Announcement

**Carl-NC** · 06-05-2009, 01:32 PM

Click here for Prior art

Since White's has already filed the patent application, I can spill a few beans on this. White's has been developing this technology for a number of years, and has a fully working prototype. It got back-burnered for a while due to other priorities, but I have moved it up to the front. Yes, it goes really deep. Yes, it discriminates. Tinkerer alluded to some of the results we get in his IB-PI posts, and now Moodz has noted the way we get there using a sinc (or psuedo-sinc) transmit.

Sorry Paul! But good thinking!

- Carl

**Mechanic** · 06-05-2009, 02:19 PM

Moodz, don't you just hate that!

**moodz** · 06-05-2009, 03:44 PM

Well thanks for raining on my parade guys ....

However I am not entirely sure that I agree with the prior art ...
But this should be an interesting subject for discussion ..

Here are my observations ...

1. The prior art makes constant referral to sampling and timings .. this is clearly time domain whereas the method I proposed is all in the complex frequency domain ... timings are immaterial. ( longer explanation required )

2. The term "half sine pulse" is used. I cannot find reference to the expression sinc ... sinc(x) .... or sin(x)/x or a numerical method for generating a waveform that is precisely of a sinc characteristic ... this is of critical importance .. I would ( will ) have to post some data to show that if you use some non numerical ( non dsp ) method to generate this waveform your phase and frequency tx reference tend to wander all over the place. Half sine pulses are not the same as sinc pulses.

3. Doh ... strike point 1 ... I see there is a catchall in part 74 of the prior art.

I reckon it stands or falls on the use of a sinc waveform vs a half sine pulse and method of utilising same ... let the games commence

regards,

moodz.

**moodz** · 06-05-2009, 03:54 PM

ah ... knew I had seen it somewhere

http://www.zonge.com/PDF_Papers/TEMposterAs.pdf

use of half sine pulses .....

err ... is this a secret division of Whites or are they violating your patent ?

... there is a spectrum of half sine pulses there .. not exactly like a sinc.

moodz

**B^C** · 06-05-2009, 10:33 PM

Gday Guys,

Isn't this just old Radar technology from the 1950's or earlier, i think you'll find that there's plenty of old patents regarding this. I can remember a detector, not sure of the name at the moment that used this in about 1984 or a bit earlier.

I know it has been looked at by many but there were issues with it, i'll try to find some old books i have regarding this.

One would think if Whites thought it was better than ML it would have hit the market already.

**moodz** · 06-05-2009, 11:47 PM

Originally posted by B^C View Post

Gday Guys,

Isn't this just old Radar technology from the 1950's or earlier, i think you'll find that there's plenty of old patents regarding this. I can remember a detector, not sure of the name at the moment that used this in about 1984 or a bit earlier.

I know it has been looked at by many but there were issues with it, i'll try to find some old books i have regarding this.

One would think if Whites thought it was better than ML it would have hit the market already.

Cool .... I would like to see links to that info if you can dig them out B^C ... I am not ashamed to say that the inspiration for this actually did come from ADSL technology ... I would hazard a guess that it can only be done effectively using DSP .. not sure if they had that capability back in the 50s.

moodz.

**moodz** · 06-06-2009, 12:09 AM

half sine compared to sinc

Hi Carl ... I coded a new function call into my reference system called halfsine ... now I am violating the patent !

It generates a half sine pulse or multi thereof .... Everything is the same on my reference system except I am generating half sine pulses.
The screenshots are taken from the recieve hardware.
The screenshot below shows in the top half the time domain pulse as emitted by the DAC and the frequency spectrum in the lower half. The DSP guys out there will notice that the frequency domain characteristic is sinc ... so the halfsine does have an underlying sinc characteristic ... unfortunately it is in the wrong domain .

now have a look at the sinc pulse ... only running to 10 Khz BW here. The time domain has a sinc characteristic ... top half of screenshot ... the frequency domain .. lower half has a rectangular pulse characteristic ... and if you are wondering ... yeh ringing still occurs even in frequency domain. There are 2048 frequency slots in that spectrum ... all of ( nearly ) equal amplitude and known phase. ... ie 2048 parallel virtual detectors.

moodz

**moodz** · 06-06-2009, 12:46 AM

Originally posted by B^C View Post

Gday Guys,
blah blah (edited

)
One would think if Whites thought it was better than ML it would have hit the market already.

the prior art patent that Carl cited has one catch all about using DSP most of the claimed analysis is done in the time domain .... nothing wrong with that ... however the DSP route simplifies the electronics however it requires heavy lifting in the CPU dept. ML and others ... Whites ? have played around the edges with DSP ... however they applied it to rectangular pulses in the time domain .... there is an old saying in engineering .... something like think outside the square ... well with regard to pulses ... no one has taken that advice literally.

moodz

**Carl-NC** · 06-06-2009, 03:33 AM

What are your thoughts on frequency-domain processing? BTW, the f-d ringing is an artifact of your finite DFT.

- Carl

**moodz** · 06-06-2009, 06:27 AM

Originally posted by Carl-NC View Post

What are your thoughts on frequency-domain processing? BTW, the f-d ringing is an artifact of your finite DFT.

- Carl

Hi Carl,
well ... I could write maybe a small book on that ... I will list some points and if anyone wants to expand we can discuss it.

1. The electronics is arguably alot more simple. Basically your coil ( IB or single ) a Transmit amplifier , Receive amplifier and DSP / operator interface block.

2. Because all the processing is done in the DSP there are a number of benefits here ...
- technically the signal to noise ..... and I mean that both practically and subjectively ...
... I dont just grab the signal off the receive amp and dft it ...it first hits a digital lockin amplifier / synchronous demod implemented in DSP that recovers the pulse from the amplified signal fed to the ADC. It can recover a pulse from below / at the noise floor. Factor in the gain of the coil amp and the further processing gain when the pulse is factored to the frequency domain and you are streets ahead of time domain techniques.

3 Adaptive equalisation or compensation of the external electronics, fancy low noise amps help but to a large degree the processing gain of the DSP compensates for middle of the road components phase / freq /amplitude compensation, environmental adjustment, snazzy user interfaces. No need to factory align .. the software does it on a continuous basis.

4 commercial benefits / author copyright / recognition ... the hardware cost is not too significant but the consumer pays for the level of software in the machine ... DRM anyone ? ... there I said it .. aaaargh .... wash mouth out with soap.

5 detector upgrade / maintenance is obviously alot easier with a firmware load.

6. Nothing I have said above is ground breaking .... what needs to be factored in here is an improved method of creating a magnetic pulse that influences a target in the presence of unwanted influences and by processing the received signal the wanted influence can be separated optimally from the unwanted influence.
What I am saying is that for a given bandwidth the sinc pulse is the most spectrally efficient pulse you can use since it presents the target with the optimal range of amplitude, frequency and phase controlled spectral energy which if the target interacts with even a single spectral component will be detectable. Of course most targets will interact with more than one component.... hmmm because of the number of analysis points involved the software is looking for 'signatures' .... ie combinations of changes in the signal. This is done .. no offence ... crudely in time domain ... eg ground balance with multipoint sampling. You will notice that I have emphasized the method of generating, receiving and benefits of the sinc pulse ... but not the methods that would be further employed to provide descrimination, ground balance etc and this is because there are either well known methods as employed in VLF etc and / or patents that already exist ( ML or Whites again ? ) that relate to these analysis methods and so these would be well known to those skilled in the art so to speak.

... of course I have a plan to bypass any IP difficulties in this area ... I know what is required to implement it ... but one step at a time.

What sort of hardware to use ... I use the guts out of compact netbooks like the eeePC by ASUS but I have had a look into what the IPOD touch can do and this would easily have enough grunt + cool user interface to do the DSP block ... well it has a reasonable SDK.
I did try out a 256 point version on a dspic at 30 MIPS ... it works but ...hey you get what you pay for.

... hey look the rant meter is moving into the red ... gotta go.

... ps the comment about ringing was a very poor attempt at humour.

moodz.

**Aziz** · 06-06-2009, 10:03 AM

Hi all,

I have my doubts on getting the spectral pulse energy into the coil. We all know, the coil has complex impedance jwL.
The reactance Xind = wL = 2*PI*f*L

So the higher the frequency, the higher the coils complex resistance. This wouldn't allow a much high current flow through the coil and we end up with diminishing frequency response.

Indeed, my broad band transmit signals in the past showed the inefficiency (low sensitivity) of this method. But a good discrimination possibility on the frequency spectrum.

On a single frequency, the reactance of the coil can be set to zero with a capacitor (Xind+Xc=0) and this is the case using the coil in resonant mode.

One also could use the chirp modulation.

Aziz

**moodz** · 06-06-2009, 01:05 PM

Hi Aziz,
there is a field not totally unrelated to metal detecting ... MRI ... magnetic resonance imaging .... a typical field coil is 1 Henry or more ... the time constant is between 2 and 20 milliseconds .... and yep you guessed it ... they use sinc pulses ... why because its the fastest way to reach maximum flux without causing EMI type artifacts in the very sensitive receivers and I would say that a 1 henry coil would have a hell of a flyback and supercooled as well
The rectangular pulse is the worst pulse to feed into an inductance because of the high dv/dt at the edges ..... however the rectangular pulse is linked to sinc by the fourier pair relationship ... the sinc waveform has its sharp edges in the frequency domain ... inductances are not as sensitive to transitions in the f/d ....
.... another way to produce a sinc pulse is to fire a rectangular pulse through a low pass filter ... thats how I discovered all this when I was trying to produce PI impulse from the sound card of my toshiba and it generated sinc pulses because of the 12khz low pass filter on audio out.

anyway if you can achieve power density with a rectangular pulse ... and we all know you can or PI s would not work ... well by inference you can with a sinc pulse .. because it comes from a rectangular pulse

... just bang a big amp on it instead of a switch .. you really only need to go up to 20KHz or so ... VLF on steroids.

.... regards,

moodz.

**simonbaker** · 06-06-2009, 04:33 PM

Originally posted by moodz View Post

Hi Aziz,
there is a field not totally unrelated to metal detecting ... MRI ... magnetic resonance imaging .... a typical field coil is 1 Henry or more ... the time constant is between 2 and 20 milliseconds .... and yep you guessed it ... they use sinc pulses ... why because its the fastest way to reach maximum flux without causing EMI type artifacts in the very sensitive receivers and I would say that a 1 henry coil would have a hell of a flyback and supercooled as well
The rectangular pulse is the worst pulse to feed into an inductance because of the high dv/dt at the edges ..... however the rectangular pulse is linked to sinc by the fourier pair relationship ... the sinc waveform has its sharp edges in the frequency domain ... inductances are not as sensitive to transitions in the f/d ....
.... another way to produce a sinc pulse is to fire a rectangular pulse through a low pass filter ... thats how I discovered all this when I was trying to produce PI impulse from the sound card of my toshiba and it generated sinc pulses because of the 12khz low pass filter on audio out.

anyway if you can achieve power density with a rectangular pulse ... and we all know you can or PI s would not work ... well by inference you can with a sinc pulse .. because it comes from a rectangular pulse

... just bang a big amp on it instead of a switch .. you really only need to go up to 20KHz or so ... VLF on steroids.

.... regards,

moodz.

1. Regarding Aziz's last comment -- what kind of forcing function would be required to make a typical metal detector coil "exhibit" a sync function? By the way, are we talking about the current in the coil, or what?

2. A delta pulse is really a limiting case of a sync pulse where the rectangular pulse in the frequency domain is infinitely wide and the sync pulse is infinitely narrow. It seems the advantage of the sync signal over a delta pulse is to limit the bandwidth in the frequency domain and increase the amplitude by compressing the power into that range -- this should make working with the data less noisy. The question is - exactly what frequency band would give us the most useful detection and discrimination.

It is in our interest to use the narrowest band possible to get the best signal to noise, yet still provide discrimination. Wouldn't it be funny if that turned out to be a sine wave!

To continue the S/N argument, the sync wave forces us to pick up noise proportional to the bandwidth. Because of that, I'll propose: it may be better to use multiple sine waves for discrimination than a sync wave because of superior signal/noise and almost as good discrimination. I know that's old hat, but it seems to make sense from s/n standpoint.

Regards,

-SB

**simonbaker** · 06-06-2009, 04:39 PM

Originally posted by moodz View Post

Hi Aziz,
.... another way to produce a sinc pulse is to fire a rectangular pulse through a low pass filterpass filter on audio out.

....
moodz.

A perfect brick wall bandpass filter would be needed, wouldn't it?

Regards,

-SB

Announcement

Missing Link between pulse and continuous waveform detectors ?

Missing Link between pulse and continuous waveform detectors ?

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment