LF Coil antenna impedance

[mgburr]

I'm looking for ideas or references that will help me design multi loop coil antennas for 1380Hz to 5000Hz range. The system is a transmit and receive system that puts out a bipolar pulse, and measures the Eddy current decay off of a piece of metal that passes between the antennas. I'm looking at what I can do to increase the sensitivity and improve the transmitted energy pulse to increase the eddy current induced in the metal for detection. Any ideas would be greatly appreciated.

 

[Higgler]

Give some more detail, like a test setup.

Sounds unusual from an antenna point of view. Why multi loop antennas. Helical, spiral and quadrafilar are most common at that frequency band.

If you want to concentrate a field at a certain point that you pass the metal between, you can make an elliptical reflector. I've had a 1" spot made at 2450 Mhz before to heat chicken. It used a one foot diameter elliptical and the spot was about one foot from the reflector. If you had back to back elliptical reflectors, that could be what you're asking.

kch

 

[mgburr]

We manufacture Metal Detectors for Industrial applications. I.e. metal traveling down a conveyor belt. We send out a 50us pulse that has a sharp rise and drop from the inductive effect from the wire wrap antenna. Our antennas are 16ga copper and 22ga copper. The transmit antenna has about 17 turns, with a 42inch length and a 6 inch gap between the long legs. so the coil is actually a rectangle. If we change the number of turns it changes the current value in the circuit, which is 3A at 2.1kHz. So I'm looking for a way to re-design the transmit antenna to be able to better predict current/radiated power with frequency change. And for the receive antenna (22ga) about 50 turns we can get about 5% sensitivity(i.e. 5% of the aperature size). 10 inch aperature between transmit and receive and we can see a metal sphere about 1/2 in diameter.

 

[Higgler]

Well, I have a bit of a picture what's going on. What's the width of the conveyor? 12"?

Have you considered making a metal box with open ends (Microwave waveguide) and locating it around your conveyor. Then have antennas radiate inside this box through a slot in the box. When RF energy flow is concentrated in a metal box like this (similar to a microwave oven), field densities increase. Hence I'm hinting changing your environment to concentrate the energy.

Alternatively, if you metallize your conveyor line and have box shaped antennas just above the conveyor, it'd concentrate the energy better too.

These solutions are suggested in part because I'm not familiar with your present antenna design and test setup.

kch

 

[mgburr]

We have some more sensitive designs that use a box around the transmit and receive antennas, but they are operating around 10-14kHz. The conveyors are variable width. The reason I am looking is the antenna designs were created around 40 years ago, but nothing has been done to improve them with the advances towards digital electronics/DSP. Right now we have "Standard" antenna designs, but I'm looking to see if there's a way I can improve the design enough to make the advantages of the Digital systems with DSP. Our website

http://www.tectron.net

 

[zappedagain]

If those antennas were designed about forty years ago for those frequencies I'll bet they are pretty good and you aren't going to do a lot better; most of the antenna work over the last 40 years has been at much higher frequencies.

Putting a DSP to work to add some post processing can probably give you a major return. sub-bit averaging can give you more range, chopping may give you a better noise floor (moves you away from the 1/f noise).

You might be able to find a quieter/more stable/simpler preamplifier than one designed 40 years ago too.

Boy, I bet there was a lot more metal on the Ford than there is on that Dodge!

Z

 

[mgburr]

My biggest frustration though is the varying widths of the conveyors means the number of turns changes to keep the current down and not drive the circuits into overload. When we wind and antenna, we pre-pour check it to make sure it will terminate correctly and also to ensure the current is within limits. If not, then we add or decrease a turn as needed. I'd like to be able to drop a freq, and antenna length into a formula to be able to get the correct current. They are all wound the same way, just varying the number of turns for the transmitter. That would make one step easier, and could be a point to make sure all the antennas are in fact being wound the same. i.e. the same number of turns "should" produce the same current signature at the same frequency, regardless of the electronics attached. If I can prove that then I can work backwards through the circuit and make sure there aren't any "hidden" spots that I can improve the electronics to help with sensitivity.

 

[Higgler]

I just realize you're in the 5 khz band, silly me thinking 5 Mhz.

One technology emerging that will give your receive antenna sensitivity a boost of about ?30 dB is non-Foster matching network. Your short antennas are essentially a capacitor or inductor and OpAmp technology at the antenna output can create an active C or L to resonate with your electrically small antenna. It can make it a tunable C or L to sweep thru frequencies too I believe.
It's more difficult to do for transmit antennas, but works well on receive antennas.
I actually haven't made any, we've discussed it at work, and considered proposing them. I have a few writeups that I researched.
If you go that way I'd be interested in what you find. Maybe I can help out a bit.

Kevin

 

[mgburr]

That sounds like it might help. Currently we "tune" each receive antenna using a resistor network to balance the incoming signal. Active wouldn't be a bad option, and I can always change the input hardware to accomidate for that.

 

[Higgler]

I looked at a few places, googling "non-foster impedance matching", seems frequencies are 10 Mhz or higher for most considerations.
here's the best article I could find, with improvements of 9, 18, 27 dB depending on antenna and frequency. Alot of good theory, but frequencies don't go down as far as yours usually.

http://www.ieee.li/pdf/viewgraphs_matching_network_design_non_foster_impedances.pdf

good luck,
kevin

 

[VEBill]

Although you're using the term 'antenna', to nature it is simply an inductor.

For transmit mode, to increase the magnetic field output you need to increase the amp-turns. The simplest way to do this is probably to add more turns, but at some point the inductance may impede your "high frequency" ;-) signal.

I would design & treat the transmitter as a current source.

For receive mode, the optimum coil might well be different. In general, I think that you can use many more turns.

You don't necessarily need to match impedance for receive mode because the signal-to-noise ratio (SNR) at these frequencies is almost certainly set by the external environment (not the amplifier input).

Careful filtering might be essential.

Another thing that might be possible (depending upon how much time you have) would be to repeat the measurement and synchronously add-up the return signal. If the target is in 'the zone' for ten pulses, then a 10:1 (10dB) SNR improvement should be possible.

Note - all above is based on what I know about LF communications systems. I believe it is applicable, but stand-by for any corrections from others.

 

[biff44]

I have been thru some of this before. It is hard to make "careful filtering" if you are simply driving an inductor. It has such a big reactance compared to the real part impedance, that things like lowpass filters are hard to realize.

Normal filters, like you look up in a table, have an input and output impedance, like 50 and 50 ohms. It is hard if the load impedance is 2 + J300 ohms!

 

[mgburr]

I had a chance to peruse the non-foster filters, It looks like it might work, however as noted the frequency is definitely below most of the designs presented.
The reason for the original question is the documentation for the original antenna designs were incomplete. I.e. It's more of a trial and error to get the right number of windings to the correct current level. I've looked at the basic formula for an inductor, but with the size and seperation of the sides of the windings, it didn't quite come out to what is realistically comming from the circuit(even with worst case performance values). If I have a good starting point for the antennas I can make sure any changes I implement later won't detract from or I can design an active tuning circuit with a "Standard" turn antenna.

VE1BILL, you were close for the Amp turns, except. Reducing the number of turns increases the current. i.e. inductor length reduced along with bundle thickness. I also like the concept of noise canceling filtering. The timing isn't too terrible, however. Not sure I'm willing to do a major board re-design just yet ;-). If I can better guess the L/C/R of the antenna before wrapping I can improve the unit design and might be able to improve the filtering with active design. Although not sure yet.

Hopefully I can find a version of the inductor formula that might be closer to what we have measureable from the antenna/coil/windings. ;-)

 

[VEBill]

Obviously the filtering applies only (hopefully!) to the receive side. The weak receive signal can be buffered and then filtered using all the usual opamp audio-filtering techniques (nothing complicated).

This layer of analog filtering is likely to provide some advantages as compared to going straight into the digital domain (even the best software defined radios would benefit from roofing filters - excuse the 'comms' view).

And with a buffer, the signal source impedance need not affect the filtering.

The only tricky bit would be making sure that the huge transmit pulse and other noise sources don't overload the front end of the receiver. There are receiver techniques to address this, basically using preamps with lots of headroom if they're before the filtering and outside the automatic gain control loop.


With respect to the relationship between the coil and the current, it makes sense that the output current should be right up against the reliable limit of the transmitter. Thus the amps should be defined by the source, and then you design a coil with as many turns as you can without affecting the current.

In general, whenever I see people experimenting with extreme magnetic fields (Google: magnetic 'can crushing' and 'coin shrinking' for some insane examples), they always use just a moderate number of turns on their coil because otherwise the inductance impedes the current.


Sounds like a 'fun' project.

 

[mgburr]

The detection uses timing pulses to enable the "look" for the magnetic decay, so the receive section is turned off during the primary pulse and the kickback pulse. The timing is critical to the detection system. Although it would be nice to increase the number of turns at a rate that is predictable for increasing sensitivity and reducing the noise induced by an incorrect harmonic.

 

[zappedagain]

Here are some handy formulas for inductors that I use. Sorry for the quality. This is a scan of a fax that I've been carrying around since 1995. I'll type the equations because they are a bit difficult to read in the scan:

Single layer

L = (r * N)^2 / (9*r + 10*l)

N = sqrt( L* ((9*r + 10*l)) / r

Multi layer

L = 0.8 * (r * N)^2 / (6*r + 9*l +10*b)

Single layer spiral

L = (r * N)^2 / (8*r + 11*b)

watch out for those lower case "L"; if it doesn't make sense as a one then it must be an "l".

Of course, someone else may have posted these on the web by now...

 

[mgburr]

I was looking at something similar. Those will definitely help with the transmit coils. The receive section is a little harder because of the dual coil setup. i.e.


CCCCCCCCCCCCCCCCCCCCCCCCCC
C C
C C ____
CCCCCCCCCCCCCCCCCCCCCCCCCC /
----------------------------------------C=====<
CCCCCCCCCCCCCCCCCCCCCCCCCC \____
C C
C C
CCCCCCCCCCCCCCCCCCCCCCCCCC

 

[zappedagain]

A HP4195 is a great tool for checking out circuits like this. With a working range from 10 Hz to 500 MHz it easily covers your frequency range. I've seen them on eBay for under $1800 (wow).

 

[mgburr]

Ouch, trying to keep costs down, don't have a lot of funding for the R&D department ;-) hence the raw calculator work :-D But you're right that would definitely make it easier to check the antennas out.