signal strength

[green]

Prospector Al made a comment in a thread I was reading. If the detector doesn't detect a nickel at 15 or 16 inches with a 11 inch coil put it back on the shelf or sell it on ebay. Might have remembered the diameter wrong. When building or buying a metal detector what are some non field tests a person should perform. How much would the distance change for different size and type coils for the nickel test.

[bklein]

Are you bring the new unit to the ******* this week?

And what is ******?

[Gwil]

If I may contribute another quotation:

"There is no sadder sight in the world, than to see a beautiful theory killed by brutal fact."

-- Thomas Huxley, biologist (1825-1895)

[deemon]

And on the 30th August 2010 I replied:

As many enthusiastic designers have discovered, with metal detectors there can be a large gap between the initial theory and what happens when you try to put that theory into practice. Your learning experience will be improved if you try building something.

http://www.geotech1.com/forums/showthread.php?16963-TX-for-metal-detectors&p=116442#post116442

Obviously, nothing has changed.

But on the other hand , one good scientist said - "Nothing is more practical than a good theory"

[Qiaozhi]

George, it is time to use term WIDEBAND or BROADBAND instead PI detector. Even MINELAB makes this.
I answered to the above your question two years ago:

http://www.geotech1.com/forums/showthread.php?16963-TX-for-metal-detectors&p=116428#post116428

And on the 30th August 2010 I replied:

As many enthusiastic designers have discovered, with metal detectors there can be a large gap between the initial theory and what happens when you try to put that theory into practice. Your learning experience will be improved if you try building something.

http://www.geotech1.com/forums/showthread.php?16963-TX-for-metal-detectors&p=116442#post116442

Obviously, nothing has changed.

[mikebg]

Please post the full schematic of a PI detector that you have personally designed using these theoretical techniques, and that has been successfully tested in the field.
We are waiting .....

George, it is time to use term WIDEBAND or BROADBAND instead PI detector. Even MINELAB makes this.
I answered to the above your question two years ago:

http://www.geotech1.com/forums/showthread.php?16963-TX-for-metal-detectors&p=116428#post116428

That means CHINELAB Unlimited had enough time to test this:

http://www.geotech1.com/forums/showthread.php?16019-Ancient-ultra-low-noise-narrow-band-preamplifier&p=101783#post101783

http://www.geotech1.com/forums/showthread.php?16019-Ancient-ultra-low-noise-narrow-band-preamplifier&p=106445#post106445

Moreover, the CHINELAB Unlimited knows the reinvented block diagram of the most sensitive metal detector:

http://www.geotech1.com/forums/showthread.php?14025-RELIC-HAWK&p=160264#post160264

If you send me via email this PDF file, which costs only $ 2, I'll send you more details:

http://www.geotech1.com/forums/showthread.php?19940-Pulse-Induction-History-and-Theory&p=167343#post167343

[6666]

My current design has a 2-stage preamp, each 20dB.

Is that the magic number Carl ?

[Carl-NC]

That was the plan, but today I completely cooked my receiver board. O.E. Now in the process of tediously converting another board.

[Mark-VA]

Are you bring the new unit to the ******* this week?

Depends on the design. My current design has a 2-stage preamp, each 20dB. Any more than that screws things up.

[Carl-NC]

Speaking from a VLF point-of-view, AGC has very limited use. We're dealing with a small target signal on top of a potentially large ground signal, so there is limited headroom to AGC the gain up higher. And, as Davor points out, you would need a fast-attack AGC to avoid instant saturation, and even then you are likely to get AM-to-PM modulation that screws up the TID. If the AGC has discrete gain steps, you will also get signal discontinuities that screw things up. This is one of those Good Ideas that doesn't quite work out when you actually build it.

What you really want is a way to null the ground signal. Literally remove it at the coil, so that the preamp never sees it. This is also fraught with pitfalls and is (so far) a Good Idea that doesn't quite work out when you actually build it.

[Davor]

Mike, you can't push only a single side of a story just because it works somewhere else. Metal detectors operate over somewhat larger dynamic span than short wave radio does, so perhaps some broadening of perspective might be in order.

You often propose AGC as a solution to all problems, yet, what kind of AGC you think about is not entirely clear. Most of the AGCs as seen in Ham solutions choke on such strong signal you encounter in metal detectors. There are a few possible options that could cope but one more exotic than the other one, using vactrols, digital potentiometers and what not. And to what end?

When using a metal detector, most of the time there is no target to observe, so the system should graze right on the noise floor anyway. Once the target comes, and boy it comes sharply in, your system must be ready for the full span, only to recover ASAP to the noise grazing state. No AGC could ever cope like that. There are attack and release times attached to the AGC concept that would render your AGC-equipped rig useless. 200ms release time? You miss two targets with that.

Shortening attack and release times to zero you obtain a simple clipper. Well hello to the metal detecting technology!

We are dealing with signals that come with such steep power law transfer characteristic you simply don't want to convert them into sound in their raw form. You need to compress them, one way or another. Binary sound is perhaps the worst form of it. Proportional sound is a somewhat compressed form of the target response turned into sound and it sounds immensely better. VCO also, within reason.

There is one technology you probably missed in Ham radio world that comes to mind, a SSB voice compressor. You have a SSB modulator producing SSB at some low-ish frequency, say 455kHz, followed by a clipper (!), followed by a simple filter and a demodulator. Albeit crude compression is involved in such compression, intelligibility of such voice is good, spectrally all the spikes are there as in the input voice signal, but oddly equalised in amplitude - the voice signal is severely compressed, and surprise! there is no AGC to be seen.

Whole point is that if you render an AGC to a clipper with no attack and release times - you get what we already have. Even the SSB IF in form of a zero IF of the metal detectors.

[Qiaozhi]

That means you use a primitive block diagram because:

1. Your preamp operates without AGC to maintain gain near to saturation. Then you can't use maximal possible gain and maximal sensitivity to small and deep targets,

2. Your TX has no P-I control to suppress amplitude modulation when TX coil moves. Then your RX receives a modulated by ground AIR signal, that is awful, and

3. There is no automatic balance in input of your RX. Then it cant operate at maximal possible sensitivity.
IMNHO, if this is true, you can't see noise and even interference in preamp output.

I posted in the forum images of preamp noise at wideband and narrow band preamps when AGC maintains maximal sensitivity.

http://www.geotech1.com/forums/showthread.php?16793-Frequency-Domain-and-Time-Domain-relationships&p=165898#post165898

http://www.geotech1.com/forums/showthread.php?16793-Frequency-Domain-and-Time-Domain-relationships&p=167545#post167545

You really achieved such output?

Please post the full schematic of a PI detector that you have personally designed using these theoretical techniques, and that has been successfully tested in the field.
We are waiting .....

[mikebg]

In most of the designs I'm involved with, AIR+GND isn't the problem; the preamp(s) is not close to saturation, even with a fairly strong ground. We could increase the preamp gain, but typically the chatter gets so bad it is unusable. The same is true in trying to apply more post-demod gain, too much chatter. So, in my experience, it is mostly a noise problem.

That means you use a primitive block diagram because:

1. Your preamp operates without AGC to maintain gain near to saturation. Then you can't use maximal possible gain and maximal sensitivity to small and deep targets,

2. Your TX has no P-I control to suppress amplitude modulation when TX coil moves. Then your RX receives a modulated by ground AIR signal, that is awful, and

3. There is no automatic balance in input of your RX. Then it cant operate at maximal possible sensitivity.
IMNHO, if this is true, you can't see noise and even interference in preamp output.

I posted in the forum images of preamp noise at wideband and narrow band preamps when AGC maintains maximal sensitivity.

http://www.geotech1.com/forums/showthread.php?16793-Frequency-Domain-and-Time-Domain-relationships&p=165898#post165898

http://www.geotech1.com/forums/showthread.php?16793-Frequency-Domain-and-Time-Domain-relationships&p=167545#post167545

You really achieved such output?

[Overtheedge]

Think I've got it now. Not really noise but rather co-channel interference from the ground signal the MD induced to re-radiate.

An analogy might be standing and shouting on a canyon rim and hoping to use the echo to determine the distance to a spire midway across the canyon. Too many extra echoes masking the desired echo.

By zero IF receiver, I'm assuming you mean Direct Conversion (DC) receiver centered on the oscillator frequency. The frequency conversion stage output being bandwidth limited to reduce noise prior to further signal processing. And then we are faced with filter Q realities. Wow.

And ground is not homogenous so a lot of different decay rates, many clustered but with some outlayers and still find the needle in the haystack just because it is more conductive. Wow. The ground return is like the spray paint hiding what is underneath. The heavier the coat, the more hidden it is until it can't be found except by excavation.

This MD theory and practice is a whole different animal completely. Between ITMD and this forum, I should get it figured out.
eric

[golfnut]

I read the word S/N (signal to noise) ratio and I think noise. 40+ years in RF communications (job and/or hobby) and I tend to get stuck thinking in those terms.

Eric,
These are generally fairly simple zero IF receivers. The IF filters are often bandpass, centered at about 10Hz, their BW is limited in attempt to reduce noise as you say.
The signals come though, post mix, at low Hz due to the sweep speed over the ground, fast sweep long shaft - 12Hz? - No motion - mixed signal comes through at 0Hz with the image at DC.
Medium sweep maybe 5 - 8Hz.


The main problem is that the ground signal comes though in the IF filter passband and appears as a co-channel type of signal.



Most IB VLFs have 2 Rx channels,

GEB - CH is phase correlated with the gnd signal - with 180deg offset in mixer LO signal so that 1/2 of detected geb signal is above 0v and 1/2 below 0 axis in order to average the gnd to zero
( this CH detects target alloys too, but to lesser extent)

Disc CH is set to max the sens on the expected target alloy - by opening up the mixer at the time - wanted alloys should be there
( it will detect gnd signal too, but to a lesser extent, as not phase correlated with gnd signals)



Both Rx CH outputs are amplitude compared to a common reference. If and only if both CH outputs are above the common ref you get a tone.



The weakness is the GND CH, it needs to have almost nothing coming out of it while there is no target -Perfect GND null, in order that a small target will give a small output above the threshold. The Disc CH will give a bigger output and will easily trip the common threshold.


Perfect GND balance is possible with a manual GND tune but only for The ground where the balance was done - as you walk away the GND signal will grow (as different ground) the level out will grow and trip the threshold and false. The user is left with having to either raise the trip threshold or retune GND bal.


With a quality auto gnd system on a mid or top machine the GND balance is done for you maybe hundreds of times per sweep. So this GND co channel signal is very low almost all of the time - this leaves the user to set a very low alloy trip threshold. So now, tiny real alloy targets are above the very low level GND residue - and allow a real alloy to trip at these low levels - more sens and depth for the same Tx power..


In the summer I fell foul of the need to regularly tune out the GND signal as best you can.

I was on a field in UK - naturally iron rich (ironstone/red soil/furnace slag) - I did not reset my GEB properly one time.

After not getting much, I dug in a regular brass Yale door key- I didnt get it at 3 1/2" ! ! !

After resetting the GEB best I could, I buried the key - flat - at 9 1/2" and I could get it on a regular sweep. In this condition - noise and preamp noise etc then becomes significant - albiet to a much lesser extent. With a normal radio these things are top drawer.

S