Houston...I have a problem! (bandido coil)

[Max]

NOW IT WORKS !

Hi all,
I think I've solved the problem in a cheapy way. My bandido seems works now.

What I have from tests in disc mode using a near-to-extrem setting of the disc pot:

- 1 eur coin at about 30cm
- 2 eur coin at about 32cm
- 1 half crown (diameter 32mm) at more than 35cm
- 1 small gold ring (1.5grams) at 24-25cm
- 1 copper cilinder thick 0.5mm, diameter 46.5mm, height 12mm at 42cm
- 1 silver coin (diameter 30mm) at 22cm

That seems pretty good... and like what I expect from Bandido.

All tests where made with fast object movements relative to the coil, tring to simulate
1m/s swings.

It sound also with iron (and rarely with steel) sometimes...but at few cms and with big masses (tools, hammer...).
So it's not perfect discrimination in my unit.

Question: have bandido total disc vs IRON ? I don't know.
Maybe it depends only of capacitors used.

Now the solution, I've used:

- eliminate as much noise as possible
Noise is really subtle.
I've done using semi-encapsulation of some part of the circuit board using a thin aluminum
sheet (insulated with plastic foils) shaped to cover upside of components/pcb. On the copper-side
I've used another PCB (single side copper downside), not etched, connected to circuit pcb using
screws and plastic spacers to give 4mm of distance from circuit copper-side.
Both metallic shields connected with a wire to a unique GND PCB trace.

This cutted out more than 95% of 50Hz noise...and now I have two symm signals at inputs of
comparators (IC8a-IC8b).
This way I get output nearly flat, at a small negative voltage, that rise when an imbalance
occourrs thus giving the audio signal to show-up.

I've noticed that bandido circuit is very sensitive to external noise also if one use (like I did)
fiberglass pcb...using solvents to remove flux after soldering etc...and that's expecially true for
the disc signal path starting at c20/c21.

Only thing to solve and avoid big troubles is Faraday shielding sensitive parts.

So, other good solutions can be using Gary's pcb with a huge upper GND layer and RF guards or use the germany
group pcb or another that have a good, wide gnd screen.

I think that Gary's PCB is very good for testing but, if one can do, to have the best he must add a second
upper layer of copper (double side) to give it a good GND plane and maybe also built in "jumpers".
Unfortunately, I can't do double or multiple side PCBs at this time... so I'll continue with metallic shields,
at least for now.

Other easy solution is insulate everything in a metallic box connected to GND - if a removable cover
present is easy enough to open and connect probes etc for testing.

IVCONIC, TRY SHIELDING AND LET ME KNOW.

Thanks again to all.
Best regards,
Max

[chemelec]

Hi all,
I think I've solved the problem in a cheapy way. My bandido seems works now.

What I have from tests in disc mode using a near-to-extrem setting of the disc pot:

- 1 eur coin at about 30cm
- 2 eur coin at about 32cm
- 1 half crown (diameter 32mm) at more than 35cm
- 1 small gold ring (1.5grams) at 24-25cm
- 1 copper cilinder thick 0.5mm, diameter 46.5mm, height 12mm at 42cm
- 1 silver coin (diameter 30mm) at 22cm

That seems pretty good... and like what I expect from Bandido.

All tests where made with fast object movements relative to the coil, tring to simulate
1m/s swings.

It sound also with iron (and rarely with steel) sometimes...but at few cms and with big masses (tools, hammer...).
So it's not perfect discrimination in my unit.

Question: have bandido total disc vs IRON ? I don't know.
Maybe it depends only of capacitors used.

Now the solution, I've used:

- eliminate as much noise as possible
Noise is really subtle.
I've done using semi-encapsulation of some part of the circuit board using a thin aluminum
sheet (insulated with plastic foils) shaped to cover upside of components/pcb. On the copper-side
I've used another PCB (single side copper downside), not etched, connected to circuit pcb using
screws and plastic spacers to give 4mm of distance from circuit copper-side.
Both metallic shields connected with a wire to a unique GND PCB trace.

This cutted out more than 95% of 50Hz noise...and now I have two symm signals at inputs of
comparators (IC8a-IC8b).
This way I get output nearly flat, at a small negative voltage, that rise when an imbalance
occourrs thus giving the audio signal to show-up.

I've noticed that bandido circuit is very sensitive to external noise also if one use (like I did)
fiberglass pcb...using solvents to remove flux after soldering etc...and that's expecially true for
the disc signal path starting at c20/c21.

Only thing to solve and avoid big troubles is Faraday shielding sensitive parts.

So, other good solutions can be using Gary's pcb with a huge upper GND layer and RF guards or use the germany
group pcb or another that have a good, wide gnd screen.

I think that Gary's PCB is very good for testing but, if one can do, to have the best he must add a second
upper layer of copper (double side) to give it a good GND plane and maybe also built in "jumpers".
Unfortunately, I can't do double or multiple side PCBs at this time... so I'll continue with metallic shields,
at least for now.

Other easy solution is insulate everything in a metallic box connected to GND - if a removable cover
present is easy enough to open and connect probes etc for testing.

IVCONIC, TRY SHIELDING AND LET ME KNOW.

Thanks again to all.
Best regards,
Max

I Totally AGREE with the Flux Removal.
Contrary to Manufacturers Claims, Flux is a Conductor in Sensitive Circuits.
(My PI Detector Also requires the Flux be Removed, Especially around Q4.)
Personally I use a Kester 331 Solder, "Its Flux is removed with JUST WARM WATER".

I'm using my Split PCB and I don't seem to have the Noise problems. (Maybe the Two PCB's Back to Back is helping this.)
I can Design a Double sided PCB and even etch them, but the Plate through holes is a PROBLEM.

As to External Pickup by the detector, A Metal Box should Prevent that. I also don't seem to have that Problem.
(Do you have a lot of Florescent Lighting?)

You MIGHT also Put a Few .001 uf Disc Type caps Across the Supply Lines at Various Points on the board to Supress Any RF in the circuit.

**Ferrite Beads Could also be Placed on the Wire Ends of Some Resistors, Where you think RF is a Problem.

Just some Possible Suggestions, Hope this Helps.

Take care.....Gary

[Max]

I Totally AGREE with the Flux Removal.
Contrary to Manufacturers Claims, Flux is a Conductor in Sensitive Circuits.
(My PI Detector Also requires the Flux be Removed, Especially around Q4.)
Personally I use a Kester 331 Solder, "Its Flux is removed with JUST WARM WATER".

I'm using my Split PCB and I don't seem to have the Noise problems. (Maybe the Two PCB's Back to Back is helping this.)
I can Design a Double sided PCB and even etch them, but the Plate through holes is a PROBLEM.

As to External Pickup by the detector, A Metal Box should Prevent that. I also don't seem to have that Problem.
(Do you have a lot of Florescent Lighting?)

You MIGHT also Put a Few .001 uf Disc Type caps Across the Supply Lines at Various Points on the board to Supress Any RF in the circuit.

**Ferrite Beads Could also be Placed on the Wire Ends of Some Resistors, Where you think RF is a Problem.

Just some Possible Suggestions, Hope this Helps.

Take care.....Gary

Hi Gary,
thanks for the hints.

"Contrary to Manufacturers Claims, Flux is a Conductor in Sensitive Circuits."
Totally agree with you. I always remove any flux in this kind of circuits. I use (outdoor) solvents based on acyclic hydrocarbons. I have in spray and also in small bottles...using some cotton with. They are little dangerous, if one knows what to do and precautions. Never used kester331: I'll try it if I find here.

"I'm using my Split PCB and I don't seem to have the Noise problems. (Maybe the Two PCB's Back to Back is helping this.)"
Don't know. Maybe I have too electrical noise around. But no fluorescent lamps in all the house. I don't like them. Too UV for me (eyes stress)...sadly...workplace has a lot of them...

"I can Design a Double sided PCB and even etch them, but the Plate through holes is a PROBLEM."
Same problem. Can't do this way. I can impress photo-sensitive films both side...but then I've to manually connect layers...(tried only one time...waste of time..."butcher's work").

"You MIGHT also Put a Few .001 uf Disc Type caps Across the Supply Lines at Various Points on the board to Supress Any RF in the circuit."
"**Ferrite Beads Could also be Placed on the Wire Ends of Some Resistors, Where you think RF is a Problem."
Tested a lot of these...no way...I have 50Hz noise. No way.
Tested too inductance filters on power lines...no way.
Tested too hi-value caps (up to 50000uF) to prevent noise from supply ripple...no way.
Tested with battery, no way, it's external noise.
After trying a lot of things...I've realized that I can't suppress totally EM noise at 50Hz without proper shielding.

Maybe is just my situation...workshop...workbench...(too noise here?) don't know.
Never had such problems before. Think that huge amplification make things go worse if too noise present.
LM393 has 200V/mv voltage gain typ (open loop)...so saturation will occurrs here at about 45microvolts differential...
on single comparator and I had much more from noise before installing the shield.

So, I suggest anyone making bandido to look at noise levels before giving up and playing with hammer if something
goes wrong.

Best regards,
Max

[Qiaozhi]

It sound also with iron (and rarely with steel) sometimes...but at few cms and with big masses (tools, hammer...).
So it's not perfect discrimination in my unit.

Question: have bandido total disc vs IRON ? I don't know.
Maybe it depends only of capacitors used.

With my original Bandido (not umax) I could detect the following:

Victorian Penny at 30cm
Liberty Dollar at 27cm
Dime at 23cm

These readings were taken with the detector set to maximum sensitivity, and were taken at a point where the signal was (just) consistent. So these are max depth readings in air. Sorry - no Euros were available for testing.

The Bandido totally discriminates against iron.

[Geo]

Qiaozhi can you tell us the diameter of every coin ???
Victorian Penny = ????mm
Liberty Dollar =
Dime =

[Qiaozhi]

Qiaozhi can you tell us the diameter of every coin ???
Victorian Penny = ????mm
Liberty Dollar =
Dime =

Victorian Penny = 30mm
Liberty Dollar = 24mm
Dime = 18mm

[chemelec]

MAX, That 50 Hz sounds like a Line Frequency Being Picked up "by the COIL".

Gary

[ivconic]

Like i thought...


Thanks people!!!
First thing i am gonna do is to make pcb and rest.....Later i'll post my experiences...
Best regards!!!!

[Geo]

Thanks people!!!
First thing i am gonna do is to make pcb and rest.....Later i'll post my experiences...
Best regards!!!!

Hi Ivconic . Why to make the bandido. It is about 90% the same with CL3. It is a little better at discrimination and cl3 is a little better at sensitivity.
Total weight .....the same
I say this because you are professional at CL3 construct.
Regards

[Max]

Hi Ivconic . Why to make the bandido. It is about 90% the same with CL3. It is a little better at discrimination and cl3 is a little better at sensitivity.
Total weight .....the same
I say this because you are professional at CL3 construct.
Regards

Hi Geo,
I know CL3...used a bought one sometimes in past...not so good with disc...I think that bandido is better (when right tuned in disc mode). Performances seems to me even better (in air) than CL3.

Anyway, we do these circuits also for curiosity...I think.
I have GS4-proto and goes well...very deep...but doesn't really disc...just ignore ground mineralization. It sounds with bricks...so I decided to make a try with (unknown for me) bandido II, as Reg suggested.
I do not expect miracles here...from vlf, but circuit is really good from an engineering point of view except maybe the voltage converter.

Best regards,
Max

[Max]

MAX, That 50 Hz sounds like a Line Frequency Being Picked up "by the COIL".

Gary

Hi Gary,
yes I was thinking too that coils picks up too much noise...but I disconnected RX...then measured noise voltages at inputs of IC8 and get same noise levels (before shielding circuit). So I realized that small noise variations were in the circuit , starting at c20/c21.
To prove that I've used my fingertip:
touching c20 I can see increase in noise in its channel ; touching c21 the same for the other channel.

Concluded that small noise asymm creates huge differential at IC8 inputs.

Then shielded and solved the problem.

Maybe coils gets too big noise...I think so. But it's nulled very well...then if the noise source is far away noise is nulled too.

Best regards,
Max

[Max]

Victorian Penny = 30mm
Liberty Dollar = 24mm
Dime = 18mm

Hi Qiaozhi,
I've tested with half crown of 1967 (queen Elizabeth II) and gives distance of
more than 35cm.

I can test too with other coins...thanks for the infos.
I have to solve too the slight iron detection. I think caps are involved.
I'm using Carl's vaues at now with tollerances of less than 2%.

Do you think I have to use Gary's values ?

Best regards,
Max

[Qiaozhi]

Hi Qiaozhi,
I've tested with half crown of 1967 (queen Elizabeth II) and gives distance of
more than 35cm.

I can test too with other coins...thanks for the infos.
I have to solve too the slight iron detection. I think caps are involved.
I'm using Carl's vaues at now with tollerances of less than 2%.

Do you think I have to use Gary's values ?

Best regards,
Max

Which caps do you think are incorrect?

[chemelec]

Which caps do you think are incorrect?

Those Values were given to me by "Carl Moreland" in an email to me.
He said he measured them Out of Circuit.

On the Other hand "Goodrat" emailed me and Said he measured a Tesordo Detector.
C10, 13, 14 and 18 were all 5pF.
C16 and C29 were 10nF.

So take your choice.

[Max]

Those Values were given to me by "Carl Moreland" in an email to me.
He said he measured them Out of Circuit.

On the Other hand "Goodrat" emailed me and Said he measured a Tesordo Detector.
C10, 13, 14 and 18 were all 5pF.
C16 and C29 were 10nF.

So take your choice.

Hi Gary,
many thanks for the interest.
"Those Values were given to me by "Carl Moreland" in an email to me."
do you mean the values in your revised schematic(s) ? Why they are different from original Carl's schematic ? Maybe he can't measure them well first time due to circuit assembly or something ?

Goodrat values are different too. There are big differences !
Why ?

Is the original pcb using SMT ?

Gary, I've found another little mistake on my circuit:
when disc pot, or gnd pot or both are close to zero ohms (extrem) and big load is on tx signal from oscillator I have a peak-doubling...means that too load on oscillator cause signal distortion visible also after analogue gates.
I think I can solve increasing TX osc. current modifing , so reducing, R1 from 1.2K to 1K or something lower.

What do you think about?

Best regards,
Max