Did some more experiments to see how I could interface a fluxgate coil assembly the best way to a microprocessor.
The original idea was to drive the 'drive coil' from between 2 pins of a PIC. The problem I ran into is that you need to drive the core into saturation, and that saturation point depends on a lot of factors such as permeability of the core material, and those factors have this tendency to change, So that sort of rules out driving from software with a *fixed* frequency. Of course I was appalled by the large number of analog components in designs I have seen, used mainly to generate sampling pulses for analog switches (in abundance too). So anyways trying to solve the constant saturation point problem, I considered using the PIC's comparator to, in an interrupt, switch the polarity of the drive voltage, the comparator used as current peak detector as in any switch mode. But the PIC also proved to not be powerful enough to drive the 100 mA or more current into the coils... So I needed 2 drive transistors anyways, But then why not leave it free running as in 'multivibrator'? Everybody seems to do that. So, OK, but then what to do with my nice peak current detector idea? Well that proved to be a stroke of genius (tm). By simply using a PNP in the supply to 'feel' the current peak that happens every half period that the core goes in saturation, and a simple resistor, you get a nice very small sample pulse EVERY HALF PERIOD, and then you only need ONE analog switch per channel (x, y, or z axis). Sensitivity proved absolutely great, stability too, added 22x gain with a LM324, output swings between +5 and 0 V for a magnet rotating north to south 20 cm away. I added a lowpass after the sample and hold cap to filer out some ripple from the multivibrator, but this likely can replace the sample and hold too. The PIC can make 1/2 U as 'zero' as there is only a +5V supply (PID has a programmable DAC output on pin 19), saves an other 2 resistors. So I think I have a working circuit, LCD works too, ADC software can be used from my other projects. Now for a real PCB and the other channel[s] (only tried North South). No complicated diodes and RC differentiators to get sample pulses, only ONE
74HC4053 for 3 axis, only ONE quad opamp (one for 1/2 supply voltage buffer, other three for x, y, z gain), And a 7805 for supply, NO SWITCHER, that magnetic field would cause havoc.End of report.
DANGER DANGER PICTURE OF HAND DRAWN CIRCUIT DIAGRAM, LOOK AT YOUR OWN RISK.
Filed under Usenet patent 001011110110ZZ fluxodecimal. Prior art you have seen it here,
All rights reserved. Donutted into the public (domain). Burn before reading.