PCB Coil?

[bugwhiskers]

I like your precise numbers.
Not sure what you mean with flat top at the TX. Do you mean more than 5 TC of the coil, when the current has reached its full potential? I don’t see how you do that with 1 Ohm and 250 uH.
Tinkerer

The coil the test was done on is a 190mm round spiral PCB coil 8.1 ohms 160uH.

I meant flat top on the back EMF, not sure how many of the TC of the coil it is. I just wind up the pulse width until there is no more benefit.

[Andy]

Near Field

Carl is correct about the near field, something I had forgotten about in my earlier post. The wire-wound coil is similar to what is called a small magnetic loop antenna. In the loop's case, the shielding and the gap have a direct effect on the near-field, in addition the amount of current "pumped" through the contained field [inside the shield] determines the intensity or "range" of the field.

If I recall correctly, the older radio books provided information on loops which had the gap opposite of the feed point. In all of the detectors I have seen the gap is at the feed point. I recall seeing references to some sort of capacitance problem if the feed point and the gap were the same. Does anyone know about this? I'll dig through my old books to find this reference [or find out I was wrong in my recollection]. I also seem to recall that the gap size had a direct impact on the near-field [could be wrong there too :-)].

Tinkerer, I was definitely wrong [yup] about the frequency in my earlier post, I had mentioned audio, but forgot about the <10 - 75 usec pulse which is the real workhorse. This puts a PCB solution in-range, albeit at the very end of the range based on current techniques. I have the same question that bbsailor brought up earlier, grounding/shielding effect. Probably the only way to resolve it is to build a coil like the one bugwhiskers has and experiment.

Bugwhiskers, is your coil now one-sided, that is you cleaned off the copper from the other side? What is the width of your spiral and what is the distance between the tracks? Also [you have to love questions] what is the thickness of the pcb material and type? You had mentioned once that you saw an improvement, was that with a coil on two sides or a one-sided one? If it was two-sided, did you by chance have only one side connected?

Thanks,

Andy

[bugwhiskers]

Reply to Andy

Hi Andy,

Of all the PCB coils I have made the ones with wider tracks and gaps work the best, sure they dont have the same inductance but they allow more current and have lower parasitic capacitance. The double sided one I made was the opposite, thin tracks and gaps. I did have it wired up correctly, ie in series with the loops going in same direction. I should have wired it in parallel to see the effect but I was so peeved by the massive capacitance that I ground the other side off the same afternoon.

The next coil I am making will be 285 long, 150 wide 2mm tracks with 1mm between and 22 turns, 10.89 metres length of track, capsule shaped (straight sides, semi circle on the ends) 2oz copper on 1.6mm PCB.

I dont suppose you have a formula to work out the inductance ?
Effectively it is 2 halves of a 150mm diameter round coil with 135mm of straight tracks joining them.

[Tinkerer]

I have a wire wound coil with the same dimensions. I wound it round, meadured the inductance, then bend it into the oval shape. The inductance diminished only slightly. Try with 15 meters of track and 2mm spacing, this gave me good results with low capacitance .
Tinkerer

[Tinkerer]

OK the flat top is from the avalanche diode clipping the flyback. I have not found any advantage in doing that. Among other problems, it gets the mosfet warm. With the mosfets I used, it also generated a burst of HF noise. With 8 Ohms and 160uH the TC of the coil is about 20uS, about 60uS TX will give you close to maximum current.
Tinkerer

[bugwhiskers]

Spacing ?

I have a wire wound coil with the same dimensions. I wound it round, meadured the inductance, then bend it into the oval shape. The inductance diminished only slightly. Try with 15 meters of track and 2mm spacing, this gave me good results with low capacitance .
Tinkerer

Is that 2mm spacing between the tracks or between the middle of each of the tracks ?

I am limited to what fits on an "A4" sheet of paper as I use my postscript laser to make the photo tool.

As for my MOSFET heating, I use a MOSFET driver and it switches in 30 ns so it doesn't even get warm.
The output up to a point keeps increasing as I increase the TX pulse width. When the output stops rising I back the pulse width off a bit.

[bugwhiskers]

MOSFET driver

If anyone is interested the link below has the specs for a very nice MOSFET driver. I have been using the 426 version for months without a problem.
I power it from an 18 volt regulator and the 5 volts ATMEL micro switches it very nicely, pure vertical lines on the CRO.

http://www.microchip.com/stellent/id...cName=en010656

[Tinkerer]

Bugwhiskers,

Thanks for the link. Switching speed is very important

A coil with 30 turns of 0.5mm track, with a spacing of 2 mm between the tracks, with the elongated shape you mentioned earlier, should fit on a sheet. It would have relative high DC resistance, that saves you the series resistor.
My guess is that the coil will have somewhere near 300uH, and low capacitance, that will give a very fast coil with the high resistance. (L/R) By fast coil I mean that with a short Tx time the current will reach near its maximum.
This coil will give high sensitivity but not very great depth.
Tinkerer

[Andy]

Collecting PCB Information

Bugwhiskers,

Thank you for the pcb information, you even included one item I forgot to ask about, the amount of copper. I do not have a readily available equation to produce an answer for the inductance. That is why I asked for the all the pcb information, if I can get enought details about what the/a board really looks like and then what the tested results are I can start to put together a mathematical model of the pcb coil. As I mentioned, all my work was at the daylight-end of the spectrum, this is right up against the dc-end so I have to learn a few things :-). The math part is easy for me, it's the field testing that is time consuming.

It would be interesting to see what other people have done with this type of coil what the details are and what the results were. What is needed is are the following:

type of board material [fr-4, etc.]
amount of copper [1oz, 2oz]
thickness of the board [1.9mm]
thickness of the trace [ 2mm ]
edge-to-edge spacing between traces [ 2mm ]
dimensions of the board [ width, length ]
measured inductance [ uH ]
number of turns on the board or total trace length

As mentioned, I have some learning to do at this lower-end, what applies to microwave is not likely to apply here. It will be interesting to see what effect [if any] the velocity factor has at the lower frequency, in higher frequencies this is about 60-70% on pcb traces [a 10usec pulse is on the border-line of being considered a high frequency signal]. Also almost all formulas assume that there is a ground plane under the trace, in the case of the coils here, you do not want a pcb ground plane under the coil.

Bugwhiskers, have you tried going off to the edges of the coil to see how sensitive it is? [Instead of going under or over the coil] You will need to make checks at different locations around the edge of the coil.

How are you delivering the signal to the coil [rg-58 cable?] and what length is the cable? This may not be relevant, then again it may be, in higher frequencies when the output mosfet gets hot while transmitting, it's usually caused by power being reflected back [ aka an impedance mismatch]. Again this may not be a factor here. I like to observe and collect all information even if it looks irrelevant, just never know...

Andy

[bugwhiskers]

Getting to the heart of PCB coils

Hi Andy,
Thankyou for your interest in PCB coils. I am commited to making them work as they have lots to offer, ease and repeatability of manufacture, very low capacitance, light weight, rugged etc.

I use fibreglass boards 1.6mm thick fr4 ???(Kinsten PCB) The 1oz copper boards have 1.34 mils (thousandths of an inch) copper and the 2 oz double that. I have a variety of boards and I will put the specs for each in a spread sheet for you... that should help as there is a very large spread to work your maths magic.

The high resistance of the PCB coils doesn't bother me as I can up the coil voltage to virtually whatever I like to belt some current into them.

The pure round spiral coils obey the inductance rules that Chemelec has on his web page but the problem with them is they are most sensitive in the centre decreasing to the outside which means a very small area is covered with each swing. The capsule shaped have equal sensitivity along the axis formed by the centre of the semi-circles at each end with a slight increase at those centres. This allows more area to be covered with each swing compared to the circular type.

Tests I have done on the spiral circle shows the field pattern is a sphere centred on the centre of the coil.

regards

Bugwhiskers

[bugwhiskers]

Coil connecting cable

Hi Andy,

As for the connecting cable, I have just started using 300 ohm ribbon cable as used a while back to run the TV antenna to the TV. This cable has a reasonable ammount of copper so as to not increase DC impedance and the conductors are a uniform 7mm apart, hopefully keeping the capacitance low.

The insulation "looks like" Teflon.

regards

Bugwhiskers

[bugwhiskers]

Further thoughts

If anyone is interested the link below has the specs for a very nice MOSFET driver. I have been using the 426 version for months without a problem.
I power it from an 18 volt regulator and the 5 volts ATMEL micro switches it very nicely, pure vertical lines on the CRO.

http://www.microchip.com/stellent/id...cName=en010656

As an aside, if your PIMD needs another voltage, ie -5 for OP amps then the spare gate could be used to generate it.

[bugwhiskers]

Double sided PCB coil

Today I made a double sided PCB coil and it works very well indeed.

Despite one sides tracks being pitted because the CAD program tried to print a "colour" by using a type of hatching/pitting resulting in a 50 % increase in impedance on that side. My old eyes couldn't see it but the UV light sure did.

The tracks are 4 mm apart and the track width 1 mm with the tracks on the opposite side being in the spaces between to minimise capacitance.


Capsule shaped 285*150mm
45mm * 180mm track free area in the middle.
Inductance is 150 uH.
13 Tracks per side 26 total for both sides. Total length of tracks both sides 14.4 metres
2oz Copper 1.6mm PCB
DC Impedance of the "good side" is 2.4 ohms so with a perfect photo tool total impedance 4.8 ohms.

Using an LM318 with 1k of gain the decay curve starts to droop at 5.5 uS allowing a very early sampling indeed.

Using my usual test targets I am getting a signal about 30-50% further out.

[bbsailor]

Coil Self Resonance Guess

bugwhiskers,

Impressive results!

I'll guess that your coil has a self resonance near 3MHz.

What TX pulse width are you using for your tests?

bbsailor

[bugwhiskers]

TX pulse

Hi bbsailor,

The TX is 19.2uS.

I have no way at the moment to test the self inductance resonant frequency as my old Freq Gen only goes up to 100kHz.

I will have to lash me up a 555 circuit.

My inductance meter's lowest range is 20 mH so I only get to see 2 digits for uH.

The early sample has now gone through the roof. In the past I used to adjust the second sample width so the output of it's integrator was the same as the output of the first sample with a ferrous target and that enabled me to discriminate. The ratio used to be about 1 part first sample to 2 parts second sample but now even 35 to 200 doesnt go close. I may have to delay the first sample longer to bring it back into range... or add more compensation capacitance which will reduce noise as an additional bonus.

I once used schottky diodes as pre-amp front end clippers but read somewhere that their capacitance is fairly high so I have switched to 1N914... they saved about 1 us from my observations. In the old days they used to recommend gold bonded germanium diodes for crystal sets, I wonder how they would perform ?


bugwhiskers