TGSL Tuning & troubleshooting

[fekemil]

Hello!
I built my first TGSL board, it's my second homemade metal detector. The first one was the famous and easy "matchless".
I used TGSL-EDU-small layout, exactly those parts that were written on the drawing. It wasn't easy to buy LM3038's, but I had luck. I measured all parts before installation. I used low ESR capacitors and 1% metallic resistors.


I could not reject ferrit, no matter what I'm doing with the potentiometers. Metal detecting working, 1€ coin 15 cm. I have douple tone.


Coils: TX 6,05 mH, RX 6,55 mH, both shielded with aluminium tape.
TX frequency 13,79 kHz, as it should be. But RX only 13,61 kHz, from my calculations it should be 16,06 kHz. I dont get it. This value is only good, if I have 6,55 mH coil and 20,88 nF capacitor. But I measured the capacitor before soldering and it was 15 nF... Need I change this capacitor?
Negative voltage source is -6,33 V. Nulling not perfect, 4 mV AC on U101a pin 7.
I could borrow an oscilloscope, but I never used before I have a MM with frequency, inductance and capacitance measures.
Any idea?

[ripsdevala]

Alu tape is bad, i use mylar blanket or graphite+glue as shielding
2 tone= reversed rx cable
Cant reject ferite= not ballanced properly
My last best air deep is 25cm for brass coin.

[Hyena]

Hello!
I built my first TGSL board, it's my second homemade metal detector. The first one was the famous and easy "matchless".
I used TGSL-EDU-small layout, exactly those parts that were written on the drawing. It wasn't easy to buy LM3038's, but I had luck. I measured all parts before installation. I used low ESR capacitors and 1% metallic resistors.


I could not reject ferrit, no matter what I'm doing with the potentiometers. Metal detecting working, 1€ coin 15 cm. I have douple tone.


Coils: TX 6,05 mH, RX 6,55 mH, both shielded with aluminium tape.
TX frequency 13,79 kHz, as it should be. But RX only 13,61 kHz, from my calculations it should be 16,06 kHz. I dont get it. This value is only good, if I have 6,55 mH coil and 20,88 nF capacitor. But I measured the capacitor before soldering and it was 15 nF... Need I change this capacitor?
Negative voltage source is -6,33 V. Nulling not perfect, 4 mV AC on U101a pin 7.
I could borrow an oscilloscope, but I never used before I have a MM with frequency, inductance and capacitance measures.
Any idea?

4mV on pin 7 is your problem. Ivconic and others measure voltage directly on unplugged rx cable with target value 5-15mV. But if you measure it after first op amp as you did - get it to 0.6V - 0.8V. You have too deep null now. That also caused reject ferrite problem - try to increase residual voltage slowly and check rejection position and air depth. Double tone is caused Rx swaped cables. Alu foil is ok. With 0.8V on U101 pin 7 I get 30cm for 1€...

[fekemil]

Thank you for your answers!
Swapped cable is strange, because I spend lots of time with the schematic. I tought RX pin1 is the coil start connector, which goes to U101a(LF353) pin5. Because that is marking with +. My coil is under epoxy now, so I could not change the overlap. Is my bad RX frequency is also came from the too deep null? I think I have to make a new DD head.
Update.
I soldered RX cable with reverse polarity, now I have only 1 tone when detecting. And the circuit is now extra sensitive for electrical smog. Are you using plastic, aluminium, or iron box? My "matchless" circuit only works well when in an iron box. Small tea box actually.

[Hyena]

Thank you for your answers!
Swapped cable is strange, because I spend lots of time with the schematic. I tought RX pin1 is the coil start connector, which goes to U101a(LF353) pin5. Because that is marking with +. My coil is under epoxy now, so I could not change the overlap. Is my bad RX frequency is also came from the too deep null? I think I have to make a new DD head.
Update.
I soldered RX cable with reverse polarity, now I have only 1 tone when detecting. And the circuit is now extra sensitive for electrical smog. Are you using plastic, aluminium, or iron box? My "matchless" circuit only works well when in an iron box. Small tea box actually.

1) I do not care about Tx and Rx start or end, just plug it in and if it produces two tones, swap Rx cable, thats all...


2) Yes, when I get my PCB out of box or just open top it gets nervous due to EMI. I use black chinese aluminium boxes. Just grind off paint on contact surfaces to make it completely conductive and ground it to GND on PCB. Via screws it grounded shaft and everything what is connected to it.


3) Now the hardest part - the null Inductances are close enough, there is only problem of size of overlap now. Even when you pot it with resin you can fine tune it with small piece of ferrite. Get some SMALL piece of ferrite. Place ferrite somewhere near cross of loops (most sensitive place so smaler ferrite is neccessary) and check residual voltage as before. Every time you move ferrite you need to check voltage, set GEB and check DISC!!! Now you can play with it - best results I got with 800mV. Ferrite rejection is in the middle of GEB potentiometer and detection depth is (for 27cm coil) 30cm in air with 1€. I put 1€ 25cm in white sand (I got it everywhere around my house) and I can even raise coil few CM over ground to get clear signal.


There is one more thing - underlap and overlap. With underlap you get 200° phase shift and detector is - I do not know why - more sensitive to ferrous targets. With overlap you have 20° phase shift and detector is more sensitive to precious metals. This you can measure only with 2 channel oscilloscope or devices you can find here on geotech or on internet. Like phase shift measurement or so.

[fekemil]

I think you're right about these cables, less thinking, more doing My 1 euro detection depth increased with about 5 centimeter just from the cable swap.


I have an unused external HDD case made of aluminium, I'm gonna use it as my TGSL box. Iron boxes are less elegant and have more weight.


What kind of ferrite is good for the fine tuning? Ferrite bead or magnetic? I have a hammer, so I can make small pieces. (Pink Floyd - One Of These Days I'm Going To Cut You Into Little Pieces)


But I think it's easier to make new search head, I have two coils already, I just have to hardening and shielding them. Perhaps I will try graphite shield. And vacuum forming too The first search head made from plastic pot saucer.
I will take more time for the nulling procedure. I left the college of electrical engineering before I learned to handle the oscilloscope. But never too late.


Poor english, I know.

[Hyena]

Nulling again

I am not sure what material ferrite beads are made of, maybe the same as those rods? I am using rods from very old radio antenas. About 8-10mm diameter, usually 10cm long, enough material for many coils and pinpointers...


As I said before... use that coil you already made. You do not need oscilloscope, use your DMM and watch residual voltage on pin 7. It does not matter if you have 20° or 200° pahse shift much. It will work both ways (e.g. every concentric coil has 200° and work well). Play with voltage and you will see where you will get best air depth and how it affect GEB...

After you handle this and confirm, that it work well on field (with GEB and DISC), make proper coil with fancy housing...

I still have my first plastic pot saucer coil in cabinet. Looks very funny, but works as well as any other vacuum formed after. And when I want to measure something, I am using this, because I do not need to disassembly anything from already completed detectors...

[fekemil]

I tried use a bit of ferrite moving on the head, but the voltage only increased with few mA, best result was 14 mA. Without ferrite, it was 4 mA on U101a pin 7. I think my problem is not just the deep nulling. I could not reject ferrite. Can you measure your RX frequency? Mine is 13,62 kHz, and very-very stable. I tried change the parallel 15 nf capacitor to 10 nF, but frequency not changed. Even without any capacitor!
In the tgsl base document we can read this: "C6 in parallel with the Rx coil tunes our Rx circuit to about 16.12 kHz (off resonance). This is required to get the proper phase relationships for disc and ground balance to work properly."
I checked every component, every soldering point with magnifying glass The funny thing is, the circuit working more or less, but no matter what I'm doing with the potentiometers.

[fekemil]

I've done some measurements according to the documentation:


1. TX freq. 14,5kHz measured 13,84kHz
2. 4024 /32 freq. 453Hz measured 425Hz
3. negative power -5V measured -6,3V
4. RX freq. 16,12kHz measured 13,62 kHz
7. Lm308 PP-noise 5mV measured 0,18V
8. sensitivity pot 0-0,036V measured 0-0,038V
11. regulated power 8V measured 7,99V
13. audio -5-+V measured -6,3-0V

[fekemil]

I've connected my new unshielded coils to the circuit. The results are the same. TX frequency 14,07 kHz, RX 13,83 kHz. The RX resonant circuit not working somehow. The LF353 has +8V on pin8 and -6,3V on pin4, and now 800mV on pin7. Strange.

[Hyena]

Well, lets take it slowly...


You wrote you got 4 or 14 mA? You measured current? Measure AC voltage of pin 7 against ground.


Frequency, which you can measure while tgsl is running, of Rx is the same as Tx, it just it pick up. But actual resonant frequency of Rx with 15nF cap is 16.XXX kHz - it is frequency, which will have highest voltage if you will connect it to freq generator and try different frequencies.


I tried higher capacitances than 15nF, it improves air depth a bit, but has huge drawbacks, so I stick to range 15nF up to 15.5nF. With 16nF it behaves still somehow well, only GEB moves from center to right slightly. Discrimination then becomes less stiff. Over 17nF or 18nF you will be almost unable to GEB and Disc will shift also. It actually shifts so much, that I was almost unable to discriminate pulltab...


To your measurement:
1. TX freq. 14,5kHz measured 13,84kHz
ok
2. 4024 /32 freq. 453Hz measured 425Hz
ok
3. negative power -5V measured -6,3V
ok
4. RX freq. 16,12kHz measured 13,62 kHz
ok
7. Lm308 PP-noise 5mV measured 0,18V
Need to look and measure it later
8. sensitivity pot 0-0,036V measured 0-0,038V
ok
11. regulated power 8V measured 7,99V
ok
13. audio -5-+V measured -6,3-0V
ok


Well... it looks well for me.


Are you sure it does not work as it should?

EDIT: Check that residual voltage on PIN7. I will do some oscillo-screenshots of behavior of this pin wit ferrite rod and DMM so you we can compare our units somehow.

[fekemil]

Well, I'm trying to understand you!
It was mV off course, not mA.
I understand, I think. Changing capacitor C6 makes phase shift. With the right value I can use my GB pot.
I can measure 30-40 mV AC on pin7, according to the position of coils. 800 mV was a measuring failure.
I made pictures.

[fekemil]

I made little steps forward. I measured my TX frequency, it was 13,86 kHz. Frankly, I don't know how can this value 14,5 kHz in the TGSL doc. The resonant circuit with 22 nF capacitor and 6 mH coil must be 13,85 kHz. Whatever. The difference between the TX and RX frequency in document is 16,12-14,5=1,62 kHz. My RX frequency must be 13,86+1,62=15,48 kHz. My new RX coil is 6,6 mH, so I need 16,02 nF capacitor. I soldered 1 nF capacitor paralell with the already built-in 15 nF.
The new coils are not fixed, I can change their coupling. I trying to learn, how can I nulling them.
But ferrite rejection and discrimintion not working well. Sensing distance is very bad. Sad. Many people built this detector without an oscilloscope...


Good news is, I made a box from an external HDD enclosure. It looks promising from the outside, but really messy from the inside. I have a volume control and an earphone/speaker switch.

[fekemil]

I spent a couple of hours with my good friend and his oscilloscope. We measured the important things on the circuit, it seems that working well. Then we set the overlap of the coils. And voilá! Now everything works just the way it should be. Ferrite rejection, discrimination OK. Sensing distance is 35 cm with 1€ in the air.


I'm not saying it's impossible to set the TGSL right without an oscilloscope, but I did not know. And as I watched the oscilloscope chart, it must be very lucky for someone to perfectly adjust the overlap without the device.
With a cheap multimeter you can not measure exact voltage at such a high frequency. And debugging will require an oscilloscope.
This is really not a beginner project, it's not enough to make the circuit and the coils well.




Thanks to everyone who helped, especially to Hyena!

[homefire]

LOL, Acually the TGSL is a quite Simple VLF detector to build. There is nothing in the line of Alignment to do other then the coil. Nothing is adjustable without changing component values.