Pulse Induction History and Theory

[mikebg]

HOW ANCIENT PI METALDETECTOR WORKS
(See circuit diagram in post #37)

Below is shown normalized locus of transfer impedance valid for all conductive nonferrous targets.
The normalized locus uses relative frequency which has no dimension because is against the time constant of object. This allows in one locus to present the frequency spectrum of signal from two targets with different time constants. This method loses information about the scale (real amplitude of each target), but on the other it shows important spectral differences between received target signals. Here's how the normalized locus can explain the different timbre of the sound of two targets, which differs in timeconstant.

The TX of metal detector TIMBRE illuminates environment with powerful magnetic pulses which in frequency domain are represented by wide band of audible frequencies. The bandwidth is even larger than the so-called ELF radio band (300 ... 3000Hz). With letter A is denoted the lowest frequency of the band (in this case 180Hz) and with letter B - the highest (in the case 3600Hz). Each boundary frequency appears twice on the locus, which is impossible when the impedance plane is not normalized.

TARGET WITH SMALL TIMECONSTANT

For this target, the whole illuminating frequency band appears in the low frequency region of target's spectral characteristic. This is below cutoff frequency, where the phase lag is small #less than 90 deg, but the amplitudes and phases depend strongly on the frequency. Magnitude OA in the lowerst end of frequency band appears significantly less than the magnitude OB in the upper end . In order not to clutter the drawing, the vectors of magnitudes OA and OB are not drawn. Only vector 01 for magnitude at cutoff frequency is shown for reference.

The top right diagram illustrates what timbre of voice give this target. At relatively small timeconstant we will hear timbre SOPRANO because the amplitudes at low audio frequencies are relatively small to amplitudes at high audio frequencies.

TARGET WITH LARGE TIMECONSTANT

For this target, the whole illuminating frequency band of TX current appears in the high frequency region of target's spectral characteristic. This is above cutoff frequency, where amplitudes and phases depend less on frequency. Phase lag is more than 45 deg, but phase differences between frequency components are small. Magnitude OA in the lower end of the frequency band is virtually identical to the magnitude OB in the upper end of the band as shown in the top right diagram. In this case the operator will hear loud audio signal with timbre ALTO.

[mikebg]

HARMONIC ANALYSIS?

Eric, can you post me this paper?
[email protected]

I searched the Web and found the article, but it turned out that it is not free. The first page is given only and unfortunately it contains three sentences. I really want to know how Sir Horace Lamb calculated the series of time constants and distribution of eddy currents in cylinder volume.

In the early 19th century, Jean Baptiste Joseph Fourier (1768-1830) created harmonic analysis - a powerful tool for solving PDE, suitable also for analysis of sensing systems with eddy current.
The method is published with large delay in 1822:
Fourier, J. B. J.; "Théorie Analytique de la Chaleur". C. F. Didot, Paris, 563 p., 1822.
The book is published in English with more than half century delay:
"Analytical Theory of Heat", translated with notes by A. Freeman, the Cambridge
University Press, London, 1878, 466 pages. The English language book of Fourier appears 5 years before the article by Sir Lamb.

I do not know if the father of eddy currents Jean Bernard Léon Foucault (1819 – 1868 ) used harmonic analysis for solving PDE with eddy currents. The method for solving of homogenous PDE is as step-down response, ie as conventional PULSE INDUCTION metal detector - cylinder volume is filled with magnetic field at the moment of cessation of a previously constant field. Then begins a free decay process because magnetic energy dissipates as heat.
I'm familiar with the harmonic analysis of eddy currents in a cylinder, because in 19th century the method was taught in German universities for the specialty "Schwachstrom Elektro Ingenieure" ("Weak current electrical engineers" means electronics). Later it is taught also in French, Russian and Italian universities.

The harmonic analysis of eddy currents shows that cylinder has series of timeconstants and that there is distribution of eddy currents in the whole cylinder volume. It shows what is minimal width of TX pulse if we need eddy currents in the whole volume.

[Ferric Toes]

Snubbers and Blockers

This subject comes up from time to time as if it were some new technology. Snubbers have been around about as long as inductors in order to limit or control the high voltage that occurs when current is interrupted in and inductor. However, the interest here is their use in a PI metal detector. The first PIs that I was involved with in 1966 used a BUY12 silicon planar transistor to switch the coil current. This had a maximum V_ec0 of 80V. A snubbing network of a 10D8 diode, 0.47uF capacitor and a 2K2 bleed resistor were required to prevent the transistor being instantly destroyed. A few years later, silicon epitaxial planar transistors became available and it was found that these would happily run without an external network in a repetitive avalanche mode. A popular device, and relatively cheap, was the 2N3055 workhorse. Many manufacturers made this device, but interestingly only the Motorola version would switch fast enough.

In 1967, I was also involved in the design of a solid state front end for a seagoing proton magnetometer. Till then, relay switching of the polarizing current was the norm, but because we wanted to get both the polarising circuitry and the preamp down in the towed fish, we opted for transistorised switching. The sensing head had an inductance of around 20mH and a snubbing network was essential both for relay or transistor switching.

With modern high voltage mosfets snubbing networks are not required, except that there is some concern for advanced designs regarding avalanche noise in a balanced coil system, where measurements are taken during the current fall / avalanche time. Also, there is the argument that improved switch off times can be achieved by using a snubber network rather than avalanche mode.

Simple proton magnetometers use a changeover relay to switch the preamp and tuning capacitor either to ground or to coil, after the coil damping period. This is similar to a PI, except for the capacitor. In effect this is a blocker, and a similar suggestion was made in the 1956 paper “A Pulsed Bomb Locator” by F B Johnson. The relay operating time constituted the delay.



Today, mosfets have replaced relay contacts in most applications, and do not suffer the limitations of mechanical devices. Therefore, switching a PI coil and the preamp input using such devices is an obvious choice in many designs.

Eric.

[Aziz]

Thanks Eric,

for mentioning the subber topic.

What is a snubber?
L (coil), R, C circuit. Active during the flyback period. Either switched in with a diode or transistor.

But what is a snubber really?
It is a pulse forming network (PFN) of first order. Yep, it is. It is operated in reserve mode however: not discharging energy but charging energy. The process is reversible: either charging or discharging.

The critical series damping condition is:
R² = 4L/C -> R = 2*sqrt(L/C)

Have a look at the PFN of radar applications. They have multiple stages and are used to discharge the energy in a PFN into the impedance load.
Now use a multi-stage PFN snubber in a PI. That should give a very fast damping feature. A flyback pulse sucker!!!

Aziz


PS: Patent-trolls: keep out (if not already invented of course).

[Aziz]

Here are some basics of PFN in a radar application:
(Just consider, we want to use it in reversed mode - sucking the flyback pulse)
http://www.radartutorial.eu/08.trans...s/tx06.en.html
http://www.tpub.com/neets/book12/49i.htm
http://www.ga-esi.com/EP/pulsed-powe...orks/index.php
and so on...

Once the PFN is charged up, the energy in the capacitors must be taken away before the next pulse is sucked again. It can be dissipated into heat with damping resistor parallel connected to the capacitors of course (high || R, slow damping).

Cheers,
Aziz

[Ferric Toes]

Maybe you could use a high Q inductor instead of a capacitor. Tune the inductor to the TX prf so that it gets a kick up the millihenries on each flyback. A secondary winding could extract or return energy. Maybe that is already done by someone?

Eric.

[Aziz]

Maybe you could use a high Q inductor instead of a capacitor. Tune the inductor to the TX prf so that it gets a kick up the millihenries on each flyback. A secondary winding could extract or return energy. Maybe that is already done by someone?

Eric.

The simple snubber ("the first order PFN") is being used by ML. Oh yes, I think QED uses it too.

The multi-order PFN: Maybe Moodz - he has patented something but I don't know, whether it is the same.


It works. I have tested it with first order PFN in a spice simulation.
The pulse width of PFN can be easily dimensioned to meet the PI flyback pulse sucking conditions.

You all have a good idea to reduce the flyback period now.

Aziz

[ApBerg]

Hello Eric,

Also some history... Newbury 1986 can remember you showing some pi detector ?..not sure if it is you on the photo ?

Best regards.

Ap

[Ferric Toes]

Hello Eric,

Also some history... Newbury 1986 can remember you showing some pi detector ?..not sure if it is you on the photo ?

Best regards.

Ap

Yes, I was there somewhere, but not in the photo.

Eric.

[Ferric Toes]

In the 1960's archaeologists in general were much opposed to the use of metal detectors. There were good reasons for this at the time, but they did not appreciate how useful a tool it would be for them, when used in a proper and controlled way. In 1968 we were invited to do a survey on Delos using DECCO 4, not by the archaeologists involved, but by a numismatist, Tony Hackens, who was studying Greek coins of that particular historical period. The archaeologists did not like us being there, and gave us part of the site where, I was told later, that they hoped that we would not find anything, and pack our bags and leave. As it turned out we found so much that had been missed by visual searching that the use of the detector could not be ignored. Subsequently we were allowed into the main excavation where two bronze figure heads were found of the gods Hermes and Bacchus. Knowing that the objects were there prior to digging, ensured that there would be no inadvertent damage by spade or trowel. Bacchus is the one with a beard likened to a bunch of grapes.

I will post the full report as a .pdf, but having trouble uploading it at the moment.

Eric.

[Ferric Toes]

1968 Delos Report

Here is the full report referred to in my previous post.

Eric.

Delos.pdf

[Davor]

I think we all know Mr. Bacchus up close and personal

Trouble with archaeologists is that they are a frustrated bunch. Mostly because their silly projects are seldom financed, and when they are, the sites are more often than not already cleaned from valuables.

[mikebg]

1915 instead 1917

The year 1917 given on this site is not true.
http://www.whiteriverprep.com/vintag...ch/bhtech.html

The event shown in the engraving is happening in 1915. The publication in the journal Popular Mechanics is in February issue 1916, which means that the engraving was made in 1915 even if the article was written in January 1916:
http://books.google.bg/books?id=o90D...gbs_navlinks_s

There are enough information in Bulgaria from 1911 and 1912 for this "esigned by "French professor" metal detector. The information given by me in 2008 and above in this thread clears something very interesting written in the Feb 1916 article.

[Carl-NC]

Probably my fault... I think I erroneously said 1917.

[Ferric Toes]

I had a magazine Junior Mechanics and Electricity Oct. 1918 (since lost, unfortunately) where it said that farmers in France and Flanders had a metal detector fixed to a plough, to give warning of buried bombs and shells which could otherwise be detonated by impact with the plough. It mentioned that the coil was excited by a trembler device, and signal was on headphones.

Eric.