Baracuda Build (silverdog kit Rev2)

[raygdunn]

I'm posting my coil info, appreciate any comments as I don't know if it is correct for the bara...

Wire = 0.5mm 24AWG (enamelled coated copper)
Coil diameter = 200mm (8") with a resistance of ~1.9ohms and 530µH (0.53mH)
Detection distance for small coins in not that great (I assume) at about 4 to 6 inches...
Larger objects such as small aluminium heatsink (1.5"x1.5x"0.5") and a small metal wrench 6" ~ 8"...

Mike

Hi Mike,
I was hoping someone with more experience would reply to you. Recent thoughts on PI coils favour 300-350µH. I think the 450uH on the Barracuda schema is that of the original. A number of people here are aiming to include very small gold nuggets and need small coils. From what I've read here, 10-11" seems to be a good compromise for coins, with the increased depth that goes with size, against increasing loss of detection of small objects.

In air, the best I've got for my Barracuda on a 10" spider type coil ( 23 turns, 333uH, 85pF, 10ohm!, 0.2mm enamelled wire) is 25cm on (the now disused) UK half penny. That's headphones and the slightest hint of something, so perhaps exaggerated a little. This with the coin Blue Tacked to thread suspended from end of metre ruler☺ I think that I read that you were doing something similar.

Have you found Qiaozhi's (improved) Coil calculator? which helps to get the number of turns right.

I've resorted to a pencil and paper to count the turns, but even then it is a frustrating task.

Hope that helps.

Ray

[raygdunn]

Additional thoughts.. Perhaps it is really intended for a 450uH coil, as current seems high for a standard low resistance coil. I've had in mind shortening the 100us coil pulse width, admittedly using a battery approaching 12V. As it is (10ohm), it is using about 150mA and things are not too hot.

Ray

[Michaelo]

@Ray...
My MOSFET is cool (warm but not hot) even with only 1.9ohms and from memory the pulse is almost exactly 100µS... I half expected it to be hotter...
Your coil resistance of 10ohms seems hight but then again your using 0.2mm enamelled wire so I guess the resistance will increase...
Think I drop a few turns to get the inductance down a bit closer to 450µH and see if I get any improvements...

BTW, did you have to change any resistor or cap values when you moved them from +5 on them to Ground (assume you did move them)...
I ask because I also changed from +5 to Ground but haven't checked timing since...

Mike

[jladre]

Its better stay on coil inductance in range of 300-380 uH. Spider wound coil or flat spiral is best.

[raygdunn]

@Ray...
My MOSFET is cool (warm but not hot) even with only 1.9ohms and from memory the pulse is almost exactly 100µS... I half expected it to be hotter...
Your coil resistance of 10ohms seems hight but then again your using 0.2mm enamelled wire so I guess the resistance will increase...
Think I drop a few turns to get the inductance down a bit closer to 450µH and see if I get any improvements...

Hi Mike
I used the 0.2mm wire because my next size up 0.63mm was large for the slots in the plywood former that I'd wound on a couple of times before. Quick fix job☺

BTW, did you have to change any resistor or cap values when you moved them from +5 on them to Ground (assume you did move them)...
I ask because I also changed from +5 to Ground but haven't checked timing since...

Mike

Yes, I had to reduce the resistors substantially. My -5V supply is a little low, so my values may be little low I suspect.
Try something like 13K for the two 33K ones and 56.6K for the 150K. For some strange reason I had to down the 13K for the second sample pulse to 11K to match them. The target width of the pulses is 45us. The gap is around 144us.(This being the gap reading on mine before the change. Possibly it should really be 140 or 150, but I don't see it being critical).

Ray

[Michaelo]

@Ray, that 144us you mention... is that referenced somewhere?

One other thing, have you ever scoped the 5 volt supply rails (+/-)...
The transmit pulse or process, introduces substantial noise on the both rails...
I'd be tempted to filter that out before doing any more changes...

Mike

[raygdunn]

@Ray, that 144us you mention... is that referenced somewhere?

I had no basis for the 144 so backpeddled my upgrade and used my original value.

One other thing, have you ever scoped the 5 volt supply rails (+/-)...
The transmit pulse or process, introduces substantial noise on the both rails...
I'd be tempted to filter that out before doing any more changes...

Mike

Hi Mike,
I checked out the supply rails at one point and failed to detect any problem. However I have real problems like you describe with the 40106. The output to the second sample pin 2, have three spikes on it. This was thought to be associated with the out of range resistor bias being discussed above. However now all my 40106s have this problem (5 or 6). Some since fixing the resistor bias. I have to assume my static measures are not good enough. I've dug out a grounded mat to add to the wrist band. (Peeved.. I ordered more and received the wrong parts. Still the unit is still finding targets my PI-1seems to have missed.)

Ray

[kt315]

My -5V supply is a little low
--
is a little hight. you can not get low than -5V from a source with -5V voltage out.

[Qiaozhi]

I checked out the supply rails at one point and failed to detect any problem. However I have real problems like you describe with the 40106. The output to the second sample pin 2, have three spikes on it. This was thought to be associated with the out of range resistor bias being discussed above. However now all my 40106s have this problem (5 or 6). Some since fixing the resistor bias. I have to assume my static measures are not good enough. I've dug out a grounded mat to add to the wrist band. (Peeved.. I ordered more and received the wrong parts. Still the unit is still finding targets my PI-1seems to have missed.)

Ray

Have a look at my post here -> http://www.geotech1.com/forums/showt...486#post215486
I suspect you've knackered all your CD40106 ICs.

[Michaelo]

@ray, I started posting but then read Qiaozhi post so I've edited this and removed comments till I read the rest of the post...
Have to say, things are beginning to clear up now...

At least we had the +5 to Ground issue fixed... Have to admit, the caps arrangement around the two regulators and 7660 had me confused, something didn't look right but again I assumed...

So it looks like C17, C20 and C22 are wrong...
C17 should go from pin 5 of U3 to Ground (positive to Ground)... C20 & C22 negatives to Ground...

Sometimes you can't see the trees for the woods

Mike

[raygdunn]

At least we had the +5 to Ground issue fixed... Have to admit, the caps arrangement around the two regulators and 7660 had me confused, something didn't look right but again I assumed...

So it looks like C17, C20 and C22 are wrong...
C17 should go from pin 5 of U3 to Ground (positive to Ground)... C20 & C22 negatives to Ground...

You must be using the schema marked 'Drawn by kt315', as there is no C22 on the board. Some useful additional components there, but there is a closer schema to the Silverdog Rev 2 board here. Though I keep finding more differences.

I've rewired the two 220uF at top right of the board (component view). I'm thinking we now really want another 220uF on the 78L05 output pin, if I can figure out a solid way to squeeze it on. It's a bit large to just hang it off the back of the board.

I suspect you have a proper oscilloscope and can check out lower noise levels than my USB based BitScope.

Just replaced the 7660 for the second time only to find it is still finger burning hot at 12V. So I guess the problem will be with the 79L05. Hopefully they will be happier following the capacitor changes.☺ Things do seem less stressed at 9V. Perhaps I should thing about buying some more AA rechargeables.

Ray

[Qiaozhi]

You must be using the schema marked 'Drawn by kt315', as there is no C22 on the board. Some useful additional components there, but there is a closer schema to the Silverdog Rev 2 board here. Though I keep finding more differences.

That's actually the one I've been looking at, but 6666 says the one with the orange writing should not be used as it's unfinished.

I've rewired the two 220uF at top right of the board (component view). I'm thinking we now really want another 220uF on the 78L05 output pin, if I can figure out a solid way to squeeze it on. It's a bit large to just hang it off the back of the board.

Yes - there's a capacitor missing.

I suspect you have a proper oscilloscope and can check out lower noise levels than my USB based BitScope.

Cannot do any further measurements just yet. Replacements CD40106 ICs are on the way.

Just replaced the 7660 for the second time only to find it is still finger burning hot at 12V. So I guess the problem will be with the 79L05. Hopefully they will be happier following the capacitor changes.☺ Things do seem less stressed at 9V. Perhaps I should thing about buying some more AA rechargeables.

Ray

The 7660 has a 220uF between two of its pins, but the datasheet specifies 10uF. It might be worth changing that as well.
The 7660 on Impulse's board seems to be ok temperature-wise.

[Michaelo]

Baracuda Rev2 (silverdog)

Because all schematics I have found are not correct for Rev 2 (silverdog), I have done a quick schematic containing fixes from Qiaozhi's post (link below)...

It's not an exact match as part numbering is different (components cover the values and using any other existing image is a non runner)...
All the component values should be correct but if anyone spots errors, please post and I will fix them...

Notes:
Ignore component numbering for now...
Different schematics appear to have the amplifier balance circuit incorrect (according the datasheet), the one presented here is correct...
There are a few decoupling capacitors added as they are required for perfect operation...
The circuit includes the additional pot (10K) for delay, mounted off the board...

The layout is not exactly like the others but over time I will try to get them close...


Qiaozhi Fixes: http://www.geotech1.com/forums/showt...486#post215486

[Qiaozhi]

It's not an exact match as part numbering is different (components cover the values and using any other existing image is a non runner)...
All the component values should be correct but if anyone spots errors, please post and I will fix them...

Here's an image of the blank PCB from the Silverdog website. It's a bit flurry, but the component values are still quite readable.

[Michaelo]

Juts doing a preliminary check and it has a TIP127 but we use the IRF9640... makes you wonder if there are more differences...
Think that's an older version as the board I have has the correct MOSFET marked on it...

Will do some googling and see if I can find some more board, you never know...