IGSL

[golfnut]

Nice App Jerry.

It shows that, say 20kHz, currents are flowing in the top 0.5mm of wire surface. (That would be both sides')


So easily, in that case, the whole cross section of a 0.5mm wire is carrying ac currents. Full current carrying capacity of the wire - on balance, no skin effect.


But at 2GHz currents are in only the top 1.7 microns of depth.. The reason silver plate is used. Full-on skin effect

Steve

[Jerry]

Here is a calculator for skin effect. Interesting to play around with the numbers.

http://www.calculatoredge.com/electronics/skin%20effect.htm

Jerry

[simonbaker]

HI Simon,

I see, with resonance of an inductor on its own, the self capacitance between turns is the cap.
One reason R seemed high is you may have actually measured a little above resonance ? this way the reactance of the L is getting really big - meybe making it look like R IS BIG who knows?

As for skin effect, it normally only becomes problematic in the hundreds or thousands of MHz - it is normallt a very high frequency problem.

The antennas we make in work can stop conducting due to skin effect (if our copper thickness has gone out of process control and is only say 5um thick) !!!



also Q=wL/R so R=Q/wL

sorry, Zd (impedance at resonance)=L/CR

steve

My instincts agree, I never thought skin effect would come into play either, so you may be right. It could have been something else, like you said. It could have been a sloppy experiment too. On the other hand the proximity effect may apply to a big bundle of wires like we have with our coils (perhaps a reason to "jumble wind" them instead of an orderly layering).

It would be interesting to check again anyway.

Thanks for the handy formulas.

-SB

P.S. I'm trying to find the original post where I calculated a greater R based on Q of coil, and I can't find it, so maybe I'm just confusing it with another case. If I observe it again, I'll make better notes.

[golfnut]

HI Simon,

I see, with resonance of an inductor on its own, the self capacitance between turns is the cap.
One reason R seemed high is you may have actually measured a little above resonance ? this way the reactance of the L is getting really big - meybe making it look like R IS BIG who knows?

As for skin effect, it normally only becomes problematic in the hundreds or thousands of MHz - it is normallt a very high frequency problem.

The antennas we make in work can stop conducting due to skin effect (if our copper thickness has gone out of process control and is only say 5um thick) !!!



also Q=wL/R so R=Q/wL

sorry, Zd (impedance at resonance)=L/CR

steve

[simonbaker]

Thx simon, its all good stuff isnt it,..

I read your discussion,, and you had a parallel resonance queation..



The series resonant case, is easier to visualise, the XL and Xc are equal and opposite sign L is +wL ohms and cap is -1/wC ohms at resonance, the reactances are identical magnitudes but opposite sign or polarity and cancel out, only the R of the coil and small losses in the dielectric of the cap remain. It looks like a short cct.

The parallel case, is a high impedance cct. The cap and L are parallel, the parts shift charge between them - at the frequency where the reactances are equal and opposite.

The charge shift from L to C is so even, and relatively low loss, the circuit is almost self sustaining and does not lose much signal. It is also relatively difficult to get energy into these circuits if they are low loss - or put another way go high Z at resonance.
The Impedance at resonance is wL/R (the R of the L) ohms, as when resonating it doesnt really want any more signal in.

steve

I see your point. Actually though, I wasn't measuring the impedance of the parallel circuit, rather I was estimating the resistance of the coil by trying to measure the Q of the resonant circuit and then use formulas to back out the R component (or something like that).

I expected to get approximately the DC resistance of the coil, but instead calculated an R much higher.

I think my problem was that I used a very high frequency to find the resonant point, because I was using the self-resonance of the coil by itself, rather than with a capacitor as it is in the TGSL oscillator. So my frequency was around 100 kHz I think. I'm thinking that at that frequency, I encountered the "proximity effect" which is similar to the "skin effect", which may have caused the effective resistance to be much higher than the DC resistance. Maybe that was the problem...

I need to redo my test at 14.5 kHz with a coil-capacitor circuit, and measure the Q of the tank, and then estimate the R of the coil and see how close it is to the DC resistance of the coil.

If you can do the experiment first, I'm interested in what you find.

-SB

[golfnut]

Thx simon, its all good stuff isnt it,..

I read your discussion,, and you had a parallel resonance queation..

Does anyone know theory why the coil resistance at resonance is 400% more than the DC value?

-SB

The series resonant case, is easier to visualise, the XL and Xc are equal and opposite sign L is +wL ohms and cap is -1/wC ohms at resonance, the reactances are identical magnitudes but opposite sign or polarity and cancel out, only the R of the coil and small losses in the dielectric of the cap remain. It looks like a short cct.

The parallel case, is a high impedance cct. The cap and L are parallel, the parts shift charge between them - at the frequency where the reactances are equal and opposite.

The charge shift from L to C is so even, and relatively low loss, the circuit is almost self sustaining and does not lose much signal. It is also relatively difficult to get energy into these circuits if they are low loss - or put another way go high Z at resonance.
The Impedance at resonance is wL/R (the R of the L) ohms, as when resonating it doesnt really want any more signal in.

steve

[simonbaker]

Also, has anybody boosted the Tx with a linear amplifer.

Or pulsed a coil using the Tx oxc as a clock source,, a bit like this..

Check the thread where Habitbraker and I were discussing simple circuits for driving the Tx coil by pulsing it with some kind of oscillator.

http://www.geotech1.com/forums/showpost.php?p=128520&postcount=356

My list of unfinished projects includes one attempting to use a Texas Instruments MSP430 microcontroller to generate the oscillator pulse and maybe do some primitive detection, trying to use as little analog components as possible.

Our preliminary tests seemed to show that with a good resonant circuit, the Tx coil could be driven to much higher voltages/currents than with the TGSL oscillator.

However, that doesn't necessarily make it more useful for a metal detector, but seems promising. One of the drawbacks may be that the voltage/current in the coil will be more easily affected by soil conditions and make a less stable reference.

Hope to get back to the project soon. The TI Launchpad is a neat, cheap experimenter board from TI.

-SB

[simonbaker]

Hi Simon, I have a little radio knowledge. Can be a hinderence actually!!

The IGSL mixes the Rx signal (in 4066) using the Tx signal as the L.O.

There are phase delays in the other L.O.s to match anticipated phase shifts from targets.



My thoughts are when any of the 4 mixers - MIX! - there will be sum and difference signals as a result.

To keep all of the IF's (if you like) clean and with good wanted-signal phase dexterity - the unwanted portion of the mixes would ideally be filtered away.

Like an IF filter in a superhet.

Do we do this on these puppies.

S

I agree the Synchronous Detector seems to be essentially an IF mixer with respect to noise signals!

And yes, the circuit after the SD is a very low frequency band-pass filter, centered at approx 8 Hz or so. I would guess that noise signals more than 30 Hz away from the TX frequency are thoroughly rejected. We can look at the actual filter Bode plot to be more exact about that.

-SB

[anz29]

question

Someone has tested Tesoro search coil 4 pin 10x14 DD?
What are your performance in the field?
Thanks

[golfnut]

Also, has anybody boosted the Tx with a linear amplifer.

Or pulsed a coil using the Tx oxc as a clock source,, a bit like this..

[golfnut]

Hi Simon, I have a little radio knowledge. Can be a hinderence actually!!

The IGSL mixes the Rx signal (in 4066) using the Tx signal as the L.O.

There are phase delays in the other L.O.s to match anticipated phase shifts from targets.



My thoughts are when any of the 4 mixers - MIX! - there will be sum and difference signals as a result.

To keep all of the IF's (if you like) clean and with good wanted-signal phase dexterity - the unwanted portion of the mixes would ideally be filtered away.

Like an IF filter in a superhet.

Do we do this on these puppies.

S

[bernte_one]

thx

good explained, looks like a scope session this night
results next day

regards

[hugo]

@Bernte_one:
Maybe you can do that in a different way.
Fix the TX cap at 330nF.
Start with a RX cap value of 100nF.
Connect the first input of an oscilloscope to Tx signal.
Connect the second input of the same oscilloscope to Rx signal.
Adjust the same amplitude and timebase on both oscilloscope channels.
Set GEB and DISC pots to zero.
Now you should be able to see the "initial" phase shift between RX and TX.
Step by step increase the value to the RX cap. until you reach the good shift value (20° for a DD coil ou 200° for a concentric coil).
ie. : 100nF >>> 104.7nF >>> 109.4nF >>> etc ...
Note : During the adjustment, you will see the RX signal increase and decrease around a special point. You shouln't adjust your coil at this point, it will not works properly. The good setting point is a bit before or after it.
Good luck,
Hugo

[bernte_one]

could be my solution...

maybe it is better to calculate by myself
here my result


what do you mean

[golfnut]

Where I left the unit, I mean I put it back on 14.5KHz

S