GDX 8000 Repair/fix Project(GPX 4500 Clone)

[brazza]

Hi Guys,

recently purchased a GDX 8000 (Gpx 4500 Clone) & as suspected the quality was appalling, Shafts, coils, accessories etc unusable & need replacing , AC battery charger was Dead on arrival etc.
The detector itself once hooked up to some real coils did work though.
Although i noticed the machine was excessively noisy & has repeating jitter & groans that can't fully be tuned out ( Talking in the middle of goldfields away from any mains or interference).
Also while it will ground balance out a hot rock i have as a test piece on the bench , when trying to ground balance in the gold fields it was completely unusable (Normal, Enhance, Gain down to min etc) & acted the same in different areas
with different types of mineralisation.
As I owned a GPX 4500 for 8 years i am well aware on how one should behave & this definitely is not it!

I Decided to investigate what is causing the machine to behave this way & opened it up to begin removing the treacherous white paint to start taking measurements & checking components.
Unfortunately i no longer have my trusty GPX 4500 to compare values & components against( & my memory is not that great).

Anyway i have a couple of findings & a few questions for anyone with some know how on these machines.

• Pre amp used: duel AD797a - I initially suspected wrong preamps but it appears as if they have used the correct ones
• Integrators op amps used- TLC2262c- I thought these were supposed to be TLE2072?? TLC2262 seems to have really poor specs (0.7mhz band , 0.55v/us slew)
• Quad Opamp next to AD797 preamps- TL064- I'm not sure what is the correct Op amp for this? but TL064 seems a little bit outdated
• Fly back Blocking fets used appear to be correct-2SK1334 N-Channel , with secondary fet in reverse bias
• Fly back was measured at 225volts - should be clamped to 185v as the blocking fets are rated to 200v i'm not sure if these would be starting to avalanche at this causing my noise issues?
• If flyback is wrong i'm guessing the snubber comparator circuit resistor dividers ref are set wrong?

I've attached a couple of marked up drawings highlighting my queries.

Does anyone know if the area highlighted on my marked up photo referencing the potential snubber circuit is correct(The snubber control for flyback ) , based around 2 x TL2022 duel opamps and 2 x TLC372 comparators?

Any help would be greatly appreciated!
Just thought this would be a good project for me rather than it going straight in the bin.
& if i destroy it in the process no big deal!

Cheers,

Brazza

[eclipse]

I wouldn't trust markings on the chips. AD797 may very well be NE5534. I would replace all OP amp chips in the front end with better /lower noise, better specs/ upgrades. Yes try to investigate the snubber circuit to reduce the voltage so the mosfets don't avalanche.

[brazza]

Thanks for the suggestion Eclipse,
come to think of it i did notice while gently scraping the white paint off the ic's some markings seemed to scrap off very easily which lead me to believe they weren't genuine ICs.
i'll start with replacing the preamps with genuine ad797 & see if i get an improvement.

[eclipse]

I have similar clone board. They've used NE5534 on the place where AD797 is expected. All other front end op amps /integrator section/ are low-end cheap op amps but on the original GPX all chip markings are deleted with CNC so nothing can be seen as far as markings go.

[Mechanic]

Brazza, the problem with the Chinese clones is that SETA does not function correctly.
SETA is used to remove Earth Field from each of the timings and is calibrated before Minelab release a detector for sale.
To test this, take a small ferrite magnet, like that from a small speaker with all metal removed. Then wave it around your room. You will find that it picks up this magnet quite some distance away. Any movement of the coil will result in moaning and wailing.

If, when they copied the software, they copied out the full contents of the memory, the SETA values may still be present, but the detector electronics not calibrated to the set values.
To fix this assuming that the calibrations are present you would need to:
Set the gain of the final front end stage correctly(resistor tolerance).
Set the EF resistors for each of the channels correctly. Good luck with that!
Failing that, there are pads for trimpots and you might be able to set EF for one timing, I'd do enhance.
I tried this with a GFX7000, but it did not work, I could not get EF cancellation to work correctly.

They make a good trouble shooting and emergency parts mule!

Cheers Mick

[brazza]

Thanks Mick,

I did notice it was picking up the Earth field i was hoping it was due to inadequate Opamps causing GB & EF subtraction to not function correctly, bummer.
Oh well, a good lesson learned i guess.

Just out of curiosity is the SETA EF subtraction from CH1, CH2, GB performed after the integrator stage? In the MCU?
As per the timing you posted in another thread i noticed CH1, CH2, GB each has it's own pos and Neg sample/samples which i assumed was fed/muxed into the +,- inputs of the integrators.
But wasn't sure whether the SETA channel was actually EF cancelation & fed into the - input of the CH1 , CH2, GB integrators (summed with the neg samples in your timing diagram maybe) or whether it was subtracted later in MCU.
(Or i could be completely wrong & the SETA channel is just used to set the sample width or gain of each neg sample for CH1, CH2,GB)

If anything i will use this board as a learning platform to better my own PI designs.

Cheers Brayden

[Mechanic]

The subtraction is all done in the microcontroller. Ch1, 2, gb etc still have their samples so EF cancellation is close, but the resolution of the FPGA that spits out the timings must not be enough to fine tune to perfection.

Cheers Mick

[Jumper]

Its rubbish I think.

[Op04]

Can someone tell me what mosfet used in tx of GDX8000 / GPX4500 or GPX5000, just curious hehe..thanks!!

[Op04]

AUTO ground balance.

what I meant to say was auto tracking analog ground balance. I've seen tepco method and apply it but that's not good enough, I also have look eclipse posted somewhere. This part is mostly done with MCU which beyond my understanding. So please if anyone have it please share here. Since I look everywhere this was never put in a subject. But I can understand if that too much to ask. Thanks all the master that already share their briliant thinking and knowledge. Regards.

[Orbit]

Гпк 9000 хттпс ://иоуту.бе/Л0Ј_ј9Зу8д4 .

[ivconic]

[Divardetector]

Hi guys
Regarding making changes from friends, could he do something to make this system a little more stable? Friends, if anyone has information, please share

[eclipse]

The amount of knowledge needed to modify the detector cost more than the device itself multiple times. If I had the knowledge I wouldn't just put it here because it's quite a lot of work to get to the point where you know what to modify.
Second hand GPX 4500-5000 now cost 1500-2000$ and this chinese crap cost 800$. Why would you want to spend 800$ on the cheap clone that just doesn't work as good as the original, and try to modify it is beyond me.

[Sundarto]