SD2100 SD2000 Auto Ground Balancing Modification

[mario]

Hi guys,

I have replaced one of the ground balancing pots in my unit with an X9118 chip. My controller runs the ground balancing smooth with the use of an encoder and separate push button (conveniently located on the handle). Here's a photo (pre install) and short video clip with a X9117 prototype.

Having said that, I would like to get your input on this issue.

Minelab does not implement AUTO ground balancing on channel 2 with their recent models. As far as I know, the 4500 and earlier (also later?) models don't provide this feature. As users know, channel 2 is "the other" channel, the one that finds the bigger stuff, deeper. For most people, channel 1 is the channel that will find ALL the stuff.

I could use another X9118 to drive channel 2; but, it seems like channel 2 will use up more resources than it's worth. Implementing two X9118s to manual ground balance is not that complicated. However, when implementing code to AUTOMATICALLY drive channel 1 AND channel 2, things will have to get much more sophisticated for very little return.

Having said that, is channel 2 worth sacrificing for AUTO ground balancing?

[IBGold]

Hi Mario,

Firstly Channel 1 is the long pulse for deeper bigger targets and Channel 2 are the multiple short pulses for smaller targets to get optimum performance on the SD2000 BOTH MUST be ground balanced correctly I definitely would opt for both.

Regards, Ian.

[mario]

Thank you Ian,

I guess the timings on those machines must be handled different for minelab to eliminate one of the channels.

I have reassembled my detector so many times, I, a long time ago, forgot which channel is which when reassembling.

This poll is not necessary and I ask a moderator to end it if at all possible.

[Mechanic]

G'day Mario,

They have not dropped one of the channels on the later models. They have just removed the channel selector switch. What you could do is make a small board, or incorporate into your board another set of filters for the GB channel, just like what is already there for ch1 or 2 without the adder cct. As Ian said, better to keep ch1, it is good to have both handy.

Cheers Mick

[Jiminy]

Hi Mario and all. I also would opt for both. This will be a great addition to a good machine. Definitely a needed upgrade.!!!!!!!

[mario]

Thank you for the input guys.

I don't want to half *** this, so your suggestions will definitely make a difference.

However, it looks like we're pushing the limits of our little Atmega. The addition of a 2nd channel for manual ground balancing will bring the total close to 21k of data space, leaving us with 11k for auto gb. I've never done any hard core signal processing, so I'm not sure what to expect.

Remember, our original intention was to use this processor for driving the DSS sub board. I chose the Atmega328 because it had a decent feature set and is fairly flexible. Had I anticipated more hard processing work, like the one we will be doing, I would have chosen something with at least 40 MIPS. Still, from what I've been readying, we definitely have something to work with, so our implementation should be fairly ok.

Also, I still don't want to post schematics because I have not experimented with the additional hardware required for processing. To date, I've only worked on the manual ground balancing part. And manual gb is not something that our detectors currently need since it's a stock feature.

All in all, I'm happy with how work is taking shape, and it's been a great learning experience, which I hope to continue even after we've completed work on this project.

I'll keep you updated.

[IBGold]

Mario,

What the 2000 does is process both channels target response then picks the strongest and passes this on to the audio that is why the two tone response either High Low or Low High which ever is the strongest channel sorry to be a spoil sport and make it harder.

Or would it be easier to leave the manual ground balance as it is as a fixed manually adjustable G/B and just add an Auto G/B mode so you could use both if that is possible I can see advantages there Just to be difficult Hi.

Best Regards, Ian.

[mario]

Mario,
...sorry to be a spoil sport and make it harder.

Or would it be easier to leave the manual ground balance as it is as a fixed manually adjustable G/B and just add an Auto G/B mode so you could use both if that is possible I can see advantages there Just to be difficult Hi.

Best Regards, Ian.

No, not at all. Like I mention earlier, I would like to do it right, so I don't mind undertaking this new challenge. I've been wanting to do multimedia processing with my uCs, so this will be a nice problem to tackle.

I needed clarification on the issue because I wasn't entirely sure how the later models handle both channels. After reading several posts, some people appear to miss the selection switch for ground balancing. I'm now going to assume those same people are misinformed with how Minelab handles the timings. Like Mick mentions earlier, Minelab doesn't eliminate any of the channels. It just implements them differently.

Also, I plan on following your suggestion and making manual ground balancing optional. I'm currently finding the range narrowing feature (seen on the video) very useful in my working unit.

However, the manual ground balancing option will be 100% digital, through the LCD/uC/RDACS, so those two 100k pots and selection switch will have to go.

I'll keep you updated and thanks for all the feedback

[gef12]

So what you are saying Mic (I think) is that they have included a summing amp of sorts for both the channels .. It would be good to have the both channels .. No doubt Mario there will be a bit of work in doing this GB mod

[Mechanic]

Hi Guys,

The thing is the micro needs to know when the coil is lowered to the ground. Currently it would be impossible for the micro to sense this, because if you were just sampling the existing channels the pitch/ voltage direction would change if we went past the gb spot. So we would need to sample the output of the bandpass filter(final signal processing stage) as little as 10 times a second to know if the ground signal is getting stronger or weaker as well as sampling both ch1 and ch2 to identify which way the gb needs to be adjusted and how close the actual gb is. There would need to be some provision for the adc to sample voltages from -5v to +5v, so there may need to be a voltage divider that keeps all of the voltages within the rails of the adc. Also it needs to be sensitive down to 5mv change on the -5 to +5 side of things.

There would be no need to remove the functionality of the channel selector switch, this is after the stage where we would need to sample and I like it as it can help give you an idea on what to expect to jump from the ground!

Cheers Mick

[Mechanic]

G'day Mario,

This sucker may be just what the doctor ordered http://www.analog.com/static/importe...ets/AD7609.pdf

Will handle +-16v on the inputs! No scaling required....

Cheers Mick

[mario]

G'day Mario,

This sucker may be just what the doctor ordered http://www.analog.com/static/importe...ets/AD7609.pdf

Will handle +-16v on the inputs! No scaling required....

Cheers Mick

Wow Mick, those are serious specs!!
+-16V differential inputs
18bit with oversampling
8 simultaneous channels
internal oscillator (!)

We have enough resources to handle both gb channels in addition to automating emi handling if we wanted to.

At 64 pin QFP package, this chip makes me want to draw a second board instead of using a soic-dip solution (like the one for the two X9118 I found on ebay). However, if we go this route, we might as well replace our little 28 pin Mega with something faster like an ARM CORTEX 3 with full speed USB (and boot loader) to better compliment our setup. That way, we may also integrate the two X9118 RDACS and even use a more affordable DSS solution instead of the pricey AD9851.

Before considering this overhaul, we have to can find something with a lower pin count. All we need is the differential inputs , two channel inputs, and at least 16 bits IMHO. Otherwise, it might be another year before completing this project (lol)...

Last DSS project was done in almost 7-9 months...I know I slacked off for most of the time, but hey, between it, the gf, my job, and house/car repairs, I kept my sanity....lol

Merry Christmas guys,

I know we'll figure something out.

[gef12]

Ye ok Mick would be good to leave the switch in situ as you could make comparisons ..
Powerful chip to by the way and interesting times ahead albeit a lot of hard work for Mario.
So Mario this would be an upgrade and replace your existing mod .. hope you all had a great xmas..
cheers

[mario]

Ye ok Mick would be good to leave the switch in situ as you could make comparisons ..
Powerful chip to by the way and interesting times ahead albeit a lot of hard work for Mario.
So Mario this would be an upgrade and replace your existing mod .. hope you all had a great xmas..
cheers

Hi Geoff,

I'll end up doing a separate board, but for a different project, and not related to our SD2XXX detectors. I need a full featured development board and haven't found one with the features I would like.

However, that wont happen for another few months, after completing this project. I would like to start on a pin pointer project Steffan had been helping me designing. I have procrastenated for too long and will likely design my board around it.

Anyway, getting back to our current work, I will likely use the LTC1859 I found at Linear's website. It has everything we need and will fit in the little smd->dip board ( I found on eBay) nicely.

[Mechanic]

G'day Mario,

Nice looking ADC very versatile. Should suit the application. In a much more practical package too.

When you get up to programming make sure you have a sufficient delay between adjusting the pot and re-sampling because when the pot is adjusted a response will be seen at the output of the channel that is being adjusted.

Cheers Mick