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

[Aziz]

Eee PC 901 consumption

Hi all,

I did some power supply measurements on the external 12V input plug of my Eee PC 901 (the LDO is working with slight modifications - more on later post's).

I have removed the internal battery from the Eee PC. WiFi and blue-tooth switched off. Following current measurements were done with 12.1V power supply using the LDO circuit, which was fed by a 12V/5A car battery charger.

Switched off: 40mA (!) (WTF should this!)

Then switched on and waited after all processes got started and the cpu got into idle mode (<5% cpu load):

Idle Mode (display off): 0.65 A
Idle Mode (display min): 0.66 A
Idle Mode (display max): 0.76 A
Standby mode: 40mA
Idle Mode + external USB sound card + display off: 0.85A
The external USB sound card consumes obviously appr. 200 mA.

Then running the detector software under full cpu load (100%), plugged USB sound card and with normal LCD brightness: 1.25A (15.125W).

The 12V / 2700 mAh Ni-MH battery pack would run at full load : 2,7/1.25 = 2.16 h
A 14.4V / 8800 mAh battery pack would run: 7 hours with simple LDO (no energy saving).
A 14.4V / 8800 mAh battery pack with efficient converter (95%): 14.4*8.8*0.95/15W = 120W/15W = 8 hours.

I suspect, that the 12 V input path is not efficient for operation. So an access into the battery plug would be convenient. But I have no specifications of this plug.

The 12 V input plug is running up savely from 12V-5% / 12V+5%. I have measured a minimum input voltage of 11.25V and a maximum voltage of 13.1V. Outside the limits, the netbook won't either start or will shut off simply.

All these measurements were done without the internal battery plugged.

The internal 7.3V/6600 mAh (48 Wh) Li-Ion battery of the Eee PC should run for 3 hours and 12 minutes (with external USB sound card and working detector software of course).

Aziz

[Aziz]

Hi all,

I did work on an improvement of the inductor-free, high current, ultra low-drop step-down switching converter for the netbook last days. The schematics went up complex but it is probably the lowest drop solution so far: max 200 mV drop. Only max. 300 mV will be converted into heat on higher input voltages.

Principle of work:

LDO as previously shown. But its input voltage will be regulated as down as possible between two threshold voltage limits:
lower limit: Uout+200mV
upper limit: lower limit +200 mV (hysteresis)
A buffer capacitor tank will be charged, if its voltage is below the lower limit. The charging will be stopped, if its voltage exceeds the upper limit. As the regulation of the buffer tank is a switched capacitor type, the drop voltage (Ubattery - upper limit) won't be wasted (heated up). Only the LDO heats the smal drop voltage (max. 300 mV average) and the FET resistors during switching.
The switching frequency is dependent on the load, buffer capacitor value and source resistance of battery. For constant loads, the regulator can even be synchronized to an external clock. The regulation ripple noise is similar to the SMPS and it depends on the load. I won't publish this schematics as it is too complex to built it (discrete solution). There is a much simpler solution with a PWM controller and a fixed switched frequency possible, which should work more better and procude lower noise.


The PI controller coil power part:

The coil power controller will be a similar switched capacitor solution. A big buffer capacitor will be charged upto to a defined voltage (user adjustable via DAC control). The start of the charging can be triggered by the user program code (synchronisation). So the higher input voltage will be efficiently used for longer operating durations. The regulation of the voltage is very accurate, so every pulse energy remains same even the battery voltage diminishes over time. It also allows lower-power search modes for lower EMI restrictions in different countries.

Aziz

[Aziz]

Hi all,

ok, the 12V netbook regulation will be efficient now even the input voltage is much higher than 12V. Only a fixed drop of 0.5-1V will be heated up from the high battery voltage.

Could be a new invention. I should check this some time.

Aziz

[Aziz]

LDO Regulator for Eee PC 901

Hi all,

the netbook Eee PC 901 needs some power supply circuit for field usage (12V car adapter regulator). You can forget all the cheap ebay devices. They are waste of money.

The Eee PC 90x series has an external 12V DC input plug (3 Ampere). The specification for this input is tight and the Eee PC detects undervoltage/overvoltage and it switches off in case of out of specification range.

I thought, it would be very convenient, to have a power supply module with regulated 12 V output for the netbook. It is ok, if the design is optimized for 12V operation and a low drop-out (LDO) regulator could feed the netbook. So a pure single supply for the controller and netbook could be realized. The PI controller will support more voltage (up to 18V) and the drop voltage (18V-12V = 6V) for the netbook would be wasted of course. The energy loss would be less, if 12V/13V input is used.

I have designed a discrete LDO circuit below, which has only max. 150 mV drop voltage and it would deliver enough high current for the Eee PC. It makes sense, that the PI controller should have a 12 V output for the Eee PC. The car adapter could be fed into the PI controller to charge the Eee PC's battery.

In the field, only one big single battery could be used for both devices (four cell 14.4V Li-Ion for instance).

Below is a proposal of the LDO regulator. I have not built it yet and please wait, until it works and I power my Eee PC with it.

Aziz

[Aziz]

DAC and Reference Voltage

Hi all,

i have choosen the following chips for the controller:

DAC:
The MCP4921/MCP4922 from Microchip seems to be quite easy to use and are very cheap (less than 2 EUR). The MCP4922 is a dual version. The DAC's are 12 bit in resolution with SPI interface. They offer a fast settling time so the VCA can be set quickly for gain changes.

MCP4921/MCP4922 (data sheet):

http://www.microchip.com/wwwproducts/Devices.aspx?dDocName=en020398

I will order both chips. The dual version will be very likely used. If they do not work well, I will throw them out of the window and look for another one.

Reference Voltage:
The reference voltage chip will be a 2.5V precision reference. I am not going to use the internal bandgap reference of the MCU. The external reference will be fed into the MCU instead.

The LT1019-2,5 should be a good one (3 ppm/°C, 5.20 EUR). This is the heart of the sensitive parts and should be a precision type of course.

http://www.linear.com/pc/productDetail.jsp?navId=H0,C1,C1154,C1002,C1243,P1130

Aziz

[Aziz]

Hi w5cdt,

How about a Bluetooth interface to your iPhone? You'd have a wonderful display and touch screen. Good developer tools. Plus you could photograph your finds and phone them in. :-)

No, this does not make sense. We already use the USB interface for the sound card and another USB interface for the PI controller is just ok.

Small handheld PDA-like devices have a very short life time. Notebooks and smaller counterparts like the Netbook's will exist longer. The latter one became very attractive for longer operating durations (>6 h) and are darn cheap (~ 250 EUR).

The software in the Netbook will support only a XP/Vista/Windows 7 OS. So it gives the possibility to the user to develop his own add-on libraries and software packages.

I even do not need the fine LC display of the Netbook. The LC display will be switched off to save battery power. Just switch on and close the Netbook display.

Aziz

[w5cdt]

USB interface to metal detector

How about a Bluetooth interface to your iPhone? You'd have a wonderful display and touch screen. Good developer tools. Plus you could photograph your finds and phone them in. :-)

[Aziz]

Happy New Year My Friends!

Happy New Year My Friends,

my best wishes to everybody. I hope the development will go faster in 2010.

Main objective for the project in 2010:
Finish of user programable PI controller device.

Slight extenstions to the current version:
- coil voltage regulator controller
- second DAC (for the coil voltage regulator controller)

The coil voltage regulator controller will make the coil voltage user adjustable. It will stabilize the coil voltage to avoid drifts due to battery voltage diminish. So every transmit pulse energy remains at same level.

Benefits:
- synchronized to cycle period
- very precise reproduceable regulation
- very low drop-out regulation (<100 mV)
- increased power efficiency (voltage drop won't be converted into heat)
- increased battery voltage range (approx. 11 - 18 V)
- low noise (!!!)

Next part for January:
- coil driver (fully differential)
- coil voltage regulator controller
- Reference voltage generator
- DAC's

Aziz

[Aziz]

PI Controller Hardware Architecture Rev 1.0

Hi all,

below is the hardware architecture of the current project. It should be a good base to start with. Even if I should not able to finish this project due to some unforeseen circumstances*, but some of you could still follow it. Some parts of the design are not finished yet and might change slightly.

Aziz

* unforeseen circumstances:
- lack of money (will delay the project only)
- illegal arrest (from criminal german government)
- ...

[nick_f]

Hi all,

I want to show and encourage you, why I am following the differential PI front end. First picture taken at both coil ends at the high voltage protection clipping diodes. You see the balanced signals, which are heavily overloaded with some noise (common mode noise perhaps?). The coil and the coil leads are not shielded. A single ended PI front end would process only one of the signals.

Now see the second picture after removing the common mode signals (noise?). This is easily done within the scope by inverting channel 2 and adding both channels together (result CRO = ch1 + (-ch2) = ch1 - ch2 ).

Can you see a significant improvement?

Aziz

Hi Aziz, that is a great experiment! Thanks for sharing.

Nicolae

[Aziz]

Communication Considerations

Hi all,

I have some comments on the RS 232 communication. Assume we have a communication speed of 9600 Baud/s.

9600-8N1 (9600 Baud/s, 8 data bits, no parity, 1 stop bit, no hardware flow control)

We have 1 start bit, 8 data bits and 1 stop bit: Makes total 10 bit for each character transfer.

So we achieve a character speed of
(9600 bit/s) / (10 bit/character) = 960 character/s

If we have a cycle frequency of 1000 Hz, then almost one character per cycle time is transferred.

The Atmel MCU has a USART FIFO of two characters (no joke, what a luxary). While the received data is in the receive data buffer, next following data can be received in the same time. So we have to read the data content just before the next data reception is finished. We could potentially double the communication speed here.

So the maximum blocking time in the CPU simulation thread would be 2/960/s = 2.08 ms. If the MCU is blocked for longer periods, then the received data would be overwritten by the following received characters. Normally, the blocking time of the MCU is not so long and there are other tasks to do before reading the received data. The baud rate of 9600 seems to be slow and fast enough for our purpose.

A typical controller firmware code is between 128 .. 512 bytes. This is transferred within one or two seconds. A generic controller firmware code allows also code modification possibility to "poke directly" in the op-code for changing parameters (delay counters, nop timings, etc.). And this is almost done in realtime (~10-15 ms, depending on the communication protocol).

I will think of implementing both ASCII and binary communication protocol. The binary version allows more efficient data packing and doubles the communication speed. The ASCII based communication protocol will be implemented first.

No doubt, I will also implement the true USB interface for the later versions. The FT232/CP210x chips are very cheap and easy to use. But difficult to solder.

Aziz

[WM6]

Great work Aziz, thanks for all explanations.

[Aziz]

Progress

Hi all,

now the timers are changed: no problems at all

Timer0: used for CPU simulation of timer delay instruction (former Timer2)
Timer1: reserved for PWM outputs
Timer2: used for 10 ms time base for the internal software timers (8/16-bit)

Care must be taken, whether the in/out or lds/sts MCU instruction is taken. Memory mapped I/O addresses must use the lds/sts MCU instruction.

I found some bugs in the AVR Simulator 1/2. Particularly the 644P device simulation implementation has problems with Timer2 and /1024 prescaler. But the target is working correct.

Next steps (firmware software):
- main thread implementation (main program flow)
- communication with laptop (very big part)
- GUI interface implementation (knobs, switches, push-bottons, ...)

The controller instruction op-codes (CPU simulation) will be finilized much later, when I know what kind of instructions are really necessary. I have approx. 60 op-codes implemented at the moment (max. 192 possible).

Aziz

[Aziz]

Hi WM6,

Hi Aziz

"So the PWM frequency can be set from 305 Hz into hundreds of kHz (16 bit timer)."

I only replying to this your statement. I don't know if there is enough time between two pulses to sample detecting information at proper point by hundreds of kHz? I am not so deep in your construction to answer myself.

The PWM is just for the cheap DAC implementation (pulse-width modulation into analog voltage conversion). The generated voltage will control the VCA's gain and is usually synchronized to the cycle frequency.
The synchronisation is very important, as the DAC will have some residual ripple noise after the low pass filter. This ripple noise should be synchronized to the cycle frequency to have always same condition at start of the cycle. There is no demand of having a fast DAC output and the settling time of it can be very large (up to one second even). However a combination of VCA & PGA circuit should make faster gain changes through the PGA possible. If we would use an integrated DAC, then the PWM will not be used anymore.

For the fast changing signals, there will be four S&H amplifiers. They will be as fast as the settling time of the S&H allows (depends on holding capacitor, the input resistance path into it and the amplifier speed). This is not much critical and we can sample where we just want.

Typical settling time of the S&H amplifier will be in the range of 1..2µs. Once a S&H amplifier holds the analog signal, the modulator will convert it into an AC signal, which is passed to the sound card for further digitizing and processing.

Aziz

[WM6]

Regarding your question:
Could you specify your question? I do not understand it right.

Hi Aziz

"So the PWM frequency can be set from 305 Hz into hundreds of kHz (16 bit timer)."

I only replying to this your statement. I don't know if there is enough time between two pulses to sample detecting information at proper point by hundreds of kHz? I am not so deep in your construction to answer myself.