Tweet
Originally posted by Aziz
View Post
Hi Max,
I have to wait until the beginning of the month, before I can order some parts: I am on welfare and have a very low budget per month for the development (~50 €).
So the hardware complexity have to be dramatically reduced due to my financial situation. The software complexity does not cost me anything.
I think, I will use very likely ATtiny2313 for dedicated timing logic. Other AVR µC (ATmega8 or smaller one) might be used for multi-tone output and man-machine-interface (user-inferface). This will reduce the total complexity of the hardware dramatically (no meter/LCD display, less switches and pots if possible).
Using a very fast ADC in digital PI's makes not sense at all. The sampled signals will be interfered with noise in time-domain and they have to be filtered out anyway. The semiconductors for this purpose are very expensive and very difficult to get. So the averaging/differential integrators will work perfect.
I will go a quite different way for PI technique:
- reducing the transmit pulse power to reduce EMI and noise and also avoiding some inherent problems of FET's due to high flyback voltage. Also a problem with magnetic susceptibilty and saturation due to higher magnetic fields. My results are showing me, that there is a better way of performance instead of the brute-force-method (high coil current).
- increasing the total battery live for more operating time
- using new low noise amplifier and techniques to reduce noise
- making the hardware as easy as possible to reduce the total number of active elements
- using of external crystal clock generator with reduced jitter and timing-errors (very critical part)
- increasing the number of pulses per second (PPS) (10k..15k or more)
- decreasing the time delay after transmit power switch off to take advantage of faster dI/dt coil current decay (see Faraday's law)
- using very fast, low capacitance coils and low power, ultra fast FET's
All this stuff and more has to be reconsidered to avoid noise and drifting problems. I am sure, by increasing the signal-to-noise ratio by avoiding the brute-force method will gain more progress.
Regards,
Aziz
I have to wait until the beginning of the month, before I can order some parts: I am on welfare and have a very low budget per month for the development (~50 €).
So the hardware complexity have to be dramatically reduced due to my financial situation. The software complexity does not cost me anything.
I think, I will use very likely ATtiny2313 for dedicated timing logic. Other AVR µC (ATmega8 or smaller one) might be used for multi-tone output and man-machine-interface (user-inferface). This will reduce the total complexity of the hardware dramatically (no meter/LCD display, less switches and pots if possible).
Using a very fast ADC in digital PI's makes not sense at all. The sampled signals will be interfered with noise in time-domain and they have to be filtered out anyway. The semiconductors for this purpose are very expensive and very difficult to get. So the averaging/differential integrators will work perfect.
I will go a quite different way for PI technique:
- reducing the transmit pulse power to reduce EMI and noise and also avoiding some inherent problems of FET's due to high flyback voltage. Also a problem with magnetic susceptibilty and saturation due to higher magnetic fields. My results are showing me, that there is a better way of performance instead of the brute-force-method (high coil current).
- increasing the total battery live for more operating time
- using new low noise amplifier and techniques to reduce noise
- making the hardware as easy as possible to reduce the total number of active elements
- using of external crystal clock generator with reduced jitter and timing-errors (very critical part)
- increasing the number of pulses per second (PPS) (10k..15k or more)
- decreasing the time delay after transmit power switch off to take advantage of faster dI/dt coil current decay (see Faraday's law)
- using very fast, low capacitance coils and low power, ultra fast FET's
All this stuff and more has to be reconsidered to avoid noise and drifting problems. I am sure, by increasing the signal-to-noise ratio by avoiding the brute-force method will gain more progress.
Regards,
Aziz
it's interesting idea. You can gain xtal stability using uC with external xtal at oscillator pins, easy and effective. Some jitter will be there cause of some uC latency but I think will be at least an order of magnitude more stable than using e.g. NE555/556 like older PI's have done in past... an example is goldscanIV, very powerful device... but now obsolete about these parts, I mean voltage converter, timing and switching device, rest of integrators could be reused or give ideas about design.
Increasing PPS will give you faster response and you'll get easy smaller stuff cause of skin effect combined with eddy currents there... consider that very good PIs by Eric Foster for beach hunting run in the 3KHz range +-200Hz(deepstarII) so using 10KHz will give fair results on small stuff.
I think will be an interesting project.
Kind regards,
Max
Comment