Barracuda Legend Kit Problems

[Old cart]

For all tests where we are looking at or comparing timing to the sample pulses let's try triggering on the NEGATIVE falling edge of the positive transmit pulse.

[Old cart]

For a good discussion how PIs work see here:

http://www.geotech1.com/pages/metdet/projects/hammerhead/HHv1p5.pdf

it is also good to buy a copy in Inside the Metal Detector described at the top of the forum. It is a great book!

[Old cart]

Hi Old cart,
Abbreviated post as managed to lose the original...

I installed a replacement 40106 and amazingly all the unwanted spikes disappeared, and the current went down from 110mA to 90mA!
Usual pins before: [ATTACH]35458[/ATTACH]

(A garden coil movement test seemed to have improved, only noticable on change of direction of swing.)

Sadly when I reconnected to grab an after image the spikes started again.. but much lower for now.
[ATTACH]35459[/ATTACH]
Gut feeling is that it gradually degrades with use. I am noting a spark like sound from headset on start up. I think that hasn't been there for long. Used to be a tone. I'm wondering whether there could be a voltage spike at start up.

Back to plan A and add those capacitors I guess.

Good range on a small GB 1p coin. so worth persisting I guess. :-)

Ray

Ray good progress. I don't think this problem is anything other than a instrumentation problem. Low cost scopes are great for hobbiest use but you can not expect them to be perfect. They can easily be very noisy, can have poor grounding, and even inject noise back into the circuit. Indeed if you are trying to make critical analog measurements like this I would not even consider a scope that cost less 5-10,000USD.

That having been said few hobbiests have such a scope nor would I recommend that they get one unless they are very serious about electronics.

So so we have to use the scopes we have as sort of a compass to help us debug and set up the detector.

Do you happen to have a US nickel? That would be a good target for testing purposes. Lacking that we can go with a 1cm square section cut from an ordinary Coke can.

One last point, I did notice your transmitter pulse changed width. This is probably due to the fact the 40106 is a digital IC just to square up and invert a signal. They were never intended to be used as precision timing devices.

[Old cart]

Hey guys I have had a chance to look at the old board and I decided to swap it for the new board to see what result I got and I had no sensitivity to movement I had the delay pot at 10k and the delay trim at 10k, I have tried a depth test and I basically have to hit the coil with my gold ring for anything to happen so i adjusted pin6 to 0v and the sensitivity was 5 inches with the same ring but I have the same issue where I cannot move it ? Shame I had a glimpse of wait I think it might actually freeking work!' Bit no lol

Have you secured the windings yet? If you have not then it is likely it will beep with movement. The reason for this was described in one of my earlier posts but basically relative coil winding movement generates a similiar signal to a target. So don't move it! Once the coil is secured and shielded then we will deal with it if it beeps. The reason it does not beep, nor does it detect the ring well when the pots are set to a high resistance is that the delay is very long and it takes a lot of metal to make a signal change down the waveform.

To illustrate this simply hook the scope to pin 6 of the 5534 and adjust for a good trace. No metal should be anywhere near the coil at this point. Now bring a large metal object like a pair of pliers near the coil and observe the change. Repeat with the ring. Tell me what you see.
then add a second trace 40106 pin two and repeat. The detector main ( target) sample is where the detector samples the signal. The more the trace moves vertically where the sample gate ( pulse ) is the better it will be detected.

You may note the the pin6 negative going pulse gets wider when a piece of iron is brought near the coil. This can cause a large vertical movement in the signal and a strong detector response.

An optional exercise is to adjust the delay and see the sample pulse move relative to the transmit pulse ( always trigger on the falling edge of the transmit pulse near the top of the positive going pulse). Take a shot where it disappears and the just reappears.

This is experiment should better help you understand how a PI detector works.

[raygdunn]

Very good. I like to reference all bypass caps to ground. There are advantages in using caps to connect +to- but it intrudes common mode noise.

Hi Old cart,
Abbreviated post as managed to lose the original...

I installed a replacement 40106 and amazingly all the unwanted spikes disappeared, and the current went down from 110mA to 90mA!
Usual pins before:

(A garden coil movement test seemed to have improved, only noticable on change of direction of swing.)

Sadly when I reconnected to grab an after image the spikes started again.. but much lower for now.

Gut feeling is that it gradually degrades with use. I am noting a spark like sound from headset on start up. I think that hasn't been there for long. Used to be a tone. I'm wondering whether there could be a voltage spike at start up.

Back to plan A and add those capacitors I guess.

Good range on a small GB 1p coin. so worth persisting I guess. :-)

Ray

[Impulse]

Hey guys I have had a chance to look at the old board and I decided to swap it for the new board to see what result I got and I had no sensitivity to movement I had the delay pot at 10k and the delay trim at 10k, I have tried a depth test and I basically have to hit the coil with my gold ring for anything to happen so i adjusted pin6 to 0v and the sensitivity was 5 inches with the same ring but I have the same issue where I cannot move it ? Shame I had a glimpse of 'wait I think it might actually freeking work!' Bit no lol

[Old cart]

Thanks Old cart,

I do hate getting things wrong... I was expecting C23 to be like the rest at 220uF. I see now that the capacitor I thought was C23 was actually correctly connected C20! So C22 and C23 missing.


I'll try tacking on those two capacitors. Then maybe look at the longish power track to the logic chip. It will be tomorrow before I can try it now.

Ray

Very good. I like to reference all bypass caps to ground. There are advantages in using caps to connect +to- but it intrudes common mode noise.

[raygdunn]

c23 should be 22 uF and looks to be correct in the schematic. It should not go to b+. C20 and c22 should be there but may not be on Silverdogs PCB layout. I don't like them going to -5v But somebody did it this way for some reason, so you try this connection first. Then if you want you could try connecting the negative side to ground and comparing the results. Sadly your scope does not have low enough noise to make a proper measurement so you wil have to test subjectively. If you have a good multimeter with true RMS you could try to use it in AC volts, which will reject the DC component and just measure the ripple. By good I mean one that costs more than $100US?
Do NOT use AC+DC volts range if your meter has it. You should note that the multimeter, unless it is VERY good wil not make an accurate measurement but it wil make a relative one.

To do the caps you can just tack solder them to the back of the board,using the shortest lead length possible. Be careful not to short traces. I would be interested to hear how this goes.

BTW bypassing is as much art as science. Sometimes it does not help at all. One such case would be if the supply is not really the source of the noise. There are ways to isolate noise but the require a lot more equipment and knowledge than most of us have.

Thanks Old cart,

I do hate getting things wrong... I was expecting C23 to be like the rest at 220uF. I see now that the capacitor I thought was C23 was actually correctly connected C20! So C22 and C23 missing.


I'll try tacking on those two capacitors. Then maybe look at the longish power track to the logic chip. It will be tomorrow before I can try it now.



Ray

[Old cart]

No my meter doesn't read capacitance, i have tried it with the cap both ways and the board stops working (no sound from the speaker). I was using a laptop until it became a brick.

Sounds like your cap is bad. If it is on the right pins it should not break the circuit. It may not help though.
It is NOT a good idea to put a cap in backward- they can explode and if they are large quite violently. Wear eye protect when first firing up a circuit where you are not sure.

[Old cart]

I've spent some time looking at this today. With the USB style BitScope, I'm just getting complete mush at 50mV per grid unit. Examining the 5V supply outputs at 1V per grid unit I get the attached. Both monitors compare the NE5534 pin 6 output pulse with the power line. I switched to a high power battery for the following but couldn't see a difference, but bags of supply power from a 1300mah Lipo at a little under 12V.

The 78L05 has just a hint of somthing towards the end of the pulse.
[ATTACH]35434[/ATTACH]
The 79L05 (supplying the logic circuit) has what I'd take to be a dip and recovery when the coil pulse ends. My estimate was 0.1V recovery from the grid.
[ATTACH]35433[/ATTACH]

I compared the latest schema with the top of PCB image showing component placement around the power supply area. (And marveled at how people can produce a schema from a populated PCB!)

C23 on schema is 220uF (not 22uF), which is good, but instead of connecting to ground it connects to B+. Does one lose anthing doing that?

Suply stabilising capacitors C20 and C22 are not part of the PCB design. Just possibly this was aimed at reducing bulky components or an error in translation? Not sure C20 would be needed, as C23 does something similar.red
[ATTACH]35432[/ATTACH]

I could try connecting C23 to ground instead it has a long separate copper track that can be cut and a wire added to the back of the board.


Ray

c23 should be 22 uF and looks to be correct in the schematic. It should not go to b+. C20 and c22 should be there but may not be on Silverdogs PCB layout. I don't like them going to -5v But somebody did it this way for some reason, so you try this connection first. Then if you want you could try connecting the negative side to ground and comparing the results. Sadly your scope does not have low enough noise to make a proper measurement so you wil have to test subjectively. If you have a good multimeter with true RMS you could try to use it in AC volts, which will reject the DC component and just measure the ripple. By good I mean one that costs more than $100US?
Do NOT use AC+DC volts range if your meter has it. You should note that the multimeter, unless it is VERY good wil not make an accurate measurement but it wil make a relative one.

To do the caps you can just tack solder them to the back of the board,using the shortest lead length possible. Be careful not to short traces. I would be interested to hear how this goes.

BTW bypassing is as much art as science. Sometimes it does not help at all. One such case would be if the supply is not really the source of the noise. There are ways to isolate noise but the require a lot more equipment and knowledge than most of us have.

[Old cart]

Ok so I have tried a shielded mono coil today and I still get the annoying beep beep beep at the end of the swing. The coil is 24 wraps 0.5 with an alloy tape wrap and connected to the shield of the coax.

What kind of alloy wrap are you using? Generally you have to use very thin lead foil which is hard to get, a graphite based compound or some other low conductivity material. Search coil shielding on this forum for lots of good suggestions.

Regardless of what you choose do NOT allow it to make a complete loop. It you do it forms a "short circuit" for the magnetic filed and the detector will not work properly.

To illustrate this point take a test clip lead or just a piece of wire and form a 20cm single turn loop with it. Leave the ends disconnected. Bring it near the coil and a touch the ends together temporarily. Even time you touch the detector will go off. You will be surprised how for you can detect this target.

[Impulse]

Ok so I have tried a shielded mono coil today and I still get the annoying beep beep beep at the end of the swing. The coil is 24 wraps 0.5 with an alloy tape wrap and connected to the shield of the coax.

[raygdunn]

Noise is is vertically polarized so the coil must be held flat for lowest noise. If the windings can move relative to one another falsing will occur. However do not pot them in epoxy, it will increase coil capacitance. You can use tape to bind the windings or expanding foam. Many commercial detectors have coil shielding. Search this forum for the various techniques.

I think any testing that involves coil movement is not very useful until all other issues are resolved and the coil is stabilized and possibly even shielded. Shielding may not be necessary at long sample delays, if the coil is small, or is to be used underwater.

i think the problems with spurious noise on the sampling lines might be affecting the overall detector noise and is due to power supply noise. Check each of the supply pins (+/- where applicable) on each chip relative to the negative side of the battery. Particularly when doing noise tests it is very important to locate the ground lead carefully ( the - side of the input protection diode is a good place) and to turn off all external sources of interference. Disconnect any PC peripherals and run the PC off batteries where practical. While checking the supply pins also check all the ground pins on each chip. Use AC coupling on the scope if it is available and the use the max sensitivity possible. Set the bandwidth limit to the lowest setting possible if available. Show me any pictures of any line that has more than 50mV of noise on them and identify them by chip name and PIN number. This is probably pushing the limits of what most USB scopes are capable of.

I've spent some time looking at this today. With the USB style BitScope, I'm just getting complete mush at 50mV per grid unit. Examining the 5V supply outputs at 1V per grid unit I get the attached. Both monitors compare the NE5534 pin 6 output pulse with the power line. I switched to a high power battery for the following but couldn't see a difference, but bags of supply power from a 1300mah Lipo at a little under 12V.

The 78L05 has just a hint of somthing towards the end of the pulse.

The 79L05 (supplying the logic circuit) has what I'd take to be a dip and recovery when the coil pulse ends. My estimate was 0.1V recovery from the grid.


I compared the latest schema with the top of PCB image showing component placement around the power supply area. (And marveled at how people can produce a schema from a populated PCB!)

C23 on schema is 220uF (not 22uF), which is good, but instead of connecting to ground it connects to B+. Does one lose anthing doing that?

Suply stabilising capacitors C20 and C22 are not part of the PCB design. Just possibly this was aimed at reducing bulky components or an error in translation? Not sure C20 would be needed, as C23 does something similar.


I could try connecting C23 to ground instead it has a long separate copper track that can be cut and a wire added to the back of the board.


Ray

[Impulse]

Very good with the carbon fiber. Is there any metal attachment hardware within a foot or two of the coil? What kind of PC are you using? A laptop or desktop? Does it have any other hardware attached to it other than a mouse or keyboard? If it is laptop can you run it on batteries?
The reason I am asking all these questions is trying to figure out how noise is getting into you system. The cap should have stopped the detector from working unless it is bad or on the wrong pins of the 40106. You want the minus on pin 7. Does your meter read capacitance?

No my meter doesn't read capacitance, i have tried it with the cap both ways and the board stops working (no sound from the speaker). I was using a laptop until it became a brick.

[Old cart]

Hi Old cart,

Thanks for your input. I've just been retrying your test idea from some posts back.
Stabilising the 40106 power supply seems a great idea to stop interaction of the pulses, by adding a '104' capacitor across the power pins. I loosely hooked the wire ends over the pins, with a bit of spring tension rather than solder. That way I could monitor the effect, when I pulled the capacitor off. Surprisingly I could see none.


The crossfeed seems all over that chip. On pin 13 next to the Delay pot. I could even see a copy of the coil pulse. How that got there is puzzling.

The power supply lines looked clean, when monitored.

It seems common to the two boards, but maybe it will work differently for Impulse.

I'll see if changing the chips (40106, NE5534) has any effect, when they arrive. Probably not, but worth a try as I ordered some.

Ray

If you look at the schematic you will see that pin13 is connected to same signal that drives the 2222 coil drive transistor. This is bad design practice but the Barracuda was designed to be simple so that concession was necessary. Other detectors use separate but synchronized sources to drive the sampling gates. See the schematic for the surf PI ( or pro) for example.

Did you look at the power supply lines using the techniques I described? Also look at the ground for each chip, that is at least equally important.