NEW !!! Full differential PI front end.

[Aziz]

Hi moodz,

I am also preparing two center-tapped coils. In this case, they will be shielded with AL mesh stripes (like the scotch 24 shielding tape).

The bigger one has 2x19 turns, 33 cm (13 inch) diameter, approx. 1.06 mH (266 µH each half), made from twisted pair cable.
The smaller coil has 2x15 turns, 20.5 cm (8 inch) diameter, approx. 400 µH (100 µH each half), made from two loosely coupled parallel cables.

The number of turns came from only the wire length, I could find at home. So they are somehow arbitrary and not intentionally choosen.

I want to see, how shielded coils perform on SSM2019 and INA163.
Aziz

[moodz]

Hi Aziz .... I will be very interested to see the effect of the shielding on your coils ... I have not used any shielding yet ..... what resistance are you getting ?? ... and your total inductance is 4 times each side ??

moodz.

[Aziz]

Hi moodz,

Hi Aziz .... I will be very interested to see the effect of the shielding on your coils ... I have not used any shielding yet ..... what resistance are you getting ?? ... and your total inductance is 4 times each side ??

moodz.

My multimeter isn't accurate enough to measure this exactly. The wire I am using is a 0.5 mm core copper diameter. With insulation, it has 1 mm diameter. The calculated resistances should be 1.8 and 3.6 Ohm (on both ends).

I am allways referencing to the total coil inductance seen on both ends of the coil. If I have a center-tapped balanced coil, then each half has only one forth of the total inductance. Inductance is direct proportional to the square of number of windings.
If both coil halves are 1:1 coupled (and they are, coupling k=1):
Ltotal = Lh1 + Lh2 + 2*k*sqrt(Lh1*Lh2), where Lh1 = Lh2 = Lh
Ltotal = Lh + Lh + 2*sqrt(Lh*Lh) = 4*Lh
Lh = Ltotal / 4 (q.e.d.)


The coils are not tested yet. Just working on a single supply version to get rid of the P-MOSFET using direct DC coupled coil signals and bipolar transmit pulses. The bipolar transmit pulses have huge inherent advantages . I also want to make same grounding for laptop and PI board to drive both devices with same power supply.

Regards,
Aziz

[moodz]

Hi Aziz ..

.... of course forgot the coupling k ... I am used to lumped inductances.

Have a look at this patent ... bipolar pulses in 1979 ... some of the ideas in this patent seem to have been 'lost' by current designs ... and even possibly re-patented under obscure descriptions and circuits ... ( not mentioning any names )

corbyn_patent.pdf


moodz.

[Aziz]

Hi moodz,

thanks for the hint. It seems to be expired and obviously in the public domain now .

BTW, I will regulate the coil supply voltage to minimize further noise and instability. That will give allways fixed transmit pulse energy emission.

Aziz

[moodz]

hi all .... fully differential bipolar drive spice sim.
features front end FET RX switching and Switched Damping resistor.
works even nicer on the breadboard.
The part types are representative ... the breadboard uses different physical semiconductors.

[moodz]

hi all ... o/p from full diff sim ....
green trace is +ve fly back into RX fet switch
blue trace is -v fly back into RX fet switch
red trace is +ve o/p from RX fet switch
light blue is -ve o/p from RX fet switch

note small < 0.5 us discontinuity due at RX fet switch on time.

the oscillations at the end are where the damping resistor FET switch turns off.
enjoy.
moodz.

[moodz]

Near ideal magnetic pulse transitions from the TX coil can be achieved if a forcing function is applied rather than a simple square wave pulse.
In the plot below you can see the current flow reversal 3.5 amps to -3.5 amps ( and hence the mag field ) that reverses in about 0.7 uS ... which is quite good ... however what is notable about this result is that the flyback decay is constrained to the current reversal time .. ie there is no significant flyback voltage outside the coil current decay time .. thus the flyback has decayed in 0.7us also.
A indication that the flyback is constrained to the decay time is the peak voltage ... nearly 5 kv in this case .. because the flyback energy has to be constrained to the coil current decay time ... ie the shorter the time the higher the voltage.

The red trace is the coil current ( coil = 220 uH 0.3 ohms 100pf )
The green trace is the fly back voltage at the MOSFET end of the coil.

The tx pulse is still 100 uS ... the forcing function is generated by a 'filter'
.... there are no resistive components in the 'filter' ... so this timing was achieved with no damping resistor at all !!

Note that this is a result from an LTspice simulation of the circuit as my CRO cannot take this kind of abuse from the actual circuit.

moodz

[Aziz]

Hi moodz,

but you need the dampening resistor to eliminate the self resonant frequency of the coil. Consider, if there is a resonant system, it is very susceptible to ground effects (capacitive effects, relative permeability, ...).

Dampening is a fundemantal benefit of a PI system.

Aziz

[moodz]

Hi moodz,

but you need the dampening resistor to eliminate the self resonant frequency of the coil. Consider, if there is a resonant system, it is very susceptible to ground effects (capacitive effects, relative permeability, ...).

Dampening is a fundemantal benefit of a PI system.

Aziz

You are right Aziz ... however an oscillating system is one that does not have constraints ... the damping resistor is one kind of constraint however it is passive ... by applying an active forcing function of known limits the oscillation may be removed in the same manner ... this is done in amplifiers for example with compensation which is not a damping resistor. The feedback in an amplifier is a type of active forcing function that modifies the input condition.
I will need to investigate further ... however my aim with this circuit is to produce a very fast reversal magnetic field in the ground / target.
So far the results are good ... the down side is the extreme voltages produced at the MOSFET ... even 5 volts of drive produces over 700 volts ... this is probably due to the 'Q' of the 'filter' circuit I am using to produce the coil forcing function.
Regards,

moodz.

[simonbaker]

You are right Aziz ... however an oscillating system is one that does not have constraints ... the damping resistor is one kind of constraint however it is passive ... by applying an active forcing function of known limits the oscillation may be removed in the same manner ... this is done in amplifiers for example with compensation which is not a damping resistor. The feedback in an amplifier is a type of active forcing function that modifies the input condition.
I will need to investigate further ... however my aim with this circuit is to produce a very fast reversal magnetic field in the ground / target.
So far the results are good ... the down side is the extreme voltages produced at the MOSFET ... even 5 volts of drive produces over 700 volts ... this is probably due to the 'Q' of the 'filter' circuit I am using to produce the coil forcing function.
Regards,

moodz.

Very interesting stuff you are pursuing, keep it up

You and Aziz make a good team... we're going to see some great new PIs one day...

-SB

[moodz]

zpi v1.0

.. been a bit busy lately but did get round to making up the diff coil demonstrator platform. Based on the good ol surfmaster design except with a diff coil front end and dspic back end ... note also the design is DC coupled all the way to the ADC.
I do have some fancy amps in the parts bin but have stuck with the good ol CA series as I have lots of em and like they say ... a good engineer dont buy what he wants he makes it out of what he can get ( or has in this case ) ... anyway the circuit works superbly with these antiques and demonstrates the benefit of the diff coil .. which is the purpose of this demonstrator circuit.
The diff coil I am using on this particular version is 22 cm in diameter, 3 ohms total resistance and has 60 turns. ( yes you read right ... six - oh ... turns ). The single control is an ALPS rotary encoder with push button switch into the dspic quadrature encoder. There is no display whatsoever. ( Oz is full of snakes and mineshafts ) Menu is accessed by pushing the controller knob and a PWM voice prompt is injected into the audio output eg 'VOL' 'DELAY' 'DISC' 'SENS' then twist knob to set level with beep rate indicating level etc thanks to stored samples in the generous ROM they give you in the dspic. Target signal is a rising click to tone with warble for discrim function.
The power supply and audio amp are on a separate board not shown below ... mainly because I have not drawn it up yet.
Moodz.

[Muntari]

Hi Modz,

Is the orientation of D5 and D6 correct?

Looks interesting though

Cheers

Brian K

[moodz]

Hi Brian ..... D5 and D6 are correct ... the PIC port will try to source current when high and sink when low ( ie current reverses ) so the diodes allow the port to pull low but the pull up resistor to +15 volts pulls the control input of the 4066 up ( ie ON ) and the diode blocks the high bias from the PIC when the PIC output is high ( ie OFF ). I am not an expert at 4066 control but it definitely does not work without the diodes.

Version 2 will do away with the differential integrator and do it all in software ... time of course is one resource I dont have at the moment.

moodz

[moodz]

doh ... just realised that explanation is wrong sort of.. when the PIC port is high the 4066 is ON. When the PIC port is low the 4066 is OFF. The diodes prevent the PIC high level from driving the port ... this is done by the pull up resistor to get correct bias.

moodz (again )