TGSL Tuning & troubleshooting

[satdaveuk]

Dont know if this will shed any light on the matter but after doing the ivonics mod with the preset in the TX stage noticed it did make quite a big shift to the disc settings ie pushed them further up the scale, hence im using both my TGSLs with disc on and disc control fully anticlock wise.
Bare in mind this happened with both machines using the TGSL circuit, so certainly has a bearing on it.
the advantage of the mod was home made coils it gives that little bit of tweeking for nulling error.
If using tesoro manufacture coils best leave as TGSL circuit described.

In real use my ground balance on both machines is around 5-10 two in clock format anything from that either falsing or low sense, that with cost is why I didnt go any further with the multiturn Idea

[Jerry]

Hi Jerry
some have used multiturn pot for the gb on here trouble is there so exspensive, maybe it would be worth trying one on disc if you have one.

Hi Dave,

I did that and it works great for ground balancing. Now I still need to figure out the best method to use. The reason I have not tried it for Disc is I like having a single turn knob that I can turn with my thumb until the beep goes away and then glance down to see where the Disc pointer is. I don't even have to look at my detector otherwise.

Thought about it some more last night and I think I am going to get a good Disc pot installed, along with the 22 pf cap mod and then look to the TX oscillator to shift the disk scale CCW. It isn't much, the red and blue dots in the picture show the amount I am talking about. It was at the yellow dot for nickels before any mods and now it is at the red dot. In fact if I get my disc scale set up an working as well as this Tesoro picture I will be happy.

Jerry

[satdaveuk]

Hi Jerry
some have used multiturn pot for the gb on here trouble is there so exspensive, maybe it would be worth trying one on disc if you have one.

[Jerry]

Hi Jerry:
You may find that just replacing the bad pot (plus your capacitor phase shift) may be all you need -- most people's TGSLs seem to work great with really ugly looking signals at u101a pin2 (input).

Note that by adding the capacitor, you probably shift the whole range of the DISC control -- so what you gain at one end, you lose at the other. I don't know if that is good or bad or indifferent. Maybe you can tell us.
-SB

I think just changing the pot and adding the cap is a good idea for now. Now I just have to find a 100K linear pot that does not have a dead band at the end. I have several new pots on hand and all of them have the same problem to some extent. It is ironic, the pot that shows the least dead band is the one I removed that was under value. The problem with that one is the threaded collar is smaller in diameter than the present one so I have already enlarged the hole in the front panel.

I don't know if I am going to have time to work on it tomorrow or not but will report back when I get everything done.

Talking through these problems is a help in thinking them through. I appreciate your help.

Jerry

[simonbaker]

I did some more checking on the 100K Disc pot looking for the reason why I was getting a constant reading of 3.3k for the first 20% or more of the knob rotation as described in my post #1039.

As it turns out, the pot is defective. The resistance reading should change from either end relative to the wiper as the shaft is turned with no dead zones. This pot, which is brand new does nothing for the first 20% as the knob is rotated clockwise. Checking from the other end, the resistance starts to change immediately as the shaft is turned. I suspect that something is wrong with the internal setup when they were made. I have several other identical pots that do the same thing.

So the mystery of why the Disc pulse would not move and it had a double bump until that point is solved. Chalk it up to bad parts!

The double bump issue is a matter of the 3.3k resistor at R19. I think increasing that slightly will make the double bump just go away and I don't think it will affect anything else with the discrimination circuit.

I plan on changing the Disc pot to a 50K and increasing the resistance of R19 to 47k to start with and see what that does to the band spread on the Disc scale. I am sure I will have to play with the numbers a little.

Now that I think a little more about R19, this is part of what Simon accomplished with his voltage divider mod which he posted here: http://www.geotech1.com/forums/showp...9&postcount=73

Jerry

Hi Jerry:

Good sleuthing!

You may find that increasing R19 is not a good solution and prevents full phase shift (off the top of my head). That's because the pot is supposed to allow variable weighting between two signals that are 90 deg phase shifted:

Asin(phi) + (1 - A)sin(phi + 90) = wiper signal

where A varies from 0 to 1. This creates a signal whose phase varies from 0 to 90.

If you increase R19 to, say, 100K, you may get something like:

Asin(phi) + (1 - A/2)sin(phi + 90) = wiper signal.

This won't make the full phase shift.

That is partly why I put the divider before both signals. The other reason was to prevent the LF353 from overdriving and causing the ringing and clipping of the LF353 output signal.

You may find that just replacing the bad pot (plus your capacitor phase shift) may be all you need -- most people's TGSLs seem to work great with really ugly looking signals at u101a pin2 (input).

Note that by adding the capacitor, you probably shift the whole range of the DISC control -- so what you gain at one end, you lose at the other. I don't know if that is good or bad or indifferent. Maybe you can tell us.

Loved the rainbow and landscape in general -- and the idea of toiling like a mad scientist in the shed while the weather rages outside - perfect.

-SB

[Jerry]

Junk Parts

I did some more checking on the 100K Disc pot looking for the reason why I was getting a constant reading of 3.3k for the first 20% or more of the knob rotation as described in my post #1039.

As it turns out, the pot is defective. The resistance reading should change from either end relative to the wiper as the shaft is turned with no dead zones. This pot, which is brand new does nothing for the first 20% as the knob is rotated clockwise. Checking from the other end, the resistance starts to change immediately as the shaft is turned. I suspect that something is wrong with the internal setup when they were made. I have several other identical pots that do the same thing.

So the mystery of why the Disc pulse would not move and it had a double bump until that point is solved. Chalk it up to bad parts!

The double bump issue is a matter of the 3.3k resistor at R19. I think increasing that slightly will make the double bump just go away and I don't think it will affect anything else with the discrimination circuit.

I plan on changing the Disc pot to a 50K and increasing the resistance of R19 to 47k to start with and see what that does to the band spread on the Disc scale. I am sure I will have to play with the numbers a little.

Now that I think a little more about R19, this is part of what Simon accomplished with his voltage divider mod which he posted here: http://www.geotech1.com/forums/showp...9&postcount=73

Jerry

[Qiaozhi]

The last picture is probably the best rainbow I have ever seen.

A good way to end the day.

Jerry

If you look to the left of the photo, you will see that it is in fact a double rainbow.
Maybe there were two pots of gold.

[Jerry]

Reference my last post. I had turned in for the night but got to thinking about that 3.3 k resistance reading on the Disc pot wiper. It really isn't apparent at first glance but there is a 3.3k dc path to ground through the coil. Coil isn't shown on the schematic so I missed it earlier.

This should pave the way for the mod i had in mind. which would be to replace the 100k with a 50 k linear pot plus 47k of fixed resistance. This would have the effect of band spreading the disc dial so that things like foil and coins would be spread out over the entire range of rotation instead of being compressed in the upper half only.

Will think on it but it is looking good.

Jerry

BTW No pot of gold Dave

[satdaveuk]

Interesting find Jerry may follow that mod at somepoint on my machines and see if i get the same result.
Did you go to the end of that rainbow and check for gold?

Regards

[Jerry]

Discrimination Issue Solved!

Well I made a lot of progress this afternoon and I think I made some discoveries as well.

I checked the scope signals at U-102 pin 2 and also at U-102 pin 1. Overall, the output of U-102 (Pin 1) proved to revealed the most information that was useful for my problem. I have noted before that this square wave (The Disc Pulse) would only move so far while turning the Disc control clockwise and it would hit a point were it would stop advancing in step with the Disc knob.

I Added a 68 pf Cap across feedback resistor R17 and now the Disk Pulse moved much further to the right as compared with the TX pulse. This increased my discrimination by a lot, too much even. With the 68 pf cap on R17, I can discriminate out U.S. Quarters! This is more than what I was looking for so I reduced it to 22 PF and that will disc out zinc pennies and not quite silence copper pennies and dimes. Exactly what I was looking for!

It did not settle the issue of shifting where nickels discriminate out. It used to be five on my very crude scale of 0 to 9 and now it is a little over six. (6.1) I may revisit the TX operating frequency but I can live with what I have now. Especially since a little further experimenting may have given me a good work around.

As of right now, my TGSL discriminates as follow. On a scale of 0 to 9:

Small Foil ........... 5
Nickel.................. 6.1
Pull Tab............... 6.3
2 Euro Coin ........ 7.0
Zinc Penny ......... 7.5
Copper Penny...... 8.0
Quarters cannot be discriminated.

Notice that the discrimination scale is compressed into about half of the rotation of the Disc control. With that in mind, I shut the detector off and started measuring DC resistance between the wiper of the Disc Pot R18 and ground. I discovered that from 0 to about 2.2, the reading stayed at a constant 3.3K. This has me puzzled since I not see a 3.3K path to ground on the schematic. Will have to check R19 for proper placement tomorrow.

Sticking with that 2.2 number a little further, earlier I had made a note that the Disc pulse would not start moving to the right until it reached 2.2 on the disc scale. This is also the same point where the pulse actually started to move the double blip in the Disc pulse went away!

I was going to mention another possible change but will hold off until I find out for sure the status of R19.

I knew this was going to get a bit long, so I dumped the pictures of U102 pin 1&2 into single images. The first one is captioned and is of the input of U102, pin 2. the second is of the output at pin 1 and the captions for disc settings would be the same.
I also included a pic of the targets I used as test subjects for the discrimination test.

It rained most of the time I was taking the pictures and I had to wait a bit so I would not get wet walking back to the house. The last picture is probably the best rainbow I have ever seen.

A good way to end the day.

Jerry

[simonbaker]

Hi Simon,

Just walked out to the shop to confirm what I was hooked up to and you are correct, it is Uio1b pin two that I took pictures of. I must have been more tired than I thought, I just hooked it up quick, snapped a few pictures, shut everything off and called it a day.

It is raining here today so I think I can get some more pictures. I will check U101b output pin 1 and also U102a pins 1 and two.

However the last series of pictures is still interesting in that it shows a huge gain in the pulse along with increased ringing with increased distortion. So it was still useful in that sense.

Will post more later this afternoon.

Jerry

Ok. BTW, I think the ringing mostly happens when the LF353 "saturates" due to out-of-bounds input, and clips the output, going nonlinear. I'm guessing that your input signal you scoped suddenly spikes when the op amp is decidedly outside the linear range. The original TGS had a lower TX oscillator voltage which I believe did not cause this condition -- we've basically "super-charged" our TGSL oscillators by removing the regulator JFet.

The TX signal "blip" also can make a really hefty blip in the LF353 output, but it doesn't actually ring I think if the LF353 is operating in a linear mode; but it can cause some weird DISC phase shift motion sometimes.

-SB

[Jerry]

Hi Jerry:

I've seem some crazy waveforms at u101b pin 1, but yours looks particularly unfamiliar - can you confirm you are measuring what I think, which is: the output pin of LF353 that connects to the DISC pot?

I think maybe you are scoping pin 2, the inverting input of U101b (LF353). The output is more revealing as to how it affects the sync pulse.

Also, to really see what is driving the sync pulse, try scoping U102a (LM393) pin 2, the inverting input of the comparator connected by resistor to DISC pot wiper. Then as you turn the DISC pot, you can really see why the phase shifts where it does. Maybe that is what you meant to do anyway.

Regarding the "double" sync pulse that happens at one extreme of the DISC pot -- I had originally thought it was due to all the "ringing" in the LF353 output. Today I checked (with dfbowers' TGSL) and noticed that the double pulse happens when the comparator is driven purely by the TX oscillator signal. My guess it is due to the oscillator voltage going beyond the "rails" of the comparator and causing the glitch when the DISC pot mostly selects the oscillator signal.

Regards,

-SB

Hi Simon,

Just walked out to the shop to confirm what I was hooked up to and you are correct, it is Uio1b pin two that I took pictures of. I must have been more tired than I thought, I just hooked it up quick, snapped a few pictures, shut everything off and called it a day.

It is raining here today so I think I can get some more pictures. I will check U101b output pin 1 and also U102a pins 1 and two.

However the last series of pictures is still interesting in that it shows a huge gain in the pulse along with increased ringing with increased distortion. So it was still useful in that sense.

Will post more later this afternoon.

Jerry

[simonbaker]

Hi Simon,

Here are a few pictures showing what happens while adjusting the TX oscillator. The changes with just a little distortion in the TX sine wave is interesting. The bump produces noticeable ringing. The upper trace is the TX signal measured at J1 and is set at 5 V/div. The lower trace is U101b, pin 1 and is set at .2 V/div. Same setting apply to all pictures.

First picture is the TX oscillator (top signal) tuned for a clean sine wave.

Second shows the TX oscillator with about as much bump at the top as was there before I added the adjustment pot.

Third is somewhat in between with just the slightest of bump at the top. Even with that, there is still ringing.

Interesting at how the Disc pulse amplitude goes up with more distortion of the TX sine wave. I suppose the next question is how much is needed for good discrimination. I am too wiped out right work on it tonight but it gives me quite a few ideas.

Thanks for suggesting this, it is interesting.

Jerry

Now that I think about it, I did not even look at the Disc knob settings..... whatever it was, it is the same for all pictures. Will check some more tomorrow.

Hi Jerry:

I've seem some crazy waveforms at u101b pin 1, but yours looks particularly unfamiliar - can you confirm you are measuring what I think, which is: the output pin of LF353 that connects to the DISC pot?

I think maybe you are scoping pin 2, the inverting input of U101b (LF353). The output is more revealing as to how it affects the sync pulse.

Also, to really see what is driving the sync pulse, try scoping U102a (LM393) pin 2, the inverting input of the comparator connected by resistor to DISC pot wiper. Then as you turn the DISC pot, you can really see why the phase shifts where it does. Maybe that is what you meant to do anyway.

Regarding the "double" sync pulse that happens at one extreme of the DISC pot -- I had originally thought it was due to all the "ringing" in the LF353 output. Today I checked (with dfbowers' TGSL) and noticed that the double pulse happens when the comparator is driven purely by the TX oscillator signal. My guess it is due to the oscillator voltage going beyond the "rails" of the comparator and causing the glitch when the DISC pot mostly selects the oscillator signal.

Regards,

-SB

[Jerry]

TGSL U101b-1 Pictures.

Hi Simon,

Here are a few pictures showing what happens while adjusting the TX oscillator. The changes with just a little distortion in the TX sine wave is interesting. The bump produces noticeable ringing. The upper trace is the TX signal measured at J1 and is set at 5 V/div. The lower trace is U101b, pin 1 and is set at .2 V/div. Same setting apply to all pictures.

First picture is the TX oscillator (top signal) tuned for a clean sine wave.

Second shows the TX oscillator with about as much bump at the top as was there before I added the adjustment pot.

Third is somewhat in between with just the slightest of bump at the top. Even with that, there is still ringing.

Interesting at how the Disc pulse amplitude goes up with more distortion of the TX sine wave. I suppose the next question is how much is needed for good discrimination. I am too wiped out right work on it tonight but it gives me quite a few ideas.

Thanks for suggesting this, it is interesting.

Jerry

Now that I think about it, I did not even look at the Disc knob settings..... whatever it was, it is the same for all pictures. Will check some more tomorrow.

[freedan3d]

Exactly.
Proper coil nulling is not an easy job. I think most people understood nulling as just simple task. Pitty, it is not like that.
At some coils i spend 5-6 hours to null them properly. Amateur conditions do not offering me comfort, so nulling is time&nerves consuming tough task.
Than again...at some coils i achieve good balance in just 20 minutes.
It is all about quality made and details.
Usually i am using hot melted plastic glue gun and put "punkts" (points) over the coil to attach it to holder good. Than i must wait hot glue to dry and get cold. Than again to correct balance by slitest (parts on millimeter) movements of some coil segments. Than again to wait to see influence of temperature changes in workshop....and again..and again. All the time RX coil is attached to millivoltmeter and observed. TX coil is attached to TGSL TX out.
Sometimes process is going pretty smooth and everything ends in 20 minutes....but sometimes it took time to be done. It can took hours sometimes.
Once coils are balanced well i use joint filler (2 components epoxy) to finish coils. It all must be done in one move, fast, not to allow temperature changes to affect balance. If you balance coils and let them overnight....those will move for sure and spoil the balance much.
Tough job to make good coil in amateur conditions. Not easy.


....

That's why i am thinking to give up of making DIY coils finally, lately. It is more cheaper to buy original coil and adapt detector to work good with it.

Hi ivconic

I see your tgsl test video and i am so excited
its very cool ,man !!!
and i decided to make that. i finish it now .
but it is not seem to your tgsl depth !
it detect a 1 penny in 20 cm in air and a gold coin (90% gold ,2.03 g) in 15 cm
is it a good depth ?
my coil tx=6 mH and rx=6.5 mH
what is the best coil rx,tx inductance ?

many tanks

freedan3d