Let's made a PC-base metal detector with usb interface !!!

[Aziz]

Why do you use 1MHz instead of say 10kHz?

Eddy currents are not the same at 1MHz as at 10kHz, and skin effect can be observed at higher freqency too.

The calculated results would be quite small. I am increasing only the dI/dt (high frequency) as a calculation trick to elevate the numbers to have more digits.
I am not taking skin effects into account. So it doesn't have an effect to the numbers.

[Aziz]

Interpretation of the Results

Hi all,

we can see, that a mono coil is performing better for the same given inductivity. The other coil types can compensate the performance losses with:

1. Higher amplifier gain:

But, but: It will amplify the induced EMI too. The amplifier will add amplifier noise. The signal to noise ratio won't increase.

2. Higher amplification through higher RX inductivity:

But, but: It will amplify the induced EMI too. Only the amplifier noise won't be added as the coil resistance increase does not contribute much resistor thermal noise.

Indeed, the induction balanced coils like the DD or concentric co-planar coils have a higher inductivity receive coil to compensate the performance losses. If we need a gain of 2, than we have to double the number of RX coil turns, which result in four times higher inductivity. In the comparison we need a gain of 2-3 to compensate the performance losses. But we would also increase the external EMI induction.

You see the benefit of a pure mono coil in a pulse induction type metal detector. It has maximum performance with less EMI induction compared to the other coil types.

I probably will add an another coil type into the comparison soon:
The anti-interference coil (so called figure-8 type coil).

I have to make lot's of target sweep calculations and have to take the maximum response of all calculations (thanks to excel, which makes it easy). Unfortunately, the target response is dependent of target orientation. This should not take a much effect but it will show, how the figure-8 coil performs. It's not going deep. You will see the results soon.

Aziz

[Tinkerer]

Hi Aziz,

thanks for the excellent work.

I have one question:

with the TX coil of 10" diameter and the Bucking coil of 5" diameter, how many turns of the bucking coil do you use for your calculations?

When I use 20 turns TX and 5 turns Bucking, I can get perfect balance, about a proportion of 300uH x 6uH.

The interaction of the 3 coils is very complex, so you probably use a simplified formula for the calculations?

Tinkerer

[Aziz]

Hi Aziz,

thanks for the excellent work.

I have one question:

with the TX coil of 10" diameter and the Bucking coil of 5" diameter, how many turns of the bucking coil do you use for your calculations?

When I use 20 turns TX and 5 turns Bucking, I can get perfect balance, about a proportion of 300uH x 6uH.

The interaction of the 3 coils is very complex, so you probably use a simplified formula for the calculations?

Tinkerer

Hi Tinkerer,

as I am using a single loop simplified coil model, the coil is more compact (no bundle thickness, the wire thickness of 1 mm only) and I have for the TX: 18.417 turns, for the Bucking coil: 5.636 turns. If I would make a more finer coil model (with bundle thickness), I would have many many wire segments and this would extend the calculation time heavily (and additionally drive me crazy in waiting for the results).
It is ok using simple coil models, which does not really affect the performance comparison.
Of course, fractions of number of turns makes not sense. One would take integer number turns and adapt the diameter of the bucking coil or the distance gap of the RX coil. For my calculations, using fractional number of turns count is ok.

No, I am not using a simplified formula. I am using the full Biot-Savart model to calculate everything.

-----------------------------

I am close to finish the figure-8 coil. Don't be much disappointed. They are the worst coil types and are not going much deep. Stay tuned...

Aziz

[Aziz]

Added Anti-Interference (Figure- Coil

Below is the new added coil type:
- Scheme of the Anti-Interference (figure-8 ) coil
- Coil Comparison of Response Values
- Coil Comparion (Log-View)

So, it's not going deep. But I have a novel AI coil, which will outperform the mono coil due to effectively no EMI and compensation with less gain only.

Cheers,

Aziz

[Tinkerer]

These comparisons are very interesting.

For some reason, my measured numbers do not coincide. What can be the reason. Below are some of my numbers. They are not of great precision, they were taken of the scope inside with a lot of 60Hz noise.

The numbers were taken at the output of the amp, to define the dynamic range.

The differences are very great.

Dynamic range:

Target 1 Euro coin at a distance of 70mm from the coil = 5V
1 Euro coin at a distance of 155mm from the coil=1.25V
1 Euro coin at a distance of 500mm from the coil=5mV

The coil diameter is about 500mm, the RX and bucking coils are 250mm.
About 300uH inductance, RX+TX coils.

Tinkerer

[Aziz]

Hi Tinkerer,

the 1 Euro coin gives a stronger signal response than the 20 mm gold wire ring. Then, the coil current may differ. Then dI/dt may differ. A lot of other factors can also be responsible.

The target (coil) has an inductive reactance XL, which don't allow to increase my target eddy current at 1 MHz. That should be responsible for different responses.

As the comparison is made under same conditions, the comparison is valid.

Aziz

[Tinkerer]

Hi Tinkerer,

the 1 Euro coin gives a stronger signal response than the 20 mm gold wire ring. Then, the coil current may differ. Then dI/dt may differ. A lot of other factors can also be responsible.

The target (coil) has an inductive reactance XL, which don't allow to increase my target eddy current at 1 MHz. That should be responsible for different responses.

As the comparison is made under same conditions, the comparison is valid.

Aziz

Aziz,

I do not doubt on the validity of your comparisons. Intuitively they look very good to me, mathematically I am not competent to give an opinion.
What I am trying to do, is to learn and understand the reason of the differences between your results and the results that I got.

So let's look at the differences and possible causes for the difference in response.

My setup runs at 30kHz, yours at 1MHz.

At a frequency of 30kHz, there is already some skin effect response. My setup measures the total response, while your calculations do not take into account the skin effect.
This could account for some considerable differences in the amplitude of the response.

Could the amount of skin effect be blamed for a factor of 4 to 10 in the amplitude of the response?

You use a mathematical gold ring, I use a 1 Euro coin.
A 1 Euro coin is really a bad choice for a test target. It is made of a combination of 2 different alloys that are unevenly distributed in the volume of the target. Weird things can happen to eddy currents in such an environment. However, I chose this target as an easily accessible test target in the European region.

You use 1A coil current, I use maybe 5A coil current.
This will obviously generate a much stronger field and response. But, will it also produce a different diminishing of the response for a given distance?

Your coil is 10" diameter, my coil is 20" diameter.
My coil should go much deeper, but, should there be a big difference between the relative responses from a distance of 1r or 2r target distance?


There is a little bit of non-linearity in your response graph for the concentric IB coil. In my numbers the non-linearity seems to be much more pronounced. In part this could be explained with the lack of precision of my numbers. On the other hand one could suspect the bi-metal alloy of the 1 euro coin or/and frequency susceptibility.

I really appreciate your mathematical work. It will give me a sleepless night, until I unravel and understand all the questions it confronts me with.

All the best

Tinkerer

[Aziz]

Hi Tinkerer,

for a more accurate model, one have to make a spice model. So time-domain effects can be seen. The induced target response then can be calculated much better. Remember, it's a complex multi-order differential equation.

Indeed, my coil model is a very dumb and simple. Even there should be a bug in the software, the comparison won't suffer much. So don't focus on the absolute values now. They are meaningless. You have to look at the relation to the other coil types only.

BTW, the log-view is a logarithmic display!
Aziz

[Aziz]

Hi Tinkerer,

it could well be a bug in the software (don't know yet). But it is very likely not affecting the comparison. I will go through the complex code and will check all again. Fortunately, spice model helps much to localise any bugs.

Aziz

[Aziz]

Ahhhhhrrrrrggg! Found the guilty one!

This happens, when making some quick changes for experiments.

E2.iExciteType = EXCITE_CURRENT;
//E2.fExciteMag  = CCTXTG.fUind / CCTG.fZL;
E2.fExciteMag  = CalcTargetCurrent(CCTXTG.fUind, CCTG.fZL, CCTG.fTC, pElectric->Frequenz);

I have tried something in the function CalcTargetCurrent() and forget to undo it.
The target current is calculated E2.fExciteMag = CCTXTG.fUind / CCTG.fZL, which is commented above.

Now I get reasonable voltages and they matches spice simulation results.
I bet, the comparison is still the same. But the voltage levels should be now higher.
Ok, let's make everything new and put the data into the tables.

Thanks Tinkerer, this was a good one.
Aziz

[Tinkerer]

Ahhhhhrrrrrggg! Found the guilty one!

This happens, when making some quick changes for experiments.

E2.iExciteType = EXCITE_CURRENT;
//E2.fExciteMag  = CCTXTG.fUind / CCTG.fZL;
E2.fExciteMag  = CalcTargetCurrent(CCTXTG.fUind, CCTG.fZL, CCTG.fTC, pElectric->Frequenz);

I have tried something in the function CalcTargetCurrent() and forget to undo it.
The target current is calculated E2.fExciteMag = CCTXTG.fUind / CCTG.fZL, which is commented above.

Now I get reasonable voltages and they matches spice simulation results.
I bet, the comparison is still the same. But the voltage levels should be now higher.
Ok, let's make everything new and put the data into the tables.

Thanks Tinkerer, this was a good one.
Aziz

I will sleep better now.

I think comparisons is good.
The concentric IB coil model that you use is realistic for an average coil.
On the concentric IB you could change the proportion of the TX-BU coil closer to 300uH x 6uH to represent a better quality IB coil. This will result in a slightly better performance.

There is another IB coil arrangement that could be very interesting to check out. I will try to draw a picture of it. It might be a very good performer.

Tinkerer

[Aziz]

Corrected Calculations

Finally, everything is calculated once more. This time more accurate.
The error was just a constant factor in the target current calculation. So it has no effect on the comparison. It's just a y-shift in the log-view. The values should be reasonable now. You can look at it if you want this time.

Let's make better coil designs with the corrected software version.

Aziz

[WM6]

Thanks Aziz, looks more convincing now.

[Aziz]

Anti-Interference Coil Thoughts

Hi all,

consider the following:

Interference cancelling coil has the potential for up to 30 dB external EMI noise reduction (this is the most limitting factor for a detector). This range could be used to amplify the signal for potential performance losses. 30 dB is a gain of 10^(30/20)= 31.62. So this would not contribute much amplifier noise.

My novel AI coil would outperform the mono loop coil . It will be a super deep and super quiet detector . Better I don't show you the performance comparision graph . You won't believe it. This could give a real breakthrough in detector performance.

Will follow it further.

Aziz