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

[simonbaker]

Hi simonbaker,



I must have missed to mention this of course. You are right. A low ESR capacitor bank will hold the energy for the coil pulses. The energy will be charged synchronously before the TX pulse starts. It is decoupled during the TX pulse from the battery voltage.

It has huge benefits, if the coil voltage is decoupled from the source voltage (I might use a SMPS). The drawback is: I need high capacitance to hold enough energy to minimize the voltage diminish. And there must be enough time to charge up the lost energy from the previous TX pulse.

Some could argue to charge the capacitors directly from the source voltage. But the high current during the charge time will end in radiating high EMI and the power losses (P=I²*R) would end in a simple linear regulator (very inefficient, even worser than linear regulator, see "two capacitor problem"). That's the reason, why I have to use a energy storing element via inductor (buck converter), which reduces the high current flow instantly.


Aziz

Ok. I agree that a buck or boost converter should be more efficient. I guess EMI maybe less because the buck/boost coil will slow the initial current pulse. Question is whether you can charge the cap as fast. Probably if done right.

Do you really need a whole SMPS, or would just a coil between the battery and the capacitor, chosen correctly, do the job pretty well with a single pulse?

Now if you could only recover the energy from the TX pulse...

Regards,

-SB

[Aziz]

Hi simonbaker,

Interested in what you are saying, but don't understand exactly.

You want controllable voltage to make coil pulse? But pulse causes voltage drop across battery or regulator outpu due to internal resistance?

Would any regulator be fast enough to work during pulse?

Seems like big capacitor is only answer, so -- don't quite understand the problem. Just interested...

Regards,

-SB

I must have missed to mention this of course. You are right. A low ESR capacitor bank will hold the energy for the coil pulses. The energy will be charged synchronously before the TX pulse starts. It is decoupled during the TX pulse from the battery voltage.

It has huge benefits, if the coil voltage is decoupled from the source voltage (I might use a SMPS). The drawback is: I need high capacitance to hold enough energy to minimize the voltage diminish. And there must be enough time to charge up the lost energy from the previous TX pulse.

Some could argue to charge the capacitors directly from the source voltage. But the high current during the charge time will end in radiating high EMI and the power losses (P=I²*R) would end in a simple linear regulator (very inefficient, even worser than linear regulator, see "two capacitor problem"). That's the reason, why I have to use a energy storing element via inductor (buck converter), which reduces the high current flow instantly.


Aziz

[simonbaker]

Hi all,

efficient power conversion causes problems. I am not happy with the coil power regulator yet.

Background:
Battery voltage diminishes or is loaded with other electronic sections and causes voltage drifts. It even may induce voltage on the power lines during the TX pulse or other EMI sources and may infect the source voltage (like SMPS's do). During the TX pulse, the emitted magnetic field energy is dependent on the coil source voltage.
I thought of using a regulator, which could stabilise the coil voltage to some accuracy to keep the TX pulse constant. Of course, temperature drift is another issue.
Using a linear regulator makes no sense. It isn't efficient enough. I also want to use the higher battery voltage to make a longer operating duration (power conversion).

I have designed a modified buck converter (step-down converter) but due to energy storing element (inductor) and switching parasitics, the power supply rejection ratio is only 48 dB (40 mV regulated variance at 10V variance of battery voltage). This isn't accurate enough. I want to have fractions of mV only.

Fortunately, the theoretical efficiency is between 90 and 92% (spice simulation results, magnetic core losses and EMI radiation not taken into account).

I will try a post regulation (shunt regulator), which makes the circuit complexity higher and reduces the efficiency slightly. Only the few milli Volt variances need to be burned into heat.

Oh, the +5V/-5V linear regulation: Ok, still buring lots of energy. But the coil power dominates here. I want to switch up to 20 or more Ampere through the coil.

Aziz

Interested in what you are saying, but don't understand exactly.

You want controllable voltage to make coil pulse? But pulse causes voltage drop across battery or regulator outpu due to internal resistance?

Would any regulator be fast enough to work during pulse?

Seems like big capacitor is only answer, so -- don't quite understand the problem. Just interested...

Regards,

-SB

[Aziz]

Problems, more Problems ...

Hi all,

efficient power conversion causes problems. I am not happy with the coil power regulator yet.

Background:
Battery voltage diminishes or is loaded with other electronic sections and causes voltage drifts. It even may induce voltage on the power lines during the TX pulse or other EMI sources and may infect the source voltage (like SMPS's do). During the TX pulse, the emitted magnetic field energy is dependent on the coil source voltage.
I thought of using a regulator, which could stabilise the coil voltage to some accuracy to keep the TX pulse constant. Of course, temperature drift is another issue.
Using a linear regulator makes no sense. It isn't efficient enough. I also want to use the higher battery voltage to make a longer operating duration (power conversion).

I have designed a modified buck converter (step-down converter) but due to energy storing element (inductor) and switching parasitics, the power supply rejection ratio is only 48 dB (40 mV regulated variance at 10V variance of battery voltage). This isn't accurate enough. I want to have fractions of mV only.

Fortunately, the theoretical efficiency is between 90 and 92% (spice simulation results, magnetic core losses and EMI radiation not taken into account).

I will try a post regulation (shunt regulator), which makes the circuit complexity higher and reduces the efficiency slightly. Only the few milli Volt variances need to be burned into heat.

Oh, the +5V/-5V linear regulation: Ok, still buring lots of energy. But the coil power dominates here. I want to switch up to 20 or more Ampere through the coil.

Aziz

[Aziz]

A Cuk Converter

Hi all,

I have found a very interesting SMPS configuration: A Cuk Converter.

Damn good solution (I have spiced this already). With synchronous rectification, the efficiency can go up to 96%. Step-Up/Down with low ripple noise and constant load to source (very low ripple).

It even makes sense to use negative coil voltage (grounded coil with N-ch mosfet configuration! Seems to be very very attractive to me now!).

An alternate is a Sepic converter.

Let's look at available controller chips (for external mosfet usage). It must have a synch input pin. If I can not find an available part, I have to make a µC or discrete solution.

Aziz

[Aziz]

Damn! I have to still wait for more other parts. This time: ferrite cores.

It seems, I can achieve at least 90% power conversion efficiency of the coil driver. The objective is:
- wide input voltage range (10 .. 28V)
- software adjustable coil voltage for operation
- high power efficiency
- low noise coil supply voltage (synchronized)

It's going to be a real challenge to achive this. Not happy with the power supply section yet.

Aziz

[BigBang]

Hi all,

I now got the new parts from the latest order. It should make some progress possible now.

I am on the way to supervised visits with my two kids for two hours. I didn't see them for a long period yet. You know, that's the reason, why I do not work for the german government anymore.

And the bailiff is coming tomorrow. They will try to rob me once more.


Aziz

Harika bir buluşma olduğunu 2 çocuk babası olarak tahmin edebiliyorum . Gözünaydın kardeşim Kokularını ciğerlerine ve zihnine doldur mümkün oldukça

[Aziz]

Hi all,

I now got the new parts from the latest order. It should make some progress possible now.

I am on the way to supervised visits with my two kids for two hours. I didn't see them for a long period yet. You know, that's the reason, why I do not work for the german government anymore.

And the bailiff is coming tomorrow. They will try to rob me once more.


Aziz

[Mechanic]

Hi Aziz,

Good stuff mate!

There's always more parts to order

I have considered making some litz wire but what a task!!

Keep plodding along.....

Cheers Mick

[Aziz]

Good News!

Hello friends,

I have some good news today. The important missing parts are now in stock and I just ordered them. They should arrive at the end of the week.

But the part's list is still not complete. I am missing following parts:
- Lot's of 0.2 mm enamelled copper wire for the coil. I am going to make my own Litz wire and coil leads.
- J-Fet multiplexer chip
- VCA chip

Damn! I forgot some more parts to order. Too late for this month.

Anyway, I will make some progress on the PI controller soon.

Aziz

[Aziz]

Project update:

I got some parts today but still missing some very important parts yet. It is so much frustrating at the moment. I am going into hibernation mode.

But I finally got the 50 Ampere connector! Let's vaporize the coil.

Aziz

PS:
r_olivarez:
I have been working with PDA's (HP). No, PDA is not a good platform. The Netbooks are the best and cheapest platforms at the moment.

[r_olivarez]

Hi,

you should decide, which type of detector you want to develop.
1. PI
The AD converter of the mentioned PIC is very likely not fast enough for PI systems. PI systems need at least 1 MSPS (Mega samples per second, 1 µs resolution). Using Flash-A/D would be better but expensive.

2. VLF
For VLF you do not need such a PIC. Just use the stereo input and output of the Laptop. You will have at least 16 Bit at 48 kHz or even with the common new Laptops 24 Bit at 96 kHz sampling rate.
10 bit A/D: SNR ca. -30 dB (in the PIC)
24 bit A/D: SNR ca. -72 dB
SNR increase on 24 bit A/D: 42 dB!!!!

Forget any suggestions with any PDA's. You will have much more problems with it (tools, compilers, specs., etc.).

Implementing the code on Laptop would be very fast and easy. You will have enough number crunching power for this purpose.

Actually a PDA is a better choice because if it's light weight and computing power. Laptops are heavy, large and overkill on processing power. A PDA is light enough that it can be mounted on a neck support or on the detector itself somehow.

'C' is a common language for programming PDAs and there are a few programs that will translate 'BASIC' to 'C' if you only know 'BASIC'. I would think it wise to included older PDAs that use the RS-232 and Parallel ports as well. That way, we could use our older PDAs for keeping logs of our finds as well as aiding in our searches.

Personally I do like the idea of using a PDA with and as a part of a metal detecting system. I do hope that others who are knowledgeable about both, pick up on the idea and give it a go. I would be interested in showcasing it on a web site.

R. Olivarez
[email protected]

[Aziz]

humanise.org

Hi all,

I have given the permission to humanise.org to publish some work from here. You should know, that there won't be any copyright issue then. Humanise.org is allowed to publish whole/parts of my published work here (for copyrighted materials).


Project update:
VLF mono version: just waiting for better weather to make field tests. Brrrrrrrrr, ist very cold here again.

PI controller version:
Project frozen at the moment. No progress yet. I have saved some money for the new parts. I probably will place an order soon.

Aziz

[Aziz]

VLF2.0 Schematics

Hi all,

this easy design makes true mono coils possible. Below is a differential mono coil for better performance. It shows only the principle and must be optimized further of course.

Aziz

[simonbaker]

Hi Simonbaker,



Amplitude and Phase (Frequency shifts)

Probably on three frequencies around the resonance frequency:
Fr-BW/2, Fr, Fr+BW/2,
where Fr = resonance frequency and BW = bandwidth (few 100 Hz).

But not decided yet. I will see this, if I made a field test.

BTW, I can increase the coil voltage up to +/-3V (6Vpp) until it gets clipped by the sound card. Using differential mode, I can process up to +/- 6V (12 Vpp) coil voltage. It is interesting, that very small changes in the amplitude and phase shift are sufficient enough to make this detector with mono coil possible. The current hardware consumes only 4.25 mA at 9V. So a 9V battery block can last for a long period.

The new coil is finished but not shielded yet. I have taken 5 strands of magnet wire, each 0.3 mm diameter. So the finished coil has ca. 20,6 cm mean diameter, 12 mm bundle thickness, 2x37 turns (74 turns total), 2.2 mH inductivity and a resistance of 1.8 Ohm total. It has a center tap and is operated in fully differential mode (differential stimulation, differential sensing). The difference is made digitally in the software of course.

Aziz

It sounds very interesting and would sure simplify coil building.

-SB