Photo of GP Extreme Badness?

[IBGold]

Hi Barry,

I don't have any left do you have an electroplater in your vicinity you could take him a new socket and get him to gold plate it that is all Woody does my supply has dried up now as they wont supply in 10 off lots any more minimum $1000.00+ order the manufacturer is Salecom see if there is a distributor in the US that is prepared to supply you one.

Regards, Ian.

[bklein]

I found a new plug at a local electronics store. I wired that up, replaced the bad parts, installed the pins 1-5 and 3-4 jumper, fired it up and...
First the BC857 shorted. Taking that out fried the left fet.
Why did the BC857 fry?
By jumping pins 3-4 and 1-5 i seem to have shorted out one of the diodes at the edge of the board - or it shorted by itself.
I'm really in a mess now. Not sure what to do next. Need to check the redone coil wiring, the jumpers on the receptacle, and take the bad parts out to get where I was originally.
I'm out of the fets - I have 4 parts that are higher Vds but lower current and higher on resistance. (I wasn't sure which one to buy so I got both)
I'm getting frustrated now.... I've always HATED FETS.

[bklein]

I checked the coil wiring and it was correct. I removed the BC857 and left fet and the behavior became much better - like earlier when I took measurements. It actually detects without these parts. I put my last parts in. Turned it on and it kept working. Put the guts in the box and powered it up - still working. Put the box and coil on the shaft and poof, quit working. Just as before.
Perhaps a short in the feed line? It was wrapped pretty tightly around the shaft originally. The feed line is not too flexible - it wants to retain the wrap shape from before. I don't have any other coils to try... suppose I could make one. The new connector fixed the intermittent issue.

The BC87's I used were 200V 400ma parts. I used em up. I also bought some that are 240V but 260ma and higher on resistance. What could possibly be blowing these but not the TX/RX fets?

[Mechanic]

Ahhh so they have some diode protection between pins 3&4 and I suspect pins 1&5 too. Probably on the DD input as well as this would make sense. Are you sure there is not a bad solder joint somewhere between the coil ground wire and actual ground?

Might be an idea to leave the fets out, solder wires to the gate and the drain, analog ground and signal ground and why not the tx wire as well. Remove the Cancel mono DD switch from the front panel and secure it to the board so it can't flap about and short out on something, hot glue , and then run the wires through the hole. Assemble the detector and see what is seeing what. Make sure that all of the voltages are constant. If you have an old analog scope laying us that as it will show up intermittent faults better than the digital. Wriggle the coil lead around etc.

The coil leads can break, within an inch or 2 from the plug and also down where it passes through the cable gland on the coil housing. I would have thought though that the protection systems in place would still have stopped the fets from seeing voltages higher than 200v. The only thing I can think of is if the coil ground was not connecting properly in the detector. Try pulling on the ground wire(in the detector) to make sure it has not broken inside the insulation and has an intermittent connection. Could explain why it only happens when you put it back together. Could also be the tx wire or an intermittent connection, bad diode between the wire and snubber circuit.

The detector will still work without the left fet as that is the mono receive fet, the right one is the DD fet. However if you try and run cancel it will not work because it needs to receive from both the Tx and Rx coils and then subtracts the responses before further processing. Also without the left fet when you try and run a mono coil the detector will not work(and probably blow the right fet). When it is in mono mode it receives from both preamps and sums the response before further processing.
Cheers Mick

[bklein]

When you ask about the coil ground wire and actual ground, you mean between the coil receptacle and the pcb, perhaps the large black wire? Or do you suggest that the blue wire that has a ground lug for a chassis screw is making flakey connection to the chassis? That would be possible I guess as it is rather haphazard the way it is just supposed to contact the
EMI aluminum shield when the screws are tightened down.
Last nite when the parts failed the first time, the pcb's were not in the box and the coil was not on the shaft. So thinking back there was a moment I think where I may have powered it without the coil attached (not sure, it was 2AM). Is that likely to kill it right there? If not, then would my new change of pins 1-5 together and 3-4 together at the box receptacle remove the diode protection and add this short risk? I'm just trying to come up with whether this is a box problem or a coil/cable problem.
Also of note of that experience - the BC857 died - not the left fet. The left fet died only after I removed the BC87 and repowered the unit. So is my BC857 under-rated? What can happen to the coil/wiring that would make this guy die? (Hmmm, maybe intermittent 100 ohm resistor? It did get hot and I didn't replace it.)
When I first saw this coil and cable, I was struck as to how the feed line has taken on the wrap form of the shaft even when off the shaft (hard to describe, easier to photograph). I'm used to the feed lines being very flexible - but this cable material is unique. It has a polycarbonate or whatever hard plastic insulation on each of the 2 coax's. I can imagine this getting so hard it cracks over time - but didn't see that at the plug side. I did notice that the ground shield wires were a bit brittle - as a few broke off when I twisted them together to wet with solder. I did cut off an inch or two to be sure it was beyond where the major stress was on it before.
As this is detecting metal in the short periods of workingness - does this mean the right fet channel is good?
I'm now suspecting the 100 ohm resistor - I'm going to swap it out just for kicks.
I don't have any more BC857's. Would most any PNP (2n3906) work here? More on order.

[bklein]

I put my ohmmeter between pins on the coil cable and see that the resistance changes from 19.8 ohms or so to 17.9 or so when I manipulate the cable about 10 inches from the coil. There is something hard in the cable near this spot - what is it?
Anyway, didn't see clear shorts or opens while doing this.

[Mechanic]

Hi Bklein,

Thats an odd observation on the coil lead, just checked mine and I can't feel anything. The fact that the resistance changes when you fiddle with the lead says that there could be a problem with it. Were you measuring between pins (1,5) and pin 2? (ground to DD rx out?)
If there is a problem in the coil lead I guess there is a chance it could damage something. Turning the detector on without a coil will cause no harm, however I would not expect the front end fets to turn on, but on second thoughts they must because you can turn on a detector without the coil and pop your finger in the plug and get the detector to sound off.

If possible try and borrow a ML DD coil from someone, if not it might be worth investing in one. Here in Australia they are $225, should be cheaper for you guys though, ML must get paid to send them over!!!! At this stage avoid a mono coil as it might take out the DD fet if there is a problem and it is not rectified first. As far as making one goes, they have very tight tolerances for their specs and coil cable(coax that is) is practically impossible to get hold of.

Have you checked the fet gate voltage with the fet installed? Might look ok when it is not under load, but perhaps there is not enough current to drive the gate properly with it installed, however I think this would be pretty unlikely.

Cheers Mick

[bklein]

Is your cable pretty pre-coiled and rather fixed that way or is it totally flexible?
After thoroughly working the cable up and down I have minimized the intermitancy of the resistance dropping. I don't get it as the two coax's are separated from each other physically. Darn - I recall a several months ago there was someone selling bad minelab coils on ebay for like $20. Would be interesting to take the cable apart to see if they have something magic in there like ferrites or something and we never expected it.
I really don't think anything is blown in the two channels - if it is I don't have much confidence I would be able to find it at this point. I put new parts in - the BC857C and one of my alternate BSS87's that are 240V parts. Outside the case it comes up and runs good. I flip the cancel switch and the wildness calms down and it detects metal things I pass over the coil. Of course there is metal and emissions all over my workbench so not a field test by any means. But I think it likely is fixed at this point. It is when it is put back in the case that must be the killer. I don't like the way the wires are just haphazardly stuffed back inside - all around switcher circuitry generating the 184V. I ran out of time last nite/morning 2am and will first cycle power a bunch of times to be sure that works. Then leave it on and manipulate wires all over the place looking for some intermittant. Then perhaps cable tie wires so that they form a consistant bussing that can be inserted in the case in a way that makes sense. Then just insert one side first and try it. If good, take that out and do the other side. That good, put the other one in and close it up. I also need to look around the edges of the board to see if something is too close that might contact the case.
I don't like the way these panel mount screws have coarse threads. They don't screw down easy - and create a bunch of metal flakes in the process.
Good point that I haven't taken scope photos in the working state. I should do that.

[mario]

Hi bklein,

I'm not sure I'm following you.

Your flyback is around 184v, but when you use the 200v fet, it blows?

And now that you have the detector working, it works fine with a fet rated at 240v?

Can you confirm the flyback for us? Also, if you look at the TX fets (usually the part numbers don't get much paint on them because they're on the side), you could easily verify as to what the flyback may be. For instance, if those TX fets are rated at 240v, then the RX fet might be rated at a higher voltage than 200v, which would explain why the ones you installed in the first place were frying.

[bklein]

No, sorry for the misunderstanding.
Digikey has two BSS87's to choose from:
240V 260ma (higher on res.) or
200V 400ma
I bought samples of both. Started with the 200V 400ma's. I burned em all up.
Now I am starting with the 240V 260ma's. Just the left fet was bad this last time so that's all I replaced along with the BC857C. It works on the bench, out of the case.
But this is normal circumstance. Typically now once I put everything together it will die on the 1st power up. But I'm going to go at it in minute steps this time to try and see what is related to it. But perhaps it works with no issue - either because of higher V part, I resoldered the 100 resistor, perhaps a flakey solder connection before, or manipulating the coil cable cleared the short... Sure would be nice to find the smoking gun.

[mario]

No, sorry for the misunderstanding.
Digikey has two BSS87's to choose from:
240V 260ma (higher on res.) or
200V 400ma
I bought samples of both. Started with the 200V 400ma's. I burned em all up.
Now I am starting with the 240V 260ma's. Just the left fet was bad this last time so that's all I replaced along with the BC857C. It works on the bench, out of the case.
But this is normal circumstance. Typically now once I put everything together it will die on the 1st power up. But I'm going to go at it in minute steps this time to try and see what is related to it. But perhaps it works with no issue - either because of higher V part, I resoldered the 100 resistor, perhaps a flakey solder connection before, or manipulating the coil cable cleared the short... Sure would be nice to find the smoking gun.

1) So your flyback is 184v, 208v?

2)You do realize that the blue wire that is hooked up to the case with a screw that goes through a Fet doesn't actually make contact with the Fet. Otherwise you get a nasty short.

[bklein]

1) So your flyback is 184v, 208v?

2)You do realize that the blue wire that is hooked up to the case with a screw that goes through a Fet doesn't actually make contact with the Fet. Otherwise you get a nasty short.

Hi Mario,
The photo on message 28 shows pulses going to 208V but they do so because of my lousy ground reference (the blue wire you mention, and the EMI shield). So really long ground wire = false inductive peaks. Hey, a PI within a PI!
The pulses are actually clamped as shown in the photo in message 38.


Please elaborate on #2. When I run the unit out of the case I typically tie that wire to the EMI shield. This wire is tied to the TX black wire at the coil receptacle. Are you saying that wire's lug should not contact the case and shield? What then?

[mario]

Hi Mario,
The photo on message 28 shows pulses going to 208V but they do so because of my lousy ground reference (the blue wire you mention, and the EMI shield). So really long ground wire = false inductive peaks. Hey, a PI within a PI!
The pulses are actually clamped as shown in the photo in message 38.


Please elaborate on #2. When I run the unit out of the case I typically tie that wire to the EMI shield. This wire is tied to the TX black wire at the coil receptacle. Are you saying that wire's lug should not contact the case and shield? What then?

It's hard to explain. Also, you haven't posted a photo of your board, so I don't know if it applies.

However, on the SD models, a screw connects the blue wire to the case ( because it is used as case ground). It goes across the tab of a big fet (the one in the snubber circuit) but doesn't actually make contact with the tab because a small plastic insulator between the screw and the tab prevents the two from making contact. Also, a thin square, rubber pad exists that prevents the back of the fet from making contact with the case.

The reason the screw goes across the fet is two fold: 1) the case needs to be grounded and 2) the big fet gets hot (it's rated at 400V), so a tight screw presses the fet against the case for better heat dissipation.

You said your machine came broken, maybe those two parts got lost along the way (the little plastic insulator and square rubber pad)?

If either of those two pieces are missing, you can make things better by replacing the fet with one in a TO-220F package. Those fets are encased in a plastic shield, so the tabs and back are not exposed.

[bklein]

Hmmm... I didn't find any loose parts other than this wire and its lug when I opened it up - but it was returned after Minelab had started working on it and maybe they kept the washer and nut. I assumed the lug attached between the case and front plate shield - I didn't actually see that the lug was sandwiched between them. (I thought it was kind of hokey)
I don't recall there being a fet with a tab that faces the case though. I think all standup fets are on the other (center) side of the board. You can see the solder pads of them I think in my first photo - near the discolored parts.
Looks like I will need some confirmation from someone who has a GP as to where this wire goes.

[IBGold]

Bklein the GP series do not have the earth lead fixed as the SD's do the lead is fixed directly to the case with a screw it is on the bottom of the case towards the front when it is standing up you should have had a screw and washer have a look at the case between where the boards slide in from memory about an inch in from the front, the case not being earthed properly would not have helped your cause.

Regards, Ian.