GPZ 7000 Tear-down

[jasong]

Also, thinking back to electronics classes, if there are that many unshielded and clearly noisy components on board then Woody's loop technique may not be picking up all the areas of potential self inflicted EMI.

Because the traces on the board itself may be acting as antennas and injecting noise into the circuit as well. Another reason it's best practice to shield and prevent self induced EMI, it might show up in places you can't really see by just probing with a loop in open air.

For instance he probes near the RX module and thinks the inductor EMI drops off. But if there is a trace in the board closer to the inductor, which also leads to the RX module, then the RX module could see higher amounts of injected noise from the trace than he sees with the loop.

[jasong]

His latest video he seems to shield all the self generated noise out of the 7000. Goes on to say someone sent him a 6000 to work on (he surmises as I did earlier, that it suffers from the same problem), and might try an Axiom. Then mentions some rumor about a new upcoming detector release being basically the 7000 put into a 6000 case.

I'd post the vid, but man these things are hard to watch and I don't want to post a link to a bunch of camera fidgeting and dead time. I really, really wish he'd spend a bit doing some basic editing. Because he does seem to have some interesting things to say and show.

Also drives me crazy he does all this work on noise reduction, then demonstrates it in a noisy room in a house with some copper sheet over the coil, using a 9V battery as a target (bad target selection for a number of reasons), with his belt and microphone moving right next to the coil too.

I find it pretty interesting the detector still detects a target at full depth through that copper fabric shielding though. That's basically a faraday cage. It must be detecting some kind of steady state EM properties, anything transient should be getting blocked, or I guess just fields directly coupled rather than radiated at least. But it does beg the question - can we do better with shielding on coils now too? That'd be the next thing I'd look at, almost curious enough to go put some aluminum tape on the top of one of my coils this summer whenever the snow melts off.

[phrunt]

If you put shielding tape on top of a GB2 coil the detector goes completely numb, can't detect a thing so it's interesting the GPZ can detect right through it.

If they put a 7000 into a 6000 case I don't want one, to me that's a downgrade as you can bet they'll change the coils to chip inside coil.  It needs to be significantly better than what I've got or I'm certainly giving it a miss, and I think that will be a challenge for them.  I don't care about ergonomics, weight, Bluetooth, whatever cosmetic things they can do to improve the 7000, I want real performance improvements over the 7000 with a range of X-coils and lots of them or it's a waste of money.

It's a wait and see, but if the 6000 is any indication of what to expect, I don't want one.  Performance differences from what I can see on anything are minimal over the 7000 if at all.

If Woody can improve shielding on the 7000 perhaps I'll take that path and send her off to be modded instead of an 8000.

[jasong]

If it was nicer out I'd go do some experiments with aluminum tape. Still winter here though. The copper fabric oddity does bring up one more potential improvement - can we shield the bottoms of coils too for more EMI prevention? I wonder if there is much EMI coming up off the ground as reflection, ground plane transmissions, etc or no?

If the GPZ is using some kinda static fields, it gives me another idea - curious what would happen if you put some kind of steel on top of the coil.  Maybe a piece of thin steel sheet, maybe just some steel nuts/washers? Would there be a depth increase?

Reason I'm curious is because ferrous metals "enhance" or magnify static magnetic fields by concentrating flux, and higher flux density means a stronger signal. It would be interesting to experiment over the RX lobes, TX lobe, and both at once.

Would be easier to try on the 6000. I'd laugh hard if a quick hack to getting a little bit more depth ended up being taping a big steel washer onto a mono coil. 😄 

[phrunt]

A bit like what some people do with their Carrots, this guy has a pretty good way on his Carrot with an antenna.

 

[jasong]

Interesting, gonna have to try that too on my carrot, first I'm seeing that. Not quite sure what's happening there, but I think it's different than the steel-on-a-coil experiment I want to try, as those antennas are probably non-ferrous (and the pinpointer is a VLF).

Guessing there is some component inside the carrot about at the point the tip of the antenna extends out to, and it's coupling to the antenna and changing some component values by proxy on the board?

Hmm, some stuff to think about though.

[jasong]

Tried some experiments with the 6000, got too curious. Putting steel on the coil definitely doesn't help, putting that right out front. 😅 

But, it was informative, Geosense seems to be a bit too smart for it's own good sometimes. Was doing some air tests on a US quarter with various chunks of steel on the coil. It'd work, but clearly it was adjusting ground balance and reducing sensitivity as a result. I could get the 6000 to be absolute dead quiet in the middle of my noisy shop by putting a bolt on the coil, because it'd adjust the sensitivity down automatically - even in manual mode.

Or at least I assume that's what was happening? The depth on the quarter would go to almost nothing too depending on the amount of steel, with one large nut sometimes it'd be 70% depth, another time it was like chunked out at 40% giving me all or nothing overload, even after I removed the nut - just stuck there with useless depth. Probably some ground balancing happening there too.

It was unpredictable though, sometimes took a bolt, sometimes 2 nuts, sometimes 1 nut (was using 3/4" hardware). 

The problem is that once it adjusted down, even after I removed the nut it wouldn't adjust back always (or ever sometimes), even if I was in manual. Sometime's it'd stick me with like 70% reduced depth, other times it'd stick me at like 30% reduced depth. And when I rebooted, sometimes I was still at some odd amount of reduced depth compared to baseline. Like literally - I have a measuring stick with a 4" variation on the max detection depth of US quarter starting from reboot (with no steel on the coil). The lack of repeatability was concerning, because now I wonder if sometimes I sit on a surface nail too long and then go the rest of the day with a reduced amount of depth due to Geosense? Dunno.

Or is Geosense reducing depth independent of the steel and just adjusting sensitivity on it's own differently every reboot due to the EMI in the shop? 4" is a huge variance, lots of missed gold too if something is happening I don't know about in the field. We know hot ground kills depth, but are we losing tons of depth in noisy EMI environments too from Geosense?

I dunno. Hasty test. Read nothing into them. Just an oddity. I'm side eying my 6000 with a bit of suspicion now though. I need trust things are doing what I configure them to do. I missed a metric ton of gold (all 1 gram or smaller, within the 6000 wheelhouse) last summer with the 6000 that I got with the 7000. I know the 7000 is just more powerful to begin with, but now I'm wondering if something else happened...

 

[jasong]

Yeah, after further testing, the 6000 definitely is having some weird highly variable automatic sensitivity adjustment thing going on under the hood (we knew this already) depending on EMI around it and also on prior exposure to big iron (3/4" nut for instance). Problem is that this seems to bleed over into the manual settings too, and it seems to sometimes exist still after rebooting even. Hard to tell though since there is no screen to look at the adjusted settings data.

The problem isn't that it's happening, it's that it stays. And is unpredictable. I mean, like 3-4" difference on a US quarter when I'm testing under identical conditions. That's too much, I can't not be knowing if I'm running 3" shallower at random times for the rest of the day, because I hit some big iron in a hole last target. Or because there was some EMI storm that went away.

Maybe I'm missing something here. Maybe it's just my machine. But I have some trust issues with my 6000 now. Honestly I'm not sure I have enough trust to keep using it. And I'm really hoping Minelab gives us an ability to turn GeoSense or automation completely off in whatever the 8000 is, and/or user ability to custom limit min/max allowed ranges for it to auto adjust sensitivity, etc within, and give us some on-screen visual gauge where we can see what adjustment levels it is at to verify, and a reset button for the Geosense adjusted settings.

Basically for Geosense: we need a "dashboard" like a car, that shows relevant settings levels so it can be confirmed how the detector is running. And we need some slider controls that let us determine the max/min levels Geosense is allowed to adjust each settings. It's too hard to trust this or know how we are running, without knowing. Total lack of control is ok for ACE250 level models, not pro models. 

[phrunt]

I always thought they should have a way to disable Geosense, I would not be surprised if it was hindering me in my mild conditions.  I like the idea of a dashboard type thing displaying what settings Geosense is changing, we know it does sensitivity and ground balance, perhaps it goes further than that accounting for sweep speed, on the older GPX we had to set that ourselves and most went with Very Slow even if they didn't swing very slow which isn't ideal 🙂  It also lets in more EMI in Very Slow so in bad EMI environments the 6000 maybe adjusting that to counter EMI which would affect performance. 

It's been said the GM1000 and GPX 6000 in their Auto + can be more sensitive in ideal conditions over being in the highest manual modes, I've not seen this at all and manual has always been more sensitive for me.

The reason they took control away on the 6000 was for their biggest market in Africa where they didn't like controls and many often didn't even understand them and just painted on the control box where to put the switch.  Feedback out of Africa would have been it needs simplified on the new model, they didn't take into account the other users that like to have some control, and over simplified it.  It demonstrates they had a lot of confidence in Geosense performing well.

Your tests are interesting and concerning and easy to replicate so I'll try put some big iron down and test a smaller target shortly after hitting the iron and see what the results are, if it's desensitized the detector and how long it takes to recover, if it's slow to recover this is a big problem if hunting in an area with a lot of junk, and makes you wonder if it will behave the same way if you hit a area with bad hot rocks, and if it does then it's a very big problem for me.

I hope if there ever is an 8000 they don't force us to use Geosense, and leave manual control of ALL settings available on it.

 

[jasong]

32 minutes ago, phrunt said:

It's been said the GM1000 and GPX 6000 in their Auto + can be more sensitive in ideal conditions over being in the highest manual modes, I've not seen this at all and manual has always been more sensitive for me.

I'm definitely not seeing that either. Just with the coil resting 4ft in air, the most mild you can get, I was definitely getting more depth on a US quarter in manual than Auto+. 

Let me know how the tests go - it's so snowy here still that I really have no good place to to test. So I just testing in my shop - 18ft ceilings, lots of space around me, but still with 60hz power, lights, routers/switches/security cams/ etc so granted not the best environment and the EMI itself could have been forcing Geosense to act like that.

I didn't notice at first, but I had marked with a sharpie various tests on a yardstick. Going back a few times later to test other things I realized that it was still desensitized, compared it to the yardstick, and definitely was still losing depth even when I switched to manual. Most of those tests I started in Auto+ though, and that may also have had something to do with it. 

The iron was literally sitting right on the coil for various lengths of time as I tested other things. Sometimes it seemed to hold over into manual, other times not. Dunno. The adjustments and what it affected was unpredictable enough that I just turned it off and stopped and resolved to not entirely trust my 6000 anymore. 🙂 I need to be able to see the adjusted settings levels...

 I'm curious what you find outside of the noisy EMI shop environment that I'm stuck in though. The high EMI may be the cause for semi-permanency?

*Also, with the 3/4" nut, there is a huge difference between one on the coil and two on the coil. One I can still get some ok sensitivity, 2 nuts completely kills it down to like 1" of detection depth, or less. This begs the question - how much would a theoretic good PI iron discrim actually help working in heavy laden iron sites with lots of big surface iron? First thing I'll be testing if such a discrim circuit is included on an 8000 or whatever. And makes me more certain of what I've said for a while - what I really want is a reliable depth discrim circuit, more than iron discrim.