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

[Aziz]

Initial Decisions for the New Netbook IB-PI Project.

Hi all,

let's start planning and making decisions for the new netbook/laptop based IB-PI project. I do not want to make the hardware part too much complex. It will be a cheap solution but still powerfull implementation of course.

Power supply:
Battery: 10-30V (12V nominal)
Battery polarity protection: simple schottky diode + fuse polarity protection with low losses.

The electronics part will draw less current (<50 mA). Simple 78xx/79xx should work well. Battery voltage will be first regulated to +8V (using LM7808CT, 1A). A 78L05 will provide +5V for analog and digital parts. +8V will be inverted into -7V and regulated into -5V (79L05). +8V will be doubled into +15V and then regulated to +12V for the mosfet driver. The DC/DC converters will be synchronised to the pulse clock coming from the netbook.
The +8V needs to be a high current one. If someone can get a LDO type, the minimum battery voltage can go down to 9V. On higher input voltages, the 7808 needs probably a heat sink. That's the reason, why it has to support more current and should provide heat sinking.

I will very likely use a mosfet driver for the DC/DC charge pumps to reduce the parts count. One ICL7667 fits perfectly in this design and offers two of them.

The RC oscillator will be implemented using a schmitt-trigger inverter. It will provide a synchronisation. The solution is already shown here.

Maybe a simple low battery voltage detection circuit will be added.

That's all for the power supply part.

Clock/Synch generation:

Two external line signals coming from the sound card will clock the hardware board. It's implementation is trivial and a similar solution is already shown here.
One line input signal will trigger the synchronisation of the DC/DC converters. The phase shift between the external signals will define the pulse width. The input frequency the pulse frequency (doubled frequency).

The RX amplifier/RX coil:

The RX coil must be a center-tapped coil. The center-tap will be connected to ground and will allow a true differential usage. I will use both input lines of the sound card to increase the dynamic range by 6 dB. This will reduce additional noise on cables too. The RX amplifier will be a "half instrumentation amplifier" implementation, which provides differential signal output. It's gain can be adjusted by a simple resistor. A dual op-amp (NE5532 or a better one) will do this perfectly. There is no need for a true instrumentation amplifier (expensive). The gain should be between 10 .. 100.

Mosfet driver/Coil driver:

A cheap solution will be implemented using a complete IC. It will be driven with +12V. I don't know, whether an ICL7667 (inverted) or MCP1407 (non-inverted) fits into the design (but I will very likely make a design with the ICL7667).

Mosfet: A low on-resistance 200V/300V n-channel mosfet should be enough. Flyback voltage should be limitted to avoid avalanching of the mosfet. I generally tend to limit the flyback voltage between 100V and 150V. Low on-resistance mosfets increase the power efficiency of course.

With some additional current sensing circuit, the efficiency can be increased further. But it increases the circuit complexity a bit and there is a 10 mOhm shunt resistor necessary, which should be difficult to build. So I will omit this feature to keep things easy.

The transmit coil: Simple single ended and shielded coil with low resistance. Something in the range of 300µH to 1mH or more (it depends on the application and timing). The coil will be driven with the battery voltage without regulation. The battery voltage diminish should cause a drift, which can be compensated in the software automatically. Maybe the +8V/1A could also provide a low power regulated supply. As the transmitter is very power efficient, it should deliver enough current. For all, who like more power in the transmit coil, just take the battery voltage.

Parasitic capacitances of coil, coil shielding and mosfet does not matter. There is no need for a low capacitance coil solution. The only critical part is the induction balance stability. A simple adjustment possibility should make it perfect.

That's all for initial decisions.

Aziz

[Aziz]

Front-end Receiver of IB-PI finished

Hi friends,

now the front-end receiver of the IB-PI project is finished and tested. I have made two configurations and tested them. There are two options:
- cheap implementation using the NE5532
- fine and quiet implementation using the INA163

The front-end receiver has a RFI-filter to keep radio frequency interference out of the receiver. The cheap NE5532 implementation achieves a noise spectrum density of 9.6 nV/sqrt(Hz) at gain of 100. The INA163 1.6 nV/sqrt(Hz). The latter one offers a true differential output and a single ended output. The single ended output will not be used. As I mentioned before, it is only a half instrumentation amplifier implementation with band limitting feature (low pass cut-off frequency: ~46 kHz).

Processing the input signal in differential mode has some benefits:
- keeping common mode noise out of the circuit/ADC converter (sound card)
- increasing the dynamic range (input voltage range)

I have choosen to use a high current rate protection diodes to keep the input resistance very low (very important for low noise operation). The additional input capacitance does not matter as the RFI-filter needs the differential capacitance anyway. The RFI-filter uses a hand made common mode chocke.

Well, the 1 EUR coin can still be detected up to 40 cm distance (NE5532, gain 98, air-test). I am sure, the detection distance can be exceeded, if the induction balance coil stability can be held and the low noise version be used.

As I have already shown in the earlier post's, the coil itself is picking up the most noise in the system. The amplifier noise is not dominating here. But the software can decode the signals almost at its noise density level, which gives superior sensitivity and detection depth. I hope, the ground and other effects will not reduce the performance too much.

I will publish the front-end receiver with LTspice-files soon, which will show its behaviour. You can make different analysis (transient, AC analysis, noise analysis). Stay tuned..

Aziz

[Aziz]

Laptop IB-PI Front-end Receiver Design

Hi all,

below is the analysis and design model of the receiver front-end. Just install LTspice and run the simulation. There are other analysis commands, which have to be activated first. Just comment the .tran command and uncomment the other command (only one allowed). Press Ctrl-Key & Left-click mouse on the command text to change it between comment and spice directive.

Aziz

[Aziz]

Just some comments:

The receive coil is not shown in the circuit above. The coils L1a/L1b and L2a/L2b are the common mode chokes only.

If you want to use the 1N4148 protection diodes, Rin must be used (100 Ohm at least). The NE5532 noise is dominating in conjunction with higher input impedance. To be save and not to damage the input amplifier, the Rin should even be higher (470 Ohm).

The UF4002 can withstand higher current and therefore the input impedance can be kept very small. The INA163 needs low input impedance to give its superb performance anyway.

Aziz

[Aziz]

INA163 is making trouble.

Hi all,

the INA163 is causing some problems. The low pass filter with the feedback capacitors Cfb2 are causing some trouble and do not work in this case. Just drop them in conjunction with the INA163. But the dual op-amp version is running well. I did not find the reason, why the INA163 makes the trouble.
The low pass filter can even be omitted (Cfb1, Cfb2 capacitors) as there is (and should be) an anti-aliasing filter in the sound card already.

On the other way, the INA163 makes a really very low noise amplifier system possible. The benefits are marginal, as the noise picked up from the coil is much larger and dominating. Even the NE5532 is working remarkable well.

Coil Noise (EMI):
It is much larger than the amplifier/ADC system performance and it is quite over dimensioned. This means, that we can measure more accurately but the signal is buried in the large noise. An alternate could be an anti-interference coil, which reduces the noise level up to 40-30 dB. It is a typical figure-8 RX coil, which cancels a large amount of noise. I have tried to make such an IB coil but remarked that it's quite difficult to balance it.

Well, if the coil noise can not be reduced significantly, we have to limit the signal bandwidth. Purely reducing the noise level to the noise density spectrum level. Lock-in amplifier and Goertzel algorithm should be the most efficient ones to get the best performance. FFT is wasting processing time and battery power for lots of unused spectrum response calculations. We just need discrete frequency responses of f0, 2*f0, .. n*f0.

To be continueed..

Aziz

[Aziz]

First Cut of Laptop IB-PI Hardware.

Hi,

below is the first cut of the powerful Laptop IB-PI detector. So you have an idea of the sections. I did not find spice models of some mosfet drivers (ICL7667) so a discrete and simple mosfet driver is shown. Also a comparable Linear Technology mosfet driver is used to get the circuit simulation running only.
You very likely will not see much details in the picture. Its not intended yet.

I will focus on the different sections later. A full running spice model will also be provided.

Aziz

[Aziz]

Transmit Coil Driver Schematics

Hi friends,

I will show you the most beatiful and genius transmit coil driver part. But do not forget to thank Tinkerer.

If it is driven in continious mode, it achieves high power efficiency. In my prototype, the current drain for the coil driver is only max. 200 mA at 12V. But it still provides coil current up to 3A.

Forget all the pulse logic generation and mosfet driver part. Look at the transmit coil driver part only.

Who recognizes the essentials of the transmit part? And why it is so much simple and efficient? WTF, where is the TX damping resistor gone?
Discussion and analysis will follow later...

Aziz

[WM6]

Hi friends,

And why it is so much simple and efficient?

Probably of coil energy buffer capacitor in resonant circuit with coil, so do not need to be damped with resistor?

[Aziz]

Current Phases of the Transmitter

Hi friends,

to get an idea, how the transmitter works, just see the graphics below. It is divided into four phases. If the mosfet is switched during the recycling phase on again, the power efficiency increases further. The voltage drop of the internal body diode can be omitted and there is only a voltage drop of the mosfet on resistor.

This one is the most efficient pulse induction transmitter I have ever seen. But it is not a resonant circuit.

More infos about this transmitter will follow. Stay tuned..

Aziz

[Aziz]

Hi,

did you all thank to Tinkerer??? Just do it.

We have to distinguish between two operating modes:
1. Discontinious Mode
2. Continious Mode

Discontinious Mode:
It happens, if the recycling phase finishes before the next transmit begins. As the coil is not damped anymore, the buffer capacitor has stored some energy, which will start to oscillate in the coil (ringing). Losses will cause a damped ringing. You could switch on a damping resistor to avoid ringing if you like. Or you could let the coil ring.
But if you place a damping resistor to the coil, the efficiency suffers heavily.

Continious Mode:
It happens, if the next transmit begins, before the recycling phase is finished. If you switch on the mosfet just after the recycling phase begins and enlarge the transmit pulse by the duration of the recycling phase, then optimal power efficiency can be achieved.


You will observe a half sine high coil flyback voltage during phase 2 and 3. The duration of the phase 2+3 is defined by:
thalfsine = 0.5*period of LC resonant circuit
fres = 1/(2*PI*sqrt(LC))
Tperiod = 1/fres

thalfsine = PI*sqrt(LC),
where L = inductivity of the transmit coil
C = sum of all capacitances seen by the coil (CEB+parasitic capacitances)

During the phase 2 and 3 the mosfet may not be switched on or avalanched. If the mosfet avalanches, a lot of the coil energy will be lost.

To get a good working continious mode, just increase the transmitter pulse width such that:
PW <= TP + HS
where TP = transmit period time (TP = 1/(Pulse Frequency))
HS = PI*sqrt(LC) (half sine flyback duration time)

It is more efficiently, if PW = TP + HS can be achieved. In this case, the mosfet switches on during the recycling phase.

Aziz

[Aziz]

Signals during continious mode operation:

[WM6]

Hi,

did you all thank to Tinkerer??? Just do it.

Dear Tinkerer, thank you very much!

Honestly, without Aziz I will never know that idea started under Tinkerer project although I am trying to follow all his posts.

[Aziz]

Some Calculations and Thoughts

Hi all,

there might arise some questions about the transmitter:

1. Is the flyback duration (half sine flyback period) dependent on coil current / battery input voltage?

No, it is only defined by the L and C of the resonant circuit.
Its period is
Tflybackperiod = PI*sqrt(L*C)

2. How much is the maximum flyback voltage?

It is dependend on the coil current I at switch-off time. Provided that, the mosfet is not avalanching. This should be avoided at any rate.
The magnetic field energy E stored in the transmitter coil L is defined by:
E = 0.5*L*I²
This energy will be transfered into the capacitor C. It will be charged until the coil has been discharged fully (I=0). The energy stored in the buffer capacitor is defined by:
E = 0.5*C*U²

BTW, at switch off, the capacitor C is already charged. It's voltage is Uin (input voltage).

If we don't take the losses into account, the equation must be equal:
0.5*L*I² = 0.5*C*U²
L*I² = C*U²
U = sqrt(L*I²/C)

So the maximum flyback voltage will be:
Umax < U+Uin (losses wont let the equation to be equal)

Note, the capacitor C was already charged with input voltage and therefore the flyback voltage is increased by this voltage.

To increase the flyback voltage, just reduce the capacitor C or increase the coil current (or increase the input voltage). Reducing the capacitor C will reduce the flyback duration time. You should avoid the mosfet avalanching. If the mosfet avalanches, a lot of the coil energy will be lost otherwise.

3. What's the purpose of the input inductor L2 and the capacitors (C14,C15,C9)?

If the transmitter is highly efficient (=less losses), it will draw less current. The most of the energy in the coil will be efficiently recycled back. During the recycle phase, the recovered energy should not flow into the battery back (may damage it or damage other circuit sections). So it is providing a high impedance path to this source. The capacitors C14,C15,C9 provides a low impedance buffers for this purpose.
It is very important to use low ESR capacitors. The coil energy buffer capacitor C should also be a high voltage, low ESR type.

4. How can the efficiency be increased further?

- Use continious mode operation (Phase 4 enhanced)
- Use discontinious mode with Phase 4 enhancement (switch on the mosfet during recycle phase additionally)
- Do not use any damping resistor or snubber circuit in the transmit coil
- Use heavy duty coil wire and leads (low resistance coils)
- Use high current rate mosfet switches (low on-resistance)
- Do not let the mosfet avalanche (limit the flyback voltage by increasing the capacitor C or use high voltage rated mosfets)
- Increase the input voltage
- Do not saturate the transmit coil current (should be operated in the linear mode)
- Use low ESR capacitors (input voltage buffer, coil energy buffer)

To be continiueed..

Aziz

[hobbes_lives]

Tinkerer's circuit reminds me very much of a class-e amplifier driving a tank circuit. The timing is a little different, but perhaps that's the inspiration of his design?

I used a class-e amplifier to drive a tank circuit for the transmit coil in a VLF detector when I first started trying to design metal detector circuits. The timing was conventional, and it was very efficient. I didn't know about coil shielding back then, so there were some stability and drift issues. I may have to revisit the circuit again in the future.

[Aziz]

Pure low voltage USB version possible!

Hi all,

the complete schematics can even be further minimized using the +5V from the USB port. Only a single +5V/100mA supply operation can be implemented with reduced coil current (requires high transmitter efficiency).
We have to use logic gate mosfets. Most part of the circuit can be dropped (voltage regulators, converters,...).

Well, I have to build both versions (kick-*** high coil current version and a minimal USB version).

Just a though.

Aziz