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

[Mechanic]

Good stuff Aziz....

This is one of the keys to producing a high performance, quiet, deep seeking detector.....

Cheers Mick

[Aziz]

Hi Aziz,

If possible, at some stage can you get away from all of the emi and do the same test with the rx coil connected. I assume that this test was done with lots of emi around.

When I do my noise tests I look at the output of the final filtering stage. I have to do it with the transmitter turned on to get accurate results. Of course in you case it is different.

Can you also re do the tests with a direct short to ground, a 1k resistor to gnd and a 20k resistor to gnd, just so we can see the difference of different input path impedance, it won't be much but there will be one

Cheers Mick

Yep, the coil measurements were done at home (high EMI environment).

Ok, below is the 1 kOhm and 22 kOhm input resistor at G=101 (input shorted). Note, that the 22k variant is quite noisy and instable (a real eye opener).

Cheers,
Aziz

[Mechanic]

Hi Aziz,

If possible, at some stage can you get away from all of the emi and do the same test with the rx coil connected. I assume that this test was done with lots of emi around.

When I do my noise tests I look at the output of the final filtering stage. I have to do it with the transmitter turned on to get accurate results. Of course in you case it is different.

Can you also re do the tests with a direct short to ground, a 1k resistor to gnd and a 20k resistor to gnd, just so we can see the difference of different input path impedance, it won't be much but there will be one

Cheers Mick

[Aziz]

Hi Aziz,

Which frequencies is the noise measurement for? 0.01 to 10hz or 1k to 150khz? Or does it not matter?

The lt1028 is a big improvement over the others.

One thing worth noting though is when I am away from all of the emi noise out bush, what I hear in my detector is the detector noise floor itself. It is comparable to running the detector with a coil and the input to the preamps connected to coil ground at the coil plug. Having high resistance on the input path to the opamp will cause the noise floor to rise considerably. A 1k resistor makes a big difference to the noise floor compared to using low R fets, about 5 ohms total input resistance.

However using a large coil does cause problems with emi. The cancel coil is a great idea and will solve the noise problems of the larger coils. Even the latest and greatest have problems with larger coils in the right (or is that wrong) conditions. It is very easy to blame EF noise as the symptom is the same, but I firmly believe it is emi levels changing as the coil is rotated causing the false response. At different times of the year this problem does not seem to exist at least not as bad as at other times.

Long live the top hat coil!

Cheers Mick

Hi Mick,

the noise floor level is the white noise spectrum density level for a 1024-point FFT at 96 kHz sampling rate (93.75 Hz bin size). We have a higher 1/f noise at low frequencies of course (I didn't take the 1/f noise into account above).

See below the noise floor level for a shorted input LT1028 in above described configuration. The second picture below is the same with RX coil connected to it.

Yes, LT1028 works indeed very fine. We can use a poor man's low noise design as well with the hand matched BC337-40 transistor pairs and a NE5532 (or the low noise types above). But this would require more parts (there isn't enough PCB space).

Yes, large coils do have big problems with EMI noise. That's the reason, why the "top hat" coil will live long.

"Long live the top hat coil! "

Cheers,
Aziz

PS: Transmitter is disabled for noise measurements of course.

[Mechanic]

Hi Aziz,

Which frequencies is the noise measurement for? 0.01 to 10hz or 1k to 150khz? Or does it not matter?

The lt1028 is a big improvement over the others.

One thing worth noting though is when I am away from all of the emi noise out bush, what I hear in my detector is the detector noise floor itself. It is comparable to running the detector with a coil and the input to the preamps connected to coil ground at the coil plug. Having high resistance on the input path to the opamp will cause the noise floor to rise considerably. A 1k resistor makes a big difference to the noise floor compared to using low R fets, about 5 ohms total input resistance.

However using a large coil does cause problems with emi. The cancel coil is a great idea and will solve the noise problems of the larger coils. Even the latest and greatest have problems with larger coils in the right (or is that wrong) conditions. It is very easy to blame EF noise as the symptom is the same, but I firmly believe it is emi levels changing as the coil is rotated causing the false response. At different times of the year this problem does not seem to exist at least not as bad as at other times.

Long live the top hat coil!

Cheers Mick

[Aziz]

Pre-Amplifier Gain Issue

Hi guys,

ok, I have the final parts values for the pre-amplifiers:
Gain = 100 (enough)
Rin = 470 Ohm
RG = 47 kOhm
Cp = 27 pF - 47 pF (take 39 pF, mandatory now)
RinAux = 22 kOhm

We can't go with Rin below 470 Ohm as it would limit the upper frequency (forming low-pass filter with RX coil inductance and Rin). Note, that Rin is becoming an active damping resistor in the inverting amplifier architecture (the negative op-amp input is regulated to AGND level).

The new dual op-amps do perform better than NE5532. The NJM2068 D is a real bargain and is performing really good (best of the bunch).

Here is the noise floor level with Gain = 100, Rin = 470 and shorted input (measuring amplifier noise):
NE5532P: -98.5 dB (reference base for noise comparison)

NE5532A: -99.5 dB (A-grade, 1 dB better than NE5532P)

µPC 4570C: -100.5 dB (2 dB better than NE5532P)
NJM4580D: -100.5 dB (2 dB better than NE5532P)
MC33068N: -100.5 dB (2 dB better than NE5532P)

NJM2068D: -101.5 dB (3 dB better than NE5532P)

If I plug a RX coil at home (16.5 cm mean diameter, 50 turns, noisy environment) the noise floor level rises approx. 10 dB. So the induced noise is higher than the amplifier noise and the Gain of 100 should be enough.

For quiet anti-interference coils, we would require more gain (500x - 1000x) and we have to change the inverting amplifier into a non-inverting AC amplifier. This is required to cancel the op-amp offset voltage (high gain -> high offset voltage) and to reduce the input resistor (less noise). It really makes sense to use a very low noise op-amp to get more sensitivity in this case.

Just for comparison with a low noise design:
If I use the low noise single op-amp LT1028 in a non-inverting configuration with Rin=47 Ohm, RG=4.7 kOhm, RG2 = 47 Ohm (Gain = 1 + (RG/RG2) = 101), the noise floor level is at -109.dB (10.5 dB better than NE5532P). You see, it makes really sense. If you wanna have more, you have to pay more (expensive low noise op-amps).

Cheers,
Aziz

[Aziz]

How about the ox semen?

[simonbaker]

Hi all,

I'm investigating a very cheap poor man's super glue for fixing the coil bundle.

This is my latest tip:

Ingredients:
- Polystyrene foam (Styrofoam)
- Solvent (Acetone, Turpentine or equivalent solvent)

Acetone dissolves styrofoam very rapidly, but the outcome solution isn't creamy enough for brushing. It dries very quickly too. To get the solution more creamy, one have to leave the mixture for a very very long time (many many months or so).

But if you put some turpentine to the acetone/styrofoam-mixture, it gets more creamy and it slows the drying speed too.

You can also use turpentine alone with the styrofoam too. But turpentine is a bit expensive and one could use a cheap substitute instead (brush cleaner, white spirit, paint thinners, etc.).

Damn it!, we have obviously a glue mafia in this country. Glue (epoxy, resin, etc.) is very expensive here!!!

Aziz

PS: Interesting links from the internet:

http://www.brighthub.com/environment/green-living/articles/20016.aspx

http://lumberjocks.com/mdedm/blog/13468

How about the ox semen?

[Aziz]

Poor Man's Super Glue!

Hi all,

I'm investigating a very cheap poor man's super glue for fixing the coil bundle.

This is my latest tip:

Ingredients:
- Polystyrene foam (Styrofoam)
- Solvent (Acetone, Turpentine or equivalent solvent)

Acetone dissolves styrofoam very rapidly, but the outcome solution isn't creamy enough for brushing. It dries very quickly too. To get the solution more creamy, one have to leave the mixture for a very very long time (many many months or so).

But if you put some turpentine to the acetone/styrofoam-mixture, it gets more creamy and it slows the drying speed too.

You can also use turpentine alone with the styrofoam too. But turpentine is a bit expensive and one could use a cheap substitute instead (brush cleaner, white spirit, paint thinners, etc.).

Damn it!, we have obviously a glue mafia in this country. Glue (epoxy, resin, etc.) is very expensive here!!!

Aziz

PS: Interesting links from the internet:

http://www.brighthub.com/environment/green-living/articles/20016.aspx

http://lumberjocks.com/mdedm/blog/13468

[Aziz]

OK fair enough, its true there are always positives and negatives to being open with ideas. For a peek at your software, I was going to offer you the hand of my first born daughter and 1 gallon of semen from the finest ox in the village, but I see you mind is set.
Midas

Aziz

[Midas]

Sorry Midas,

there is no writeup. Only several postings in some threads. Spread over different forums:
http://www.geotech1.com (here)
http://australianelectronicgoldprosp....com/index.php (requires membership)
(+ old GoldProspectinginOz (Invision Plus)-forum, unfortunately offline now)

To speed up the calculations, I'm using simplified coil models (wire element model). You have to trust in my coil software or you have to check the facts by building the coil prototypes or you have to code your own coil software.

I haven't given/sold the coil software to anybody yet (BTW, there were a lot of interesting offerings ). But this decision was chosen, that any coil related findings & inventions should publicly be shared instead of being locked by the patent trolls. Well, I can guarantee the publication. Even the coil modelling & analysis causes a lot of work, I'm doing it with pleasure, if I see a (reasonable) potential.

Cheers,
Aziz

OK fair enough, its true there are always positives and negatives to being open with ideas. For a peek at your software, I was going to offer you the hand of my first born daughter and 1 gallon of semen from the finest ox in the village, but I see you mind is set.
Midas

[Aziz]

..
Where can I read more about your coil software, have you done a bit of writeup on it somewhere? I'm interested in all the different factors its takes into account.
..

Sorry Midas,

there is no writeup. Only several postings in some threads. Spread over different forums:
http://www.geotech1.com (here)
http://australianelectronicgoldprosp....com/index.php (requires membership)
(+ old GoldProspectinginOz (Invision Plus)-forum, unfortunately offline now)

To speed up the calculations, I'm using simplified coil models (wire element model). You have to trust in my coil software or you have to check the facts by building the coil prototypes or you have to code your own coil software.

I haven't given/sold the coil software to anybody yet (BTW, there were a lot of interesting offerings ). But this decision was chosen, that any coil related findings & inventions should publicly be shared instead of being locked by the patent trolls. Well, I can guarantee the publication. Even the coil modelling & analysis causes a lot of work, I'm doing it with pleasure, if I see a (reasonable) potential.

Cheers,
Aziz

[Midas]

Hi Midas and et.al.,

your suggested concentric co-planar IB-coil is in principle possible. But that is the only good news.

Standard concentric co-planar IB-coil:
TX: 1R
RX/TX-Bucking: 0.8R (4 mm z-distance between them)

Modified concentric co-planar IB-coil:
TX: 0.8R
RX/TX-Bucking: 1R (4 mm z-distance between them)

The modified concentric co-planar IB-coil produces a strong signal at the very near detection distance region (good for pin-pointing). The "benefit" is diminishing quickly and is overtaken by the standard coil variant. At far detection distance regions, the standard coil variant produces twice more target response signal.
So the modified coil variant isn't performing good.

Aziz

Thanks Aziz. That's helped me get my head round the different factors a bit better.
I have another question. Where can I read more about your coil software, have you done a bit of writeup on it somewhere? I'm interested in all the different factors its takes into account.

Cheers
Midas

[Aziz]

Hi all,

I have just got the new parts.

*LOL*
I had luck, that I have got the required 1000µF/25V electrolytic capacitors (~10 mm diameter and <= 20 mm height). They should fit into the PCB.

Now looking forward to test the new op-amps. I'll let you know soon.
Cheers,
Aziz

[Aziz]

Concentric Co-Planar Coil Comparison

Hi Midas and et.al.,

your suggested concentric co-planar IB-coil is in principle possible. But that is the only good news.

Standard concentric co-planar IB-coil:
TX: 1R
RX/TX-Bucking: 0.8R (4 mm z-distance between them)

Modified concentric co-planar IB-coil:
TX: 0.8R
RX/TX-Bucking: 1R (4 mm z-distance between them)

The modified concentric co-planar IB-coil produces a strong signal at the very near detection distance region (good for pin-pointing). The "benefit" is diminishing quickly and is overtaken by the standard coil variant. At far detection distance regions, the standard coil variant produces twice more target response signal.
So the modified coil variant isn't performing good.

Aziz