GPZ 7000 Tear-down

[phrunt]

Some may find this interesting, a tear-down of a GPZ 7000.

 

[jasong]

He has some other vids investigating the 7000 with a scope, appears it also has unshielded components putting out a bunch of noise like the 6000, but in this case it's in higher frequencies that don't seem to interfere with itself. Hopefully Minelab spends a bit more time on the next machines thinking about these things now though.

It's been my assertion for ages that the way we can make advances with current "stagnant" tech is by EMI/signal processing which is much easier today with modern MCUs, reducing noise anywhere possible should be job #1 now IMO. Lower the noise floor and you can boost RX amplification, and even without any RX boost it still becomes a magnitude faster/easier to detect for each magnitude of noise reduction so it's a win/win. Hopefully this is an aim with the 8000. Certainly they learned a noise lesson with the 6000 anyways.

It looks like this guy is doing some mod to reduce noise, so same general idea, will be kinda interesting to see what he comes up with I guess, though I doubt I'll be mailing my 7000 off to Australia. An external noise reduction module or something along that line would be interesting though. 

[phrunt]

The fact you have to be careful where you place a Bluetooth transmitter on the side of your GPZ as it causes it to go nutty shows it's not overly well shielded, I can put a Bluetooth transmitter on my GPX 5000 anywhere on the sides of the box and it's all fine and dandy.  In saying that the GPX 5000 is pretty poor at handling EMI by comparison to the 7000, so they improved the EMI capability even with lesser housing shielding.

I don't know why Minelab seem to often miss the simple things yet they're brilliant at the difficult things.

[jasong]

Ever seen Star Trek? In it is a race of beings called the Borg that are half computer and half alien. They act like computers insomuch as they won't attack an actual invader on their own ship unless they calculate them to be a threat, they'll literally let trespassers walk around among them and prefer to spend time by paying attention only to relevent matters at hand, not seeing the potential problem it might cause in the future having intruders there. But as soon as they change their minds and see you as a threat, you will be rapidly assimilated or destroyed, but often it's too late. And they pretty well assimilate all technology around them and make it into super Borg tech. No one debates they don't do it as well or better than anyone else in the universe though.

In a way, Minelab reminds me of the Borg. It seems like there are a number of very obvious things they simply choose to ignore because they simply aren't viewed as causing problems...yet. But when it's finally made obvious to them a problem exists, then they dedicate all sorts of resources to fixing it, but often much too late. They sure do like some good ole assimilation via patents from what I can tell too. 

[Rivers rat]

9 hours ago, jasong said:

Ever seen Star Trek? In it is a race of beings called the Borg that are half computer and half alien. They act like computers insomuch as they won't attack an actual invader on their own ship unless they calculate them to be a threat, they'll literally let trespassers walk around among them and prefer to spend time by paying attention only to relevent matters at hand, not seeing the potential problem it might cause in the future having intruders there. But as soon as they change their minds and see you as a threat, you will be rapidly assimilated or destroyed, but often it's too late. And they pretty well assimilate all technology around them and make it into super Borg tech. No one debates they don't do it as well or better than anyone else in the universe though.

In a way, Minelab reminds me of the Borg. It seems like there are a number of very obvious things they simply choose to ignore because they simply aren't viewed as causing problems...yet. But when it's finally made obvious to them a problem exists, then they dedicate all sorts of resources to fixing it, but often much too late. They sure do like some good ole assimilation via patents from what I can tell too. 

MINEBORG

[phrunt]

Here is the second part of the video

Some may find it interesting, others probably boring 🙂

I quite liked it, showing the noisy components on the board and he's working on mitigating noise to improve performance rather than increasing gain.

The next part will be interesting, comparing side by side a modded and standard GPZ.

[jasong]

The recent video he comes to the conclusion the 7000 is potentially interfering with itself - much in the same way the 6000 does from what I can tell - and his mod is shielding the inductors. Which is, probably uncoincidentally, the exact thing the 6000 "Fix" Minelab does from all the info we have.

He seems to know what he's talking about and they are interesting vids, but I really wish he'd spend a few moments editting his videos because like 80% of them end being him fiddling with the camera, talking about coffee, etc. He could have made this 35 minute video more like 3 or 4 minutes. 

[jasong]

I half suspect he might be reading posts here actually. 😅 Or at least aware of what the 6000 fix was.

Because later in the video he literally says exactly what I said here and have been posting for ages: lower the noise floor, you can boost RX gain. And this is how you can get better performance out of detector tech most people say is now at it's maximum already.

My assertion for years been that we don't have a power problem, we have a noise problem - people say we can't go deeper because of the X^6 exponential decay issue with TX power. But we don't need more TX power, we just need to lower the noise floor and there is a lot of RX boost to be had still. We listen for incredibly diminishingly tiny signals in astronomy by using tricks like receiver arrays. There is a lot of room to improve with detectors here too - up to and including using actual physical arrays of "coils" if someone really wanted to make a sensitive professional level detector. There is room for improvement still.

I can't believe no one has tried to seriously experiment with coil arrays, either stationary on a  ground grid, with a central pulsing TX coil, or arrays within a single detector coil. You could achieve both X-Y-Z positional data, gradiometry, and differential noise cancelling. FPGA's are more than powerful enough to deal with the amount of data computation now. 

[phrunt]

It seems to me EMI mitigation should be the number one priority for detectors now that they are really looking like they're at the peak when it comes to performance especially depth.  Even looking for smaller targets the key is often crank that gain as high as you can but you're limited by EMI more than anything.

People are already coming out saying they prefer the Nugget Finder coils over the stock Minelab coils for the 6000 as they seem to handle EMI better, allowing more gain and there have been aftermarket coils around for  years that handle EMI better than the standard supplied coils from Minelab.  This seems a common theme, the coils that handle EMI well are the ones people say are going deeper or allowing more gain and finding them more gold.  The Concentric coils on the GPZ are a good example of that too.

The Manticore has a gain of 35 when the Nox was 25, yet few people will ever get near 35, when most could get to the 25 on the Nox in certain places, 10 more numbers of sensitivity that are available if conditions allow.  I have spots here I can run mine at 35, remote spots but it shows if EMI is not an issue you can run it at maximum without the circuit producing the noise, it runs as quiet as a mouse on 35 away from EMI.  I'd love to run 35 when its quiet and stable on some of my good spots, unfortunately too much EMI there for that.  The long noise cancel method on the Manticore is something they should implement on all detectors, the 6000's super quick method is a fail in my opinion, why did they not come up with the long press idea when the 6000 was being developed, it would make a nice difference being able to hold that button down and give it some time to get a clean channel rather than pressing it over and over again hoping.  The long press narrows it down far better and when you feel it's settled on a number you let go of the button and done. 

CONTINUOUS AUTO NOISE CANCEL
In some instances, there may be more than one ‘quiet’
channel suitable for detecting all with similar levels of EMI
noise — you might notice this after repeating the Auto
Noise Cancel procedure multiple times, with each resulting
in a different channel being selected.
Continuous Auto Noise Cancel continuously repeats the
Auto Noise Cancel process over a longer period of time to
find the quietest channel.

If they've done all they can do performance wise looking at improving EMI resistance would be very beneficial I think, and I guess they learnt an expensive lesson with the 6000, and as Woody has pointed out in these videos too the 7000 has a similar problem as the 6000 just not near as bad but again, something they could have improved during the design stage had they focused on EMI reduction more.

He shows with his oscilloscope how easy it is to find components causing noise, I'm sure the Minelab techs already were familiar with this loop technique.

His mods for the 7000 will end up being superior shielding for a quieter running detector rather than the hardware mods he was doing on older models, he may end up changing a component or two for a better shielded version.

If you put your phone near your detector its interfering with the electronics, it's not the coil causing the EMI trouble its the detector itself, you can easily demonstrate it putting your phone right up near the control box, this is an extreme example of it letting EMI into the electronics that better shielding could improve and if it improves your phone causing EMI it's improving everything.

[jasong]

Proper shielding and isolation is like design 101 for sensitive radio electronics (like a detector). I can't convince myself that Minelab didn't know about this during design - I mean why intentionally put the power supply right next to the RX module?? It's not hard to put the power supply in it's own entirely shielded case and seperate board. It half makes me wonder if the intention was to produce a lower RX gain 6000 and 7000 using cheaper parts, then later models would come out with upgraded shielded components, lower noise, and a higher RX gain as a 6000 "Pro" and 7000 "Pro" using the same boards, or something along those lines, using substantially the same circuit but repackaged in a more modern ergonomic shaft/control unit. 

I guess probably not, since it's been 8 years since the 7000 release and doing a "Pro" style release could be done quickly. At this point, with that much delay between models, I'd hope the 8000 is something better. But I'm still left scratching my head.

*I agree with the noise cancel thing, but it sure is nice to have the illusion of speed on the 6000 quick cancel. What I'd like to see - given the power of the FPGA's today - is an automatic noise cancel that just always scans for the quietest channel in the background 24/7 so that you can switch to a quieter channel instantly if you want to.