IGSL's here and it would be no effort for me to apply your suggestions on those and check them.
But next few days i am very busy with something else, i think you will do that much faster than me.
Cheers!