Barracuda Legend Kit Problems

[eclipse]

Hey guys i have my pc back to full health yay! so i can continue with the barracuda build, in the mean time i have made a coil with 0.5mm enamelled wire 14 turns wrapped with 1mm plastic spiral wrap and a mesh sheild taped on with paper tape, OK! so i have tried the coil and the movement issue is 90% gone but the coil is so slow i may as well just dig random holes at the beach with my son i will have better luck! but it is a step in the rite direction .

what wire would you sugest i buy? i have found this?

http://www.ebay.co.uk/itm/4008140503...%3AMEBIDX%3AIT

any good? or junk?

try new coil with 22-24awg teflon stranded wire. Insulate the windings with teflon tape (used by plumbers) as much space as you can afford between the coil and shield. Use low capacity coaxial cable, no more than 100ish pf per meter and make it as short as possible ie 1.2m is optimal. I havent built this device but response is faster when adding a diode in series with the mosfet drain.

[eclipse]

Also dial in a highest frequency for the transmitter, it will give better response.

[kt315]

Also dial in a highest frequency for the transmitter, it will give better response.

highest freq gives NOTHING, this is pulse induction not VLF, you get only the overrun of the capacity of the batteries.

[Qiaozhi]

Can someone please probe pin8 of the 40106 with a scope on channel 1, and then pins 12, 6, 4 and 2 respectively, so that we can see the relationship between the TX oscillator output (pin 8 ) and each of these signals.

I can see that running the 40106 between 0V and -5V will drive the TX stage correctly, but I'm not certain about the main and EFE samples. Clearly, several people have already built this design and claim that it works, so is there a discrepancy between the schematic and the actual board? There's something a but fishy here that needs to be clarified.

[Impulse]

Does your shield have a gap in it so there is not a complete ring or closed loop around the coil? What kind of shielding are you using? As for wire that should be fine. It may not be as fast as more exotic solutions but it should be fast enough since coil speed is primarily driven by interwinding capacitance which is primarily determined by the spacing between each turn of the winding and also by the spacing between each turn and the shield. In your case you might try to layer two layers of the spiral wrap around the coil and apply the shield on the a outside of that being very careful to leave a gap of maybe 8mm between the ends of the shield.
Did you ever try the single shorted loop of wire experiment I described earlier?

Hi Old Cart i have used this for a shield and there is a gap of 20mm i have soldered a piece of stripped 0.5mm enameled wire to the shield and it is connected to the neg of the coax cable. and there isnt much (if any) gap between each turn as the spiral wrap holds the coils tight together so this is probably the main cause of the slow coil. i will order the new cable from china and try another once it arrives with a double layer of spiral wrap and PTFE tape.

http://uk.rs-online.com/web/p/cable-...67206D65736826

[darkfibre]

That shield is a bit over the top, maybe you should do some tests with it removed.

[6666]

[QUOTE=Qiaozhi;214953 Clearly, several people have already built this design and claim that it works, .[/QUOTE]

One that I repaired
https://www.youtube.com/watch?v=sqMtzIPxv9I

21 seconds of my own home made Bara working

https://www.youtube.com/watch?v=7b47WYDqwdw

[Impulse]

One that I repaired
https://www.youtube.com/watch?v=sqMtzIPxv9I

21 seconds of my own home made Bara working

https://www.youtube.com/watch?v=7b47WYDqwdw

NICE! mine works well without the shield as darkfiber said its probably way over the top, im going to get some graphite spray and use some styrofoam strips (as a spacer) and paper tape all the way around then place some stripped enameled wire on the paper tape, then when i spray it the wire will be in direct contact with the graphite spray/shield. Then i hope i will have a coil that will function properly eg with no movement issues or false signals and good depth.

[Old cart]

NICE! mine works well without the shield as darkfiber said its probably way over the top, im going to get some graphite spray and use some styrofoam strips (as a spacer) and paper tape all the way around then place some stripped enameled wire on the paper tape, then when i spray it the wire will be in direct contact with the graphite spray/shield. Then i hope i will have a coil that will function properly eg with no movement issues or false signals and good depth.

Yes that shield is likely too conductive. The general rule is that if the detector can detect the shield held near an unshielded coil then it is no good. The graphite spray is fine but frankly I would not worry about any shielding until we get everything working without the shield.
Can we get the measurements Quiozhi asked for?

[Old cart]

Can someone please probe pin8 of the 40106 with a scope on channel 1, and then pins 12, 6, 4 and 2 respectively, so that we can see the relationship between the TX oscillator output (pin 8 ) and each of these signals.

I can see that running the 40106 between 0V and -5V will drive the TX stage correctly, but I'm not certain about the main and EFE samples. Clearly, several people have already built this design and claim that it works, so is there a discrepancy between the schematic and the actual board? There's something a but fishy here that needs to be clarified.

AGREED! There are several schematics and several boards. The schematic in post 324 appears to be mostly correct but references the input pull ups for gates c,e,and f to Vdd. Vdd on the board is (+5). But in CMOS logic no input should exceed the most positive supply voltage which in the Bara is 0 volts. The outputs I have seen all go 0 to -5volts. However some of these chips generate runt pulses at the transmit pulse transitions which typically go from -5 to maybe -3. I'm guessing that Pulling the inputs to +5 causes a gradual degradation in the chip, maybe damaging the input protection diodes. Once these break the runts start. The runts may not be big enough to actually switch on the sample gates but they should not be there. Not all 40106's degrade in this way presumably due to a more robust input structure. I don't know which brand(s) are more robust.
This would explain why some people's Bara's work and others do not. You would not see the runts if you don't have scope and know how to use it.

Bottom line is I believe the design is marginal.

I certainly don't want to start a war but the original Baracuda was designed by an individual and I don't know, but suspect, he was not a very rigorous designer.

If this does turn out to be the problem a simple mod should fix it. Maybe Silverdog could then update the board.

[Qiaozhi]

AGREED! There are several schematics and several boards. The schematic in post 324 appears to be mostly correct but references the input pull ups for gates c,e,and f to Vdd. Vdd on the board is (+5). But in CMOS logic no input should exceed the most positive supply voltage which in the Bara is 0 volts. The outputs I have seen all go 0 to -5volts. However some of these chips generate runt pulses at the transmit pulse transitions which typically go from -5 to maybe -3. I'm guessing that Pulling the inputs to +5 causes a gradual degradation in the chip, maybe damaging the input protection diodes. Once these break the runts start. The runts may not be big enough to actually switch on the sample gates but they should not be there. Not all 40106's degrade in this way presumably due to a more robust input structure. I don't know which brand(s) are more robust.
This would explain why some people's Bara's work and others do not. You would not see the runts if you don't have scope and know how to use it.

Bottom line is I believe the design is marginal.

I certainly don't want to start a war but the original Baracuda was designed by an individual and I don't know, but suspect, he was not a very rigorous designer.

If this does turn out to be the problem a simple mod should fix it. Maybe Silverdog could then update the board.

You may also have noticed that there's a 100nF cap connected to the gate of the 2N2222 (which drives the MOSFET). I presume the idea behind this is to prevent the MOSFET latching on if the TX oscillator stops working. However, this also means that the TX on-pulse is actually shorter that the applied driving pulse, which results in an extra delay between the time when the drive pulse goes low and the main sample starts. This is why I'd like to see some scope shots of the various timing signals, as I suspect things are not quite as they seem.

[ivconic]

Barracuda looks indeed "marginal" on paper.
But in reality it's better than that.
From the simple PI designs it is number one on my list.
Made dozen and more so far, all those still in use on real terrains.
Occasionally i got excellent reports from users.
So... using the real device is one thing and "farting" too much about it on forums is quite another thing!
It picks up literally all the targets down to 30-40cm depths in soil, with average coil size (~30cm).
600pps, 0.3mH coil, Al foil Faraday cage, matched fet pair.... usual things to carry for.
Adding multi turn potentiometer for Delay ... and it can even be adjusted to reject some unwanted stuff.
Overall it is splendid little design enough simple and cheap.
Only possible drawback is weak audio in speaker. That's why earphones are better choice.
I haven't followed this topic (building such detectors is like riding a bike, you made first one... than you made them all the same) and i am not sure exactly by what pcb draw you are making it;
but i can repeat here again 100% checked and proven pcb which i used (as i recall drawn by our nice friend ApBerg (which should be respected and printed on each pcb copy done&sold by others here)):

[raygdunn]

Here is a good example. Impulse has .55 volts on pin 1 of the 40106. This is an actual measured voltage, not a simulation. This clearly exceeds the absolute maximum input voltage rating of TI. This 40106 also is not working correctly. It introduces additional pulses that should not be there. They are runt pulses ( smaller than normal) the occur on the rising and falling edges of the transmit pulse. These may have no effect on the Baracuda operation, but they should not be there!

One simple solution would be to cut the trace that leads to the +5 volt regulator very near the 33k resistor attached to pin 1 of the 40106. Cut the trace between the 33k and the regulator NOT between the 33k and pin 1. This will allow R17,18, and 19 to float. Then tie this point to pin 14 of the 40106 with a thin jumper wire. This hold place pins 1 ,3 and 5 at 0 volts which is still a logic high and will not exceed the chip specs. However it will likely alter the timing of the two timing pulse and require resistor value adjustments.

if someone can look this over it advise if I have this right that would be great.

Hi Old cart,

I've moved the timing resistors supply from V+ to Earth. (Removing the chip before I started :-) )
* Initial sample pulse start point looks unchanged
* Initial sample pulse width increased from about ~40us to 100us (KT schema reads 45us)
* Delay start of first to start of second sample now ~480us.
* Second sample width was about 40us now ~130us
* This mod has no effect on the rounding of the leading edge of the first sample pulse. I thought it might. This separate oddity arises when the Delay is moved to an early position and starts somewhere close to optimal. This is on a standard 10" coil. Strangely it appears to cope better with the very short delay on a spider wound coil.

Pin readings from 40106 are very different:
1: -4.60V
3: -4.53V
5: -3.40V
And without further adjustment I still get 8" range in a GB half penny, (which I thought was a 1p in a recent post, so quite small).

Retiming this, I'd guess would be by soldering a couple of leads across a timing resistor (150/33/33K) for easy access, add a 100K pot (plus possibly another fixed value) and adjust until the scope shows original width. Then calculate the combined value of the two parallel resistors.

Trace showing wider primary sample pulse
Trace showing extended delay from start of primary to secondary sample pulse

Ray

[kt315]

not mine post.

General spec measured by me
-transmiter pulse Fcy= 640 PPS
-Transmitter pulse width= 100 uS (microseconds)
-pulse delay =begin at 22 uS to 60 uS .
-Position knob Delay 1= 22 uS, position 4 =40 uS, Position 9 = 60uS
-sample pulse window 45 uS
Current consumption - 100Ma at 10,5V

[Old cart]

Mods part two

Hi Old cart,

I've moved the timing resistors supply from V+ to Earth. (Removing the chip before I started :-) )
* Initial sample pulse start point looks unchanged
* Initial sample pulse width increased from about ~40us to 100us (KT schema reads 45us)
* Delay start of first to start of second sample now ~480us.
* Second sample width was about 40us now ~130us
* This mod has no effect on the rounding of the leading edge of the first sample pulse. I thought it might. This separate oddity arises when the Delay is moved to an early position and starts somewhere close to optimal. This is on a standard 10" coil. Strangely it appears to cope better with the very short delay on a spider wound coil.

Pin readings from 40106 are very different:
1: -4.60V
3: -4.53V
5: -3.40V
And without further adjustment I still get 8" range in a GB half penny, (which I thought was a 1p in a recent post, so quite small).

Retiming this, I'd guess would be by soldering a couple of leads across a timing resistor (150/33/33K) for easy access, add a 100K pot (plus possibly another fixed value) and adjust until the scope shows original width. Then calculate the combined value of the two parallel resistors.

Trace showing wider primary sample pulse [ATTACH]35557[/ATTACH]
Trace showing extended delay from start of primary to secondary sample pulse[ATTACH]35558[/ATTACH]

Ray

Well done!
i would not expect any performance improvement only that the detector would be more reliable becuase the IC's are run within spec.
if you want to continue the experiment parallel a 100K pot with each of the resistors below adjust for the timing stated. If you know how to calculate the parallel combination do it and replace the each resistor with that calculated value. Please report the results to us.

RESISTOR. ADJUST. MEASURE PIN ON 40106
R19. 45uS. 2
R18 45uS. 4 should closely match width on pin 2 value should be similiar
R17. 150 uS between the two pulses above. 2,4- you may have to use a larger value pot than 100 K.

That should do it. The delay should work as it normally does and should move both samples the same amount and not alter their values. Since you are measuring I would adjust the delay trimmer on the board to set the minimum deal ( with delay pot set to 0 ) to the delay we discussed earlier with the fastest coil you can build. Then use the separate pot to set the delay to reject small foil etc if you are search for big stuff. You always want to use the smallest delay possible when searching for low conductivity items like gold.

This mod also might make the delays more stable. Most people do not think about this but delay jitter causes noise. You might retry the bypass caps we tried earlier if they have been removed. Mainly the ones between pin 7 and 14. The reason this may help now is that it will remove noise on the minus supply. It did so before BUT the delay was referenced to and affected by the noise on the plus 5 supply, which we did not bypass.

cheers!