AMX TX

[IBGold]

Hi Ian ... yep I can send you one. Do you need the programmed CPU chip as well ?

I think I used to have your address ... PM me where you want it sent.

moodz

Will do and yes on the chip you know programming is not one of my strong points.

Regards, Ian.

[moodz]

Hi Paul sorry should have posted in other thread I meant the Moodz V3 PI board that's what happens when you get old.

Regards, Ian.

Hi Ian ... yep I can send you one. Do you need the programmed CPU chip as well ?

I think I used to have your address ... PM me where you want it sent.

moodz

[Willy Bayot]

On the real XMIT circuit, one should carefully select the best types of available MOSFET.
As I have understood, it is not only a requirement related as usual to Id, RDS(on) and Vds but also internal capacitance and integrated diode behaviour which are critical to the good switching process.

[IBGold]

Hi Ian ... AMX boards aren't designed yet ??

Hi Paul sorry should have posted in other thread I meant the Moodz V3 PI board that's what happens when you get old.

Regards, Ian.

[moodz]

Do you have any boards available.

Regards, Ian.

Hi Ian ... AMX boards aren't designed yet ??

[IBGold]

Do you have any boards available.

Regards, Ian.

[moodz]

Thanks Moodz,
Pardon my ignorance but what is ZP damping..?

have a look at this post ... https://www.geotech1.com/forums/foru...tion#post16662

[Dean Sarelius]

Thanks Moodz,
Pardon my ignorance but what is ZP damping..?

[moodz]

Hi Moodz,
Just curious about the TX and RX coil arrangment you are using to sample the 0.1gram nuggets..?

Just a regular ( is it 10 inch ? ) minelab mono ( 300 uH / 0.5 ohm ) and the coil has to be really close to the nugget ... prob about the same sensitivity as a ML GPZ7000 which can just detect the same nugget at 1 - 2 cm.

Its the active ZP damping that makes the difference .. this coil is a "standard" Minelab mono coil that is probably good for 5 or 6 microsecond sampling with a regular PI and would never detect a nugget this small.

ZP damping shortens the sampling to 1.4 microseconds thereabouts ... effectively making this a "fast coil" even though its not.

I have a 20 inch basket wound coil using litz wire and the ZP damping achieves just under a microsecond. So fast coils are faster also with ZP damping.

moodz

[Dean Sarelius]

My concern from Carl's post #52 was that the AMX design is specifying high speed sampling ADCs and complex dsp schemes ( which is kind of against the requirement of a cheap detector for artinsal gold fossicking ) whereas if you take the basic idea of #52 combined with the bipolar TX but OC / open circuit the RX coil during TX flybacks ( target charge times ) then using the long L/R of a shorted coil RX times you have plenty of time to sample small ( and big ) targets.

There is absolutely no need for high speed ADC / DSP.

I do have a ZP design built up and it has about 20 bits worth of sampling integration and gain in the front end and 12 bits of "real" ADC in the back end to give around 30 bits of real sensitivity. It damps in around 1.4 microseconds and can detect 0.1gram nuggets with ease.

I am going to post the schematics ... soon ... documentation has not been tracking hardware

moodz

Hi Moodz,
Just curious about the TX and RX coil arrangment you are using to sample the 0.1gram nuggets..?

[moodz]

Are you sure that this circuit enables to compensate the fast variations of energy absorption?

Is it not possible to inject some extra controlled current over the one supplied by V4?

You are kind of limited by the laws of physics ... the recirculating high voltage flybacks "initialise" the next pulse current and with no losses are at the right voltage so that there is ideally no ramp in the coil current. Now if you want to correct that ramp from the start the required energy would be similiar to a flyback ... just increasing the voltage on the power supply wont work because inductors dont want to change current suddenly unless you hit them with something big ... like a flyback pulse. So you cant do on the fly current correction even more so where the energy loss is non-linear over the duration of a single pulse ( eg ground that absorbs magnetic energy for half the pulse duration but then saturates ).

Additionally even if you could design some sort of high speed ( many Mhz ) feedback loop to "track" the current ramp you are literally trying to compensate mA variation in an absolute current of Amps. Feedback loops need some error to work from ...they cant track on zero error.

A sensitive metal detector state of the art is probably sensitive to sub microamp variations ( V=IR V=1e-6 x 500 = 0.5mV ) ... yeh definitely well under a microamp. So your ramp control would have to be much better than 1 microamp variation in the control loop to prevent falsing and noise.

Pretty difficult to do in hardware ( noise and expense ) when I will practically guarantee you a solution can be done in software the signal can be sampled. Just need the correct dynamic range on the input.

As an example I can hook up my CRO connect to the RX coil ... set max sensitivity / AC coupled and offset adjust / set averaging on / Fine conversion / uber sampling rate and I can easily SEE targets coming near the coil..... and it only has an 8 bit ADC. The scaling and offset in a CRO allow for huge input dynamic range which allow examination of sub millivolt features in a 100 volt input signal.
If the 8 bit CRO can do that ... then the 24 bit AFE will do much better by several orders of magnitude.

The short answer is that the solution i posted does correct pulse by pulse ... but not sub pulse variations.

moodz

[Willy Bayot]

...getting back on track. One way of adjusting for losses is to boost the loss with a parallel ride on inductor inside the detetor with a variable resistance.



By varying R3 the "ramp" of the CC pulses in L1 will be adjusted. Of course this doubles the power but this particular circuit with 1.2 volts has 500 volt flybacks and uses 1 watt.

This will only be good to adjust for certain types of losses and only to a certain level.

moodz

Are you sure that this circuit enables to compensate the fast variations of energy absorption?

Is it not possible to inject some extra controlled current over the one supplied by V4?

[moodz]

"Cheap" was not a stated requirement, Tony's comment was that it should not cost more than an ounce of gold. Keep in mind that these markets are currently paying Minelab prices so even a detector priced at 2 ounces might sell well. But for an $1800 (USD) detector you want a total production cost of $450, or 25%, for decent margins. A $100 PCB would easily fit this requirement. A direct-sampling approach, even with a $30 ADC and a $10 micro, is still quite cheap, no doubt cheaper than a TDI board was to make.

That said, I'm very interested in learning more about your scheme.

You are right. I realise I need to discriminate between cheap and cheaper.

[Tinkerer]

My concern from Carl's post #52 was that the AMX design is specifying high speed sampling ADCs and complex dsp schemes ( which is kind of against the requirement of a cheap detector for artinsal gold fossicking ) whereas if you take the basic idea of #52 combined with the bipolar TX but OC / open circuit the RX coil during TX flybacks ( target charge times ) then using the long L/R of a shorted coil RX times you have plenty of time to sample small ( and big ) targets.

There is absolutely no need for high speed ADC / DSP.

I do have a ZP design built up and it has about 20 bits worth of sampling integration and gain in the front end and 12 bits of "real" ADC in the back end to give around 30 bits of real sensitivity. It damps in around 1.4 microseconds and can detect 0.1gram nuggets with ease.

I am going to post the schematics ... soon ... documentation has not been tracking hardware

moodz

open circuit the RX coil during TX flybacks
This sounds interesting.

[Tinkerer]

"Cheap" was not a stated requirement, Tony's comment was that it should not cost more than an ounce of gold. Keep in mind that these markets are currently paying Minelab prices so even a detector priced at 2 ounces might sell well. But for an $1800 (USD) detector you want a total production cost of $450, or 25%, for decent margins. A $100 PCB would easily fit this requirement. A direct-sampling approach, even with a $30 ADC and a $10 micro, is still quite cheap, no doubt cheaper than a TDI board was to make.

That said, I'm very interested in learning more about your scheme.

If the detector finds the gold and is rugged enough to last one full season, it would even sell well at US$ 5000, but until proven in the gold fields, the initial price should be around US$1800.