IGSL

[Davor]

Ergo the path to a much smarter IGSL. I LOVE the Fe channel and what it does. It is not just iron you "see" with it, but also Al foil by overlapping audio - when I set discrimination that way. With proportional audio I get a feeling of being immersed in a terrain of a kind, like the soil becomes transparent.

I think I can understand why you love Deus so much.

[ivconic]

Yeah, I noticed that. But also that chatter is much quieter than the target response and is not much of a problem. It also gives you some insight about the soil qualities. Proportional audio takes care of that. Funny, but that's kind of my goal. Believe it or not ....

Not funny at all, but true and smart! I agree.
Deus's chatters are least problem, because outdoor on real soil those chatters are not that noticeable nor annoying. Also can disappear when Sensitivity is lowered a bit.
IGSL is calmer (apparently) ... but also much weaker and less sensitive than Deus.

[Davor]

...
For example; at same table i made video "Deus&chain" and if you watch that video you can hear Deus chattering all the time.
...

Yeah, I noticed that. But also that chatter is much quieter than the target response and is not much of a problem. It also gives you some insight about the soil qualities. Proportional audio takes care of that. Funny, but that's kind of my goal. Believe it or not

The whole chain of channel gain stages (without preamp) has sensitivity at ~6µVpp set by the Disc sens setting at 3mV at maximum sensitivity. I wouldn't go any lower with disc sense to avoid confusing the comparators, but I know I can go with more gain. Considering the bandwidth of these stages, I know for sure I can do with more gain, and get better sensitivity, hence more depth. I can always adjust Disc sense in case of problems, otherwise - pedal to the metal. My perfect candidate for this to happen is a second gain stage, that has gain set at 20 (or less when diodes kick in). So that's what I'm playing with now. When I'm ready I'll post a video. I'm not in much of a hurry and I have other things to do as well so don't hold your breath till I do.

BTW, the cars that produce EMI on my li'l island are real racks, no registration or insurance... for ages, and it happens only at maximum sensitivity - not enough incentive to play with coil shielding at this point. Partly the problem was on my wrong GEB setup, but it gave me a valuable insight on what exactly chatter sounds like on my rig. 10k finetune single turn pot was a nice idea, but I need more than that to cover full span between GEB on ferrite and GEB on red soil or sea water. I'll put a 100p in parallel to the existing one so the 10k pot will cover more span, just like 20k pot would do now.

Otherwise, even without proper GEB I found various small objects at depth ~20cm on a beach in red soil saturated with salt. It works

[ivconic]

Would be good if you now do make another coil, but this time shielded.
Good to compare between that and the one you are having now.
I have made maybe 60 TGSL's so far, maybe 70, not sure.
I have noticed chatters only in case when coil&machine were not matched good but not due any EMI you are talking about here.
TGSL design is pretty immune on EMI if done neat and properly.
Same case with IGSL.
Made only 4 IGSL's so far. The one with original Musketeer coil is rock stable and calm.
No EMI issues at all.
Yet another, with my hand made coil, tends to produce occasional chatters when Sense adjusted to maximum.
Is all about coil.
Video clip i uploaded on Youtube, showing IGSL testing, is made in my workshop; pretty noisy environment.
Next to the table where coil was placed, there is ADSL line; terrible source of noise!
For example; at same table i made video "Deus&chain" and if you watch that video you can hear Deus chattering all the time.
So IGSL is rock stable considering "outdoor" influences.
Next; i checked mods you are suggested, funny i can't get the results you are talking about here!?
On contrary; IGSL turns worse in behavior!?
Would be good if you draw schematic with your exact modifications so others here can check and test too.
Cheers!

[Davor]

Yeah, I get good depth even with wrong GEB I just didn't leave enough margin to account for highly mineralised soil: red soil soaked in sea salt. I couldn't get anything from the sea though - my GEB pot with poor margins didn't reach it. Waving a coil over the sea without a real ground balance produced a very funny series of tones.

Be assured that I noticed chatter already, and I know what you are referring to. I get it too when a car with bad electricity passes by, but I can wipe it off by adjusting Disc sense - in effect I am trading sensitivity for noiseless operation. I guess the original antichatter would wipe off small signals too because it is too slow in recovering. I get chatter, but it is a manageable nuisance.

There are two cures for EMI that I can think of.

Electronics tinkering way around this would be a noise blanker, but it would require either a tuned tank with high Q (impractical) or a delay line to be effective. Once I modded the old FT101 SW transceiver's noise blanker and got miraculous results. Can you imagine - they've put a diode in reverse so the original blanker never worked

The other would require a modified, differential coil, where Rx coil resembles figure 8 of a sort. There is a phasing problem with this approach, but not so much if Tx is running a differential coil too. This will have to be further examined. The idea is that differential coils are much more sensitive to the local signal than the far field, so in effect they are cancelling EMI. Guitar pickups are typically differential coils, so they don't pick up hum, ergo Humbucker pickups.

[simonbaker]

It wouldn't happen if you short the differential inputs, not directly the inputs of the opamp. Anyway, I'm not in a mood of doing any of these at the moment. I don't see how I can benefit from it. Instead, I am doing some field tests, and today was my first day at the beach. Got one almost deserted, but brim full of thrash. Found several pieces of Al foil, Al caps, no pulltabs a piece of lead, some odd piece of iron, and a piece of concrete with iron in it (at first I thought it was a hot rock of a kind).

Anyway, when my IGSL starts chattering, it means that some old car with bad electric circuitry is passing by. Otherwise it behaves nicely. That means a permanent solution for this kind of chatter would be a noise blanker. I'll give it a thought.

I also learned that I need to adjust the GEB gross adjustment because I was missing some quarter of a turn (I guess) at fine tuning for perfectly balancing the sea water. Also the red soil is somewhere at the very edge of my setting, so I really need to fix this. Bobbing the coil says it all - it is that easy. With sea water and red soil you can stick the ferrite up your ... you know, but it is a good reference for a start.

The relevance of EMI for most of us is that it can directly affect our depth tests. Perhaps not so much with your modifications, I don't know.

The problem seems to be that, because of that "chatter filter" we discussed at length, noise is not just "additive" to the target signal; rather, it causes breakup of weak target signals into smaller pulses; the smaller pulses cannot get past the "chatter filter", because the chatter filter has effectively a time gate that blocks short pulses.

So many people think there is something wrong with their circuit or coil when it could be EMI. I tested one of dfbowers TGSL mds in my workshop and get about 15 cm air depth. If I go 10 miles into the mountains, I get over 30 cm air depth.

In your case you get good depth, so probably EMI not a problem; I was just curious.

-SB

[simonbaker]

Hi SB

Do you really mean short the input pins of the opamp? This usually makes the output slam over to one or other of the supply rails because it removes all feedback and the input offset voltage is then amplified by the open loop gain, which is a lot. What has been your experience?

Gwil

I think you have a point; In some tests, I possibly did short the actual pins without thinking, and I don't recall looking at the output of that amp. However, if the amp did swing to one rail, the effect would still probably be to effectively remove all signal from the input of the final LM308, which is AC coupled to the previous stage, so it would serve as some kind of comparison to the full RX-connected case. But I agree it is not really a useful test; a better comparison would be to disconnect the RX coil and short the connector terminals to see what kind of background circuit noise there is without the coil. However, even then the connector and resistors pick up significant EMI in my locale I think, so it is hard to really isolate the pure circuit noise.

-SB

[Davor]

It wouldn't happen if you short the differential inputs, not directly the inputs of the opamp. Anyway, I'm not in a mood of doing any of these at the moment. I don't see how I can benefit from it. Instead, I am doing some field tests, and today was my first day at the beach. Got one almost deserted, but brim full of thrash. Found several pieces of Al foil, Al caps, no pulltabs a piece of lead, some odd piece of iron, and a piece of concrete with iron in it (at first I thought it was a hot rock of a kind).

Anyway, when my IGSL starts chattering, it means that some old car with bad electric circuitry is passing by. Otherwise it behaves nicely. That means a permanent solution for this kind of chatter would be a noise blanker. I'll give it a thought.

I also learned that I need to adjust the GEB gross adjustment because I was missing some quarter of a turn (I guess) at fine tuning for perfectly balancing the sea water. Also the red soil is somewhere at the very edge of my setting, so I really need to fix this. Bobbing the coil says it all - it is that easy. With sea water and red soil you can stick the ferrite up your ... you know, but it is a good reference for a start.

[Gwil]

Hi SB

Do you really mean short the input pins of the opamp? This usually makes the output slam over to one or other of the supply rails because it removes all feedback and the input offset voltage is then amplified by the open loop gain, which is a lot. What has been your experience?

Gwil

[simonbaker]

The "perfect mount" is made using a "Pipe clamp with rubber and nut", a screw and a nut. I'm missing two washers to relieve strain on the box, but they'll come eventually. So I drilled a hole in a device box, put the screw into it it so that it protrudes outside, secured it with a nut, and screwed into a nut of a clamp. I can unscrew it after I detach the coil cable.

Arm cradle and grip are made from a forearm crutch. Got two of them for ~13US$ with explanation that "granny does not need them any more". These come with a plastic arm cradle and grip made in one piece, and are supposed to withstand enormous weights. The rest of it are telescopic metal tubes that I discarded (for the time being). Instead I use a Φ22mm standard (literally) garden variety wooden broomstick (~2US$, new) that I trimmed a bit. Fits perfectly.

Regarding noise ... I'm not equipped with a real lab at the moment, so all my measurements are done in situ with very basic equipment. With IGSL it means that Tx must operate in order to obtain negative voltage. What I observed while balancing a coil was a uniform noise floor at some distance (say 20dB) below the unbalanced minimum which was at below 1mVpp. For these to become meaningful, I'd have to play with spectrometer to make it scan with much narrower window, do some averaging and whatnot, but the point is - I am measuring with a laptop soundcard, and it is noisy by definition. It is fine at 1mV, but not so fine 30dB below.

Pipe clamp is nice! Better than the hose clamp which I've put to use .

Laptop soundcard scope should be fine because typical noise voltage at LM308 output can be 20 to 40 mV with RX coil connected, and maybe 5 to 10 mV with RX coil shorted (actually short the differential inputs to the op amp). And the bandwidth is very narrow (couple hundred Hz max?), so won't exceed the soundcard bandwidth at all.

You could probably even measure with a decent RMS AC voltmeter.

-SB

[Davor]

The "perfect mount" is made using a "Pipe clamp with rubber and nut", a screw and a nut. I'm missing two washers to relieve strain on the box, but they'll come eventually. So I drilled a hole in a device box, put the screw into it it so that it protrudes outside, secured it with a nut, and screwed into a nut of a clamp. I can unscrew it after I detach the coil cable.

Arm cradle and grip are made from a forearm crutch. Got two of them for ~13US$ with explanation that "granny does not need them any more". These come with a plastic arm cradle and grip made in one piece, and are supposed to withstand enormous weights. The rest of it are telescopic metal tubes that I discarded (for the time being). Instead I use a Φ22mm standard (literally) garden variety wooden broomstick (~2US$, new) that I trimmed a bit. Fits perfectly.

Regarding noise ... I'm not equipped with a real lab at the moment, so all my measurements are done in situ with very basic equipment. With IGSL it means that Tx must operate in order to obtain negative voltage. What I observed while balancing a coil was a uniform noise floor at some distance (say 20dB) below the unbalanced minimum which was at below 1mVpp. For these to become meaningful, I'd have to play with spectrometer to make it scan with much narrower window, do some averaging and whatnot, but the point is - I am measuring with a laptop soundcard, and it is noisy by definition. It is fine at 1mV, but not so fine 30dB below.

[simonbaker]

I can post a schematic, but in fact it will be two schematics - a delta against the original so that the existing boards can be utilised or upgraded, and a clean one. Fe channel correction performs so good, and you only need to violate two LM339-s to enjoy it. I don't know - maybe the best way to present everything would be to go through mods and explain what they do.

Regarding EMI, I think this area is quite quiet, but truth is - I have no other rig to compare this one with. I'd say most of the chatter is self-produced, mostly when batteries are running on fumes. This would be a weak spot of this design. I'd say that an oscillator that would run more symmetric would kill most of the chatter. I kinda expected this - it is the PWM problem.

In a moment of epiphany I realised a perfect mount for the device on the shaft. So it became complete. OK, a few washers wouldn't hurt, but a nut and a screw just proved the concept.

I have a silly notion of giving names to all devices at home, and i christened this one "Raskov 2012". Have a look (temporary straps are still on the probe):

Looks interesting.

1. What exactly is the "perfect mount" and what is it made of?

2. Also, how is the arm cradle and grip made?

3. To evaluate EMI effect, I look at LM308 (or whatever chip you use) outputs for background noise signal, and compare the levels with the case where preamp inputs are shorted (preferably as close to chip as possible -- even jumper clips can pick up some stuff). Can you look and give approximate size of noise signals for those two cases?

Regards,

-SB

[Davor]

I can post a schematic, but in fact it will be two schematics - a delta against the original so that the existing boards can be utilised or upgraded, and a clean one. Fe channel correction performs so good, and you only need to violate two LM339-s to enjoy it. I don't know - maybe the best way to present everything would be to go through mods and explain what they do.

Regarding EMI, I think this area is quite quiet, but truth is - I have no other rig to compare this one with. I'd say most of the chatter is self-produced, mostly when batteries are running on fumes. This would be a weak spot of this design. I'd say that an oscillator that would run more symmetric would kill most of the chatter. I kinda expected this - it is the PWM problem.

In a moment of epiphany I realised a perfect mount for the device on the shaft. So it became complete. OK, a few washers wouldn't hurt, but a nut and a screw just proved the concept.

I have a silly notion of giving names to all devices at home, and i christened this one "Raskov 2012". Have a look (temporary straps are still on the probe):

[simonbaker]

Thanks, I'll take a day or two to decide on final touches and then you'll see it all.
Let me just gather some thoughts on differences against the original IGSL.
- I joined GEB of both Cu and Fe channels together, and also ported it to a 10k potentiometer for fine tuning. It is less confusing this way. Now that I think of it - there is a spare channel doing ... exactly the same job as the other GEB channel, and it could work as an "all metal" (GEB) with time constants improved for pinpointing ... just a thought.
- discrimination potentiometers are as they are in original.
- I joined "Disc sens" (in reality low signal threshold) to a single pot. Ikebana mod allows me to do that, now that both channels work by the same principle. Again, this is less confusing this way.
- "threshold" controls are now ported to a stereo potentiometer and provide some volume control. I'll have to put a few kohm in a series because the binary sound sucks.

About the guts...
- frontend is a true differential amplifier with input impedance at 6k8 - I found resistors combination from E6 that do that nicely. It can work happily with regular coils, but I arranged my DD coil Rx as a center tapped bifilar. So far I did not notice any problems with static, and I guess it is fine. On LM833 output it produces 40mVpp so this seem to be OK.
- I joined both GEB controls to one by connecting pins 1 and 13 of the phasing comparators, and disabling the other one by forcing it high.
- LF351 is not used, yet by inertia of populating a board first - decide on mods later, I left there a few components that are not used.
- gain stages were set with both diodes in a second stage, and with feedback capacitors removed from the second stage. I can write about it whole day and more, but the consequence of this setup will be more obvious in a video.
- I removed all unnecessary bits from the comparators section, so there are only 100k resistor, and a 22n capacitor. The diode is replaced with a short. Together with modded gain stages this stands for full duration of beep over a target. Disc sens controls are merged to one.

About the coil...
It is balanced by hand-squeezing it here and there, from outside to the inner overlap, fixing it with a crazy glue to a plexiglass form. At the end only a small portion of it remained unglued + a small curl for finetuning. I noticed that the curl was not of essence, and that most probably the coil would have been nulled even lower without it - it breaks coil symmetry. Now I have less than 1mV of unbalanced signal, and I just can't make it go lower. I'll not put any shielding on.

More to come...

Sounds very good, many useful simplifications can also be applied to TGSL too.

Can you post final schematic?

What is the EMI noise level like in your area? It is very bad where I live, so designs with increased bandwidth may pull more noise. I think the mod with your second amp/filter stage slightly opens up more bandwidth at upper end if I recall.

-SB

[Davor]

It must pass the author's examination and approval first. There are many mods here.