More FBS Geekery

Non FBS geeks please drive thru.
;^)

OK, it's been another crappy wx day so I decided to do some more research. I downloaded a free audio spectrum analyzer program (SpectrumLab V2.77) and hooked my computer's mike input to a couple wire loops around my ETrac's X-1 probe. After fiddling around awhile I was finally able to learn some very interesting information. I was able to see the actual harmonic frequencies as transmitted by the ETrac. The analyzer program is only useable up to 8KHz but some very interesting data was able to be captured. Too bad the program dosen't work all the way up to 25KHz.

I posted this in another thread but here is the actual transmitted signals of Noise Channels 1, 6, and 11 as viewed on an Oscope. Notice again NC #1 is the lowest frequency and NC #11 is the highest:



Here are the harmonic spectrums of each of those Noise Channels between 0HZ and 8KHz captured by the software analyzer:



Notice how the spikes (harmonics) are spaced. The harmonic located near 1500Hz for NC #1 moves up to ~1750Hz for NC #11. The other harmonics are shifted up similarily. I assume all the other harmonics above 8KHz are shifted the same.

So what does this mean? Well it reinforces my belief that the FBS technology uses harmonics of a 25KHz pulse train to generate harmonics starting at ~1500Hz for NC #1, which in turn can be increased to harmonics starting at ~1750Hz for NC #11.
Does this mean that high conductors will respond better at CH #1 and low conductors better at CH #11? My speculation is that is the case.
If anyone has any more insight please chime in!

Now, with all that research done, I'm ready for a couple adult beverages.

This is fantastic, and what I've wanted to see for a while. I'm curious, are these measurements repeatable upon manual selection of NC channel?

Yes, they are repeatable thru all Noise Channels and totally cool to look at!

Good info...

Thanks for sharing...

I always thought my Explorer XS got the best depth on high conductor's on 1 and and vice versa...


But what was strange was you could not turn it on and get great depth till the noise cancel was done then it would get good depth...then you could change the channel...I think there was more going on than just simple freq shift....

Freq shift is part of it but I believe other parameter's are also adjusted.....Just dont know exactly what they are..

Good info...

Keith

Silently lurking, watching, reading, studying, pondering and anticipating the next post.

I'm with markg...

Steve

Another sound-card scope/spectrum-analyser program you might try is a free German one called SpecOszi. On one of my laptops, I was able to run it a 192000 samples/second. This should allow you to see more high frequency detail.
Edit: I only have it on a Packard-Bell at the moment, the soundcard max. sampling rate on this is 96000 samples/sec, the spectrum analyser display goes up to half this, ie. 48KHz, so plenty enough to show the 25KHz signal.



Edited 1 time(s). Last edit at 02/11/2012 06:18PM by Pimento.

Beautiful work Yeasty, if Minelab won't tell us then you will! Keep it up and hopefully you can get some software that works up to 25kHz. Interestingly Carl from Whites said the same thing about the E-Trac and 25kHz being the upper frequency. (Or something along those lines). But he also said that changes between the channels would be minimal, more for EMI reduction and not really having a a noticeable effect on high or low conductors.

Can you see what the actual range of frequencies are on each channel? I mean are they off by just a a fraction of a percentage or is it wider than that, justifying the better response on high conductors at CH1 and low contuctors on CH11. In other words if the differences are small we are just splitting hairs, but if the frequency set difference is from e.g. 7kHz on one channel to 15kHz on another channel (at the low end), then that is significant. And I think on the MLO forum a poster mentioned that to be the case.

Great work
EMS

Great thread........... and all without me having to 'infringe'!



Edited 1 time(s). Last edit at 06/18/2012 12:40PM by NASA-Tom.

It does look like chan 11 is about 20 - 25% higher in frequency than channel 1. Equivalent to a change from 13 KHz to 16 KHz, or from 5 to 6 KHz, for example. The difference is subtle, but probably enough to make a difference on those on-the-limit signals. Worth considering, though, is the question - Is the detector optimised for the middle channel 6 frequency, and might operating at 1 or 11 be sub-optimal for that reason?

Im about to ask a dumb question.... how exactly are you determining which kHz the machine might select as the BEST setting? I see it shifting so can you say chanel 1 is like 1.5 kHz, channel 2 is 3 kHz? That would kind of match the BBS which is a set 1.5 kHz increment. Ive also read it selects 3 of the best freqs and one may be GB which might answer Keiths question about more going on. Bet ya cant tell im dirt dumb on this lol.

Dew

Dew, all the harmonics are there all the time.. It uses them all at the same time and sorts out the target information on the receiver side. Auto Noise Cancel is used to find the quietest channel.

For FBS, MineLab claims coverage between 1.5KHz and 100KHZ and I believe them. The 25KHz square waves also generate harmonics above 25KHz. I just wish I had a spectrum analyzer to see them.

I love science class.

Does your graf show 28 freqs at random settings rather than fixed? So its sending 28 harmonics and receives the same harmonics then processes them for TID? This and magic are about the same to me lol. But i really really enjoy the information you guys provide.

Dew

This thread states the '28 frequencies' claim is more advertising than technically useful reality. There will be lots of harmonics generated by a square-wave signal. What you do with them is the important thing.
[www.dankowskidetectors.com]

Well I drove thru as you lost me on OK....Indeed the winters are long in my neck of the woods and again whatever works and imagine a light went on for some of our members reading the thread.

Your not the only one Dan. I read these posts in the hope that I may understand something....and sometimes I do, imagine that. Every so often something clicks, even in my no tech brain.

Picked up this photo of the frequency.


[i1124.photobucket.com]

Yeasty Wrote:
-------------------------------------------------------
> Non FBS geeks please drive thru.
> ;^)
>
> OK, it's been another crappy wx day so I decided
> to do some more research. I downloaded a free
> audio spectrum analyzer program (SpectrumLab
> V2.77) and hooked my computer's mike input to a
> couple wire loops around my ETrac's X-1 probe.
> After fiddling around awhile I was finally able to
> learn some very interesting information. I was
> able to see the actual harmonic frequencies as
> transmitted by the ETrac. The analyzer program is
> only useable up to 8KHz but some very interesting
> data was able to be captured. Too bad the program
> dosen't work all the way up to 25KHz.
>
> I posted this in another thread but here is the
> actual transmitted signals of Noise Channels 1, 6,
> and 11 as viewed on an Oscope. Notice again NC #1
> is the lowest frequency and NC #11 is the
> highest:
>
> [p.briggs.home.comcast.net]
> HANNELS1-11.JPG
>
> Here are the harmonic spectrums of each of those
> Noise Channels between 0HZ and 8KHz captured by
> the software analyzer:
>
> [p.briggs.home.comcast.net]
> PECTRUM.JPG
>
> Notice how the spikes (harmonics) are spaced. The
> harmonic located near 1500Hz for NC #1 moves up to
> ~1750Hz for NC #11. The other harmonics are
> shifted up similarily. I assume all the other
> harmonics above 8KHz are shifted the same.
>
> So what does this mean? Well it reinforces my
> belief that the FBS technology uses harmonics of a
> 25KHz pulse train to generate harmonics starting
> at ~1500Hz for NC #1, which in turn can be
> increased to harmonics starting at ~1750Hz for NC
> #11.
> Does this mean that high conductors will respond
> better at CH #1 and low conductors better at CH
> #11? My speculation is that is the case.
> If anyone has any more insight please chime in!
>
> Now, with all that research done, I'm ready for a
> couple adult beverages.

With snow on the ground the speculation continues.

From what we've seen above, FBS detectors transmit harmonics. So now I wonder how the machine uses them. Here's a couple quotes from MineLab's website:
"Utilising FBS technology with simultaneous frequencies ranging from 1.5kHz - 100kHz the signal received from the detectors coil is analysed from a wide range of responses. This allows E-TRAC's advanced signal processing to analyse more target information so that target identification is more accurate."
"Minelab’s FBS technology simultaneously transmits 28 multiple frequencies from 1.5kHz to 100kHz."

So here's some more speculation:
FBS technology monitors 28 of the harmonics between 1.5KHz and 100KHz in the receiver (shifted slightly up or down by the selected Noise Cancel #). I strongly suspect that FBS machines act like a spectrum analyzer. It analyzes the target's response like I analyzed the transmitted signal, but on a much broader scale. If the target is a high conductor the amplitude of the received signal will be higher at the lower harmonics of the recieved spectrum and conversly the amplitude of a low conductor will be at the higher received harmonics. Where the target's highest response falls in the received spectrum is measurable and a value (TID) can be assigned by the software.

Here's a representation of how I think the ETRAC responds to high and low conductive targets and how it assigns its TID.

PLEASE NOTE!! This picture IS NOT an actual measurment!





Again, it's only my speculation. If anybody has some more insight or think I need a sanity check please chime in!



Edited 1 time(s). Last edit at 02/12/2012 07:28PM by Yeasty.

Well, the snow hasn't melted so I started thinking about how MineLab figures out the FE properties of a target.

Actually, it's thoroughly explained in this MineLab document:
[www.minelab.com]

This is page 24 of that document where it explains how they do it.



From my understanding of their explanation, FBS uses the received wide pulse (in red) in front of the shorter (25KHz) pulses to measure FE properties. Apparently the more ferrous the target the deeper the slope of the received wide pulse. In other words if the target is a silver dollar there will be hardly any slope (and give a 0 FE reading) and conversly a big old hunk of iron the slope will be very deep and give a 35 reading).

Here's an expanded and modified portion of the above picture and how I think the FE properties are determined and scaled using the wide pulse. They simply determine the amount of the slope and the software assigns a corresponding FE number.




DONT FORGET!!
THERE WILL BE A QUIZ TOMMOROW!!



Edited 2 time(s). Last edit at 02/12/2012 11:23PM by Yeasty.

Hi Yeasty,

Unfortunately, much of your speculation is incorrect. Two frequencies, that's all they process. No more.

In your first post, your spectral plots show severe aliasing problems due to the narrow bandwidth. You need a real spectrum analyzer (1MHz or so) to get a decent picture of the signal content. In any case, the harmonics are a side effect that (unlike the Fisher CZ's) aren't used as "frequencies." I recall when Garrett released their Infinium they (tongue in cheek) advertised it as a "99 frequency" detector, an obvious poke at Minelab's advertising. No one took Garrett seriously, of course, because everyone knew what it was all about. White's briefly (and not very seriously) considered calling the V3 a "43 frequency" detector (or some silly arbitrary number, I don't remember) but we decided to call it what it really is.

- Carl

Thanks for the reply Carl.
Here's what I think. The method used by MineLab for measuring Ferrous properties is published by MineLab and to me is completely reasonable. Now, as far as Conductive property measurements, in my opinion, the measurements I made were reasonable and followed what I suspected. The fact that there is a 25KHz pulse train, and my measurements showed that there indeed were harmonics of the 25KHz, makes me believe that they use the harmonics as I described. You are correct in the fact that I need a wider bandwidth spectrum analyzer. Until someone can do a better analysis, that's my claim and I'm sticking to it. If I am proven wrong I will learn something with open arms, which is good. As far as Fishers, Garretts and Whites, I have absolutely no experience with them nor their advertising claims. MineLab, in their advertising and claims, intrigued me and I set out to see if they were true. To date my investigations seem to go along with what they claim. Also, I have absolutely no affiliation with MineLab except for being a happy ETrac owner, curious, and get bored on days with crappy weather.
Pete



Edited 2 time(s). Last edit at 02/13/2012 05:19AM by Yeasty.

Hey Yeasty....Carl is the Engineering manager for Whites Electronic's...

Keith

Keith, thanks, that's cool.
My experience is a 40 year career in Air Force avionics maintenance engineering.
Carl should have a better capability to prove or disprove (and post) what I have posted with his own analysis (with pics).
Again, I want to learn.
Pete



Edited 6 time(s). Last edit at 02/13/2012 02:57PM by Yeasty.

I'm enjoying this thoroughly, Yeasty. Not understanding it all completely, but enjoying nonetheless...

Steve

Pete. Just curious. Where did you detect/sense the 'whine' from Carl? ((( I have been blind before ))). Yes..... I agree that 'supporting evidence/data' is very important.......... but I could only detect that Carl was stating 'matter of fact' from his viewpoint...... and may have been limited on time from a more lengthy post. ((( I fail badly in this department...... more than anyone would know! I only wish I had more time to make 'comprehensive' posts! ))).

Tom. I agree with you and have edited my post.

Pimento Wrote:
-------------------------------------------------------
> It does look like chan 11 is about 20 - 25% higher
> in frequency than channel 1. Equivalent to a
> change from 13 KHz to 16 KHz, or from 5 to 6 KHz,
> for example. The difference is subtle, but
> probably enough to make a difference on those
> on-the-limit signals. Worth considering, though,
> is the question - Is the detector optimised for
> the middle channel 6 frequency, and might
> operating at 1 or 11 be sub-optimal for that
> reason?

Yeasty and others, what do you all think about the above statement, regarding NC Channel 11 being around 20-25% higher than NC Channel 1? IF that is true then there is a substantial difference between the lower and upper channels, at least theoretically but I would think there would be noticeable differences on test targets. Reminds me of a video showing NC Channel 11 was more sensitive on small gold than NC Channel 1. It was quite clear there was a difference.

EMS, my understanding is that the entire spectrum is moved when selecting a Noise Channel. The lowest harmonic seems only to move ~250Hz between NC #1 and NC #11.

Yeasty - Thanks for the reply. I don't understand then, how could there be a noticeable difference in NC 1 and NC 11 on a e.g. small gold earing when the kHz difference is 1/4 kHz (or 250Hz)??? Sorry I'm not an electronics guy.
I am a bit mixed up because at the top you said the harmonics start at 1.5kHz and move up to 17.5kHz and I don't understand at a practical level how this relates to the overall picture of response on a target and such.

And for us wet behind the ears here, what does it mean when you say the entire spectrum is moved? You mean of the 25kHz square wave? And if so (of if not!) what does that mean to a target. I'm really just trying to see what your research means in the field but am also interested in the theory, though I lack this kind of knowledge...

Thanks again,
Albert