FPGA based UNIPI system

[Altra]

I am using straight up physics but from first principles ... If you mean like descrim ?
It seems to now reject ferrous and hotrocks ( which it did already ) but any conductive metal registers as a target ( the more conductive .. the stronger response )

I was joking about the physics, I am in awe of your abilities. The ferrous question was serious, if it was still going in opposite direction to non-ferrous or the same. But no reaction is even better.
Thanks

[Carl-NC]

So I investigated why the hotrocks were not being detected ... the sensitivity was certainly high enough as I was able to detect a .05 gram gold target but not a hotrock ??.
nitially I thought I may have had a target "hole" and it was centred over maybe a range of targets including hotrocks.
More testing and code tweaking ... revealed that the hot rocks were not being detected because the code was already partially working and trying to reject ferrite responses ( ie opposite to conductive ) and this partial ferrite rejection was resulting in non detection of hotrocks.Once this fact was known the code was adjusted to cancel the ferrite response entirely and understanding this now allows a simple switch function (eg CANCEL ON/OFF) ... not sure when you want to detect ferrites though.

I'm not sure I understand, but if you are rejecting ferrite & hot rocks based on response polarity then that typically won't work in ground conditions. A large negative ground signal will mask a small conductive signal.

[moodz]

I'm not sure I understand, but if you are rejecting ferrite & hot rocks based on response polarity then that typically won't work in ground conditions. A large negative ground signal will mask a small conductive signal.

The code never looks at the polarity .... the target signals are whatever they are .. .I dont use thresholding or subtraction.
All the math is in signed integers ( large signed integers ! ) and so there are no rounding "losses" as you get in floating point.

and here is another thing to consider ...many designers ( including me upto about a day or so ago ) use an IQ demodulator to work out their magnitude and phases.
There is a usually a statement... its the first line in some older texts on IQ demods ( seems to have been lost in newer writeups I have seen ).
Generally it goes like this " The IQ demodulator was invented for situations where the phase of the transmit carrier is not known at the receiver"
People use them because they are ubiquitous and work well if the transmit phase is known or unknown.

In a metal detector .... the phase of the transmitter is known though .. and this fundamental piece of information means you dont have to use an IQ demod hence there is room for alternatives and even the possibility that vital information is masked by the IQ demod process.

This is still a work in progress ..

[moodz]

I was joking about the physics, I am in awe of your abilities. The ferrous question was serious, if it was still going in opposite direction to non-ferrous or the same. But no reaction is even better.
Thanks

Thanks Altra .. just to be clear it rejects ferrites and hotrocks but not large iron ... which registers as conductive since it is probaby more conductive than ferrous I am in uncharted territory ... so maybe full disc is just around the corner.

[moodz]

I'm not sure I understand, but if you are rejecting ferrite & hot rocks based on response polarity then that typically won't work in ground conditions. A large negative ground signal will mask a small conductive signal.

so thinking about this if i hold a ferrite at the coil and wave a target ... the target will be masked ? if the algorithm is wrong.

[Orbit]

Hi Moodz! If you take ferrite and iron at the same time and rub it in front of the coil, does it show as colored or uncolored?Usually on IB it turns the phase and if we wave both ferrite and iron in front of the coil it shows it as colored!The fact that you will hold the ferrite statically and swing an iron object, I don't think it will give a true picture of reality!?

[Carl-NC]

so thinking about this if i hold a ferrite at the coil and wave a target ... the target will be masked ? if the algorithm is wrong.

No, place a small nugget on the hot rock and sweep both past the coil, see if the hot rock masks the nugget. Generally, all true GB methods I know of require either phase notching (freq/phase domain) or signal subtraction (time domain) and will exactly cancel only one loss angle. It is possible to then use blanking techniques to expand loss angle rejection but that runs the risk of masking targets.

[Orbit]

Last winter I played with samples on the pulse device, in one setting it did not detect iron or a hot stone, however if I waved an iron and a hot stone in front of the coil at the same time then the iron would be detected, it did not detect colored objects at all neither without a hot stone nor together with him.

[Repwoc]

... it did not detect colored objects at all ...

What do you mean by "colored" in this context?

[Orbit]

What do you mean by "colored" in this context?

I mean coins, silver, copper, gold, bronze, etc.

[Repwoc]

OK - so non-ferrous metals.

[Orbit]

OK - so non-ferrous metals.

I know how you think in Britain! I happened to watch a video of a man who moved to England, he said that every year, whenever someone from the company goes on vacation, he brings something exclusive and something special from his country, it was the turn of that man from Serbia and he bought some black candies on with the name black odjacar, ordinary candies have a little spiciness but an unusual taste, everyone wondered why they were called black odjacar, as I understood from his story, they thought it was something like an insult to people with a different skin color. However, candies have nothing to do with that at all! It has nothing to do with valuation etc. those candies are a brand like Coca Cola, I hope you understand me?

[moodz]

No, place a small nugget on the hot rock and sweep both past the coil, see if the hot rock masks the nugget. Generally, all true GB methods I know of require either phase notching (freq/phase domain) or signal subtraction (time domain) and will exactly cancel only one loss angle. It is possible to then use blanking techniques to expand loss angle rejection but that runs the risk of masking targets.

I did that ...

Ferrite by itself = no response
Target by itself = target response
Ferrite + target = target response ( no signal strength reduction for non iron .. possibly a signal reduction for large iron )

Same result for hotrocks.

If a large ferrite touches the coil ... it cant cancel it because the internal numbers in the ADC / samples overload. But backing the coil off 2 cm or so recovers straight away.

Instead of waving the target I can just tape the targets to the coil spaced by a block of styrofoam as the detector is non motion and quite stable for measuring purpose.

There is no disc ... I rotated the "phase" through 360 degrees and the ferrite cancelled at one only one slot. Could not achieve for any metallic conductor.
There are probably other things I could try in the code but I have not thought of them yet.

In the mean time I need to integrate some communications stuff into the code like SPI so I can read and write things in the FPGA on the fly.

For me ferrite rejection is my main goal ... I dig everything in the field ...so disc does not matter ( to me )

moodz

[moodz]

To compare waveforms ... I changed the transmit code so we get half sine pulses. Seem to get a much stronger ( but monopolar ) signal as the signal to noise is about double the single frequency mode.
The method I used to cancel ferrite also works on half sine ( I was a bit surprised ).

Below is pic of TX waveform on CRO and target response on red trace VGA from the FPGA. ( there is no ferrite trace because its cancelled )
I think its time to build up a field unit for testing outside.

moodz

[Mdtoday]

To compare waveforms ... I changed the transmit code so we get half sine pulses. Seem to get a much stronger ( but monopolar ) signal as the signal to noise is about double the single frequency mode.
The method I used to cancel ferrite also works on half sine ( I was a bit surprised ).

Below is pic of TX waveform on CRO and target response on red trace VGA from the FPGA. ( there is no ferrite trace because its cancelled )
I think its time to build up a field unit for testing outside.

moodz

Looking good Moodz! getting very interesting now….

cheers