ARMD (ARMRADIO based Metal Detector) VLF IB PROJECT

[moodz]

Tone multiplexing. Does this mean that no external circuit will be required for AIB?

I am thinking there is enough horsepower in the DSP to either multiplex tones in the frequency domain and or the time domain to provide an IB "error" signal to cancel the IB imbalance.
Once the RX signal is digitised and with full knowledge of the TX waveform there are no restrictions except processing capability. If the TX is driven by a quadrature signal then any modulation ( ie correction ) can be done in DSP.

[moodz]

A few more days of slaving over a hot computer ... Everything inside the dotted box is working.
ARMD version 1.
There is not much in it really the DSP inside the ARM chip takes care of everything wiht bandwidth to spare.

The AIB I will leave out for now ... the LAP amplifier blocks the AIB signals. I could use a regular amp here ... but the log function provides 60 db of dynamic range.

moodz.

[KingJL]

moodz
Are you using a "tuned" RX coil?

[KingJL]

@moodz:
Are changes to the demodulator section of ARM RADIO determined yet?

[moodz]

moodz
Are you using a "tuned" RX coil?

I am going to start testing with an IB coil that was orginally intended for a PI detector.

Two reasons.

1. Its not tuned ... wide band response from DC to nearly 1 Mhz.

2. The TX coil can handle lots of current ( eg upto several amps ).

[moodz]

@moodz:
Are changes to the demodulator section of ARM RADIO determined yet?

Mostly I hope.

1. Added the NCO out from DAC2 ( this was audio out ). The NCO or LO must be brought out as it is the synchronised drive required for the TX.
The NCO is a nice sinus wave so it is already in the rignt format to amplify and feed to TX coil via a low distortion power amp.
Its possible to set the frequency to what ever you want. ( eg range is at least 1 kHz to 100 Khz ) The frequency steps are slightly better than 1 Hz ( due to sampling speed ).

2. The VDI was just an additional type of demodulator. It works on the X and R ( IQ ) buffers that are in the code.

3. There is target tone audio ... it sounds like a PI. Considering mapping VDI to a tone bank. ( not done yet ).

4. The log amplitude phase recovery circuit ( LAP ) which is external to the ARM processor does have a tuned element in it and due to the circuit simplicity it limits the recovery frequencies.
The benefits though are that the input RX voltages to the ADC are limited by a log response ( preventing overload ) whilst preserving phase information for VDI. So I will test it with and without this circuit.

5. The power amp for the TX is not resolved. Looking at CCPI sine wave generator also. It needs to be wide frequency range ( eg from 1 khz to 100 khz )

6. I need to add more debug code to the circuit. Small changes can upset the signal chains very easily as the important ones are driven off interrupts and timing is everything. I also want to add some useful code tools for testing like a frequency counter and modulator for the TX waveform.

7.The individual parts are working ... I need to lay them out on a breadboard for field tests.

8. The CUBEIDE development platform works OK however at the moment not using the HAL library. The HAL library autogenerates the device code ( like ADC / serial etc etc ) However it lets you select combos which dont actually work and it generates code that compiles with no errors etc .... but does nothing. Unless you know the low level dependancies of the ARM hardware the HAL libraries are useless plus the code generates alot of bloat code.

moodz

[moodz]

Added a debug serial port to the codeblock on USART3 to drive a bluetooth HC05 dongle for tracking variables on the go and debugging ( printf function mostly LOL )

Added a reciprocal frequency counter to the code block with 5 ns accuracy for measuring frequencies with 1 Hz resolution or better. This is needed for tuning the analogue frontend.
When the detector starts up there are some very narrow bandwidth features in the analogue front end to make the log amplitude and phase recovery work and they need to be measured at startup and probalby from time to time to track drift. The internal digital NCO will adjust to compensate.
moodz

[moodz]

Testing the CCPI TX module for suitability for the ARMD project. I have only tested one frequency as there is a lot of gotchas when you move frequency.
The selected test frequency is 8.33 Khz and the TX circuit develops 4 Amps peak to peak in the IB Tx coil with approx 10 millivolts of unbalance in the RX coil.
The CCPI support Pulse induction mode as well ... so I thought why not. By switching a single capacitor in / out the CCPI converts from PI to VLF. The DSP sampling
front end remains practically unchanged and instead of IB balance we can do ground subtract. ??? maybe

I was pretty happy with 4 amps in the TX coil and less than 4 watts consumption from the battery. ( in PI or VLF mode ). There are some linear regs in there .. so room for improvement.

Note the 4 amps peak to peak is in sinewave VLF mode ... the PI mode is about 1.5 amps as the flyback is 2 microseconds and the mosfets dont like it above that.

Here is the latest blocky of the ARMD detector.

[pito]

Can you post a picture of coil which can handle 4A ?

[pito]

[QUOTE= TX circuit develops 4 Amps peak to peak in the IB Tx coil with approx 10 millivolts of unbalance in the RX coil.
[/QUOTE]
4A and 10mV = very very impressive

[moodz]

Can you post a picture of coil which can handle 4A ?

The coil is a minelab DD ... Its made for PI use and I believe the TX coil is wound with Litz wire. There is no capacitor across the coils ( like you may find in some VLF coils ).
Its a wide band low capacitance coil and the TX reactance at 8.33 Khz is about 15 ohms - 20 ohms and real ohms is 0.4 approx.

It develops 60 volts peak - peak across the coil @ 4 amps.

I should have quoted the IB leakage as 10 millivolts per amp ... so the unbalance at 4 amps is 40 millivolts not 10 millivolts. However 40 millivolts at 4 amps is OK for my application.

Below is pic of waveforms from the testbench ...

Yellow = TX clock
Pink = RX voltage ( 40 mv pp )
Light Blu = TX voltage ( 60 volts pp )
Dark Blu = TX current ( 400mV pp ) sensed by 0.1 ohm resistor in TX leg.

[moodz]

...and here is the PI mode by flicking the switch. Universal detector PI and VLF together
Same probe points ... adjusted the CRO scaling. The TX current is 2 amps peak to peak. The flyback time is 4 microseconds ( till sampling ).
Note the RX is shown here from the first preamp. ( note theres no diode clamping at the input ).

With a 5 volt supply PI mode current is 540 mA. In VLF mode current is 360 mA.

Another cap can be switched out to improve the flyback by 2 microseconds ( 2 microsecond sampling ) however if this is done then the supply voltage must drop to about 2.5 volts due to mosfet avalanching. This would be the "fine gold" mode LOL.

[pito]

40 millivolts With IB NULL - how much lower you can get ?
Your sines do not look so good, try TC4420.

[Mdtoday]

I’m late to the plate, however the project sounds like a good challenge.
Nice work as usual Moodz.

[pito]

For PWM filtering I use this filter, 10k and 1nF, it gives a pure sine wave. How clean is your signal at the Tx amplifier input?



It develops 60 volts peak - peak across the coil @ 4 amps. = For calculation you should use RMS not pp voltage