Fast PI coil by Joseph J. Rogowski a good choice for MiniPulse Plus?

[moodz]

Moodz,

Anyone looking at the MOSFET COSS charts will see that the effective MOSFET capacitance tends to decrease with increased voltage. But, since the high voltage diode also has capacitance, and it is in series with the MOSFET COSS, this combination represents a total lower capacitance in the TX circuit and thus has the potential to sample at lower delays with a higher damping resistor value. If the bleeder resistor between the diode and MOSFET is not the best option, is there better option to allow a faster TX spike turn-off with a faster RX turn-on to sample at lower delays? This is a forum to help PI builders learn about what works and why. This is the approach I took in writing my fast coil article.

Thanks for the feedback.

Joseph J. Rogowski

Hi BB ... the diode capacitance is also minimal at higher reverse voltages ( see attached pic for fast recovery diode 1N4937 ) so the high voltage trapped between the diode and the drain of the mosfet is actually beneficial to a certain degree. No need to drain it off.


My patent AU2013101058

shows you how to damp a coil in 1 TC or better ...improvement over 5 TC approx with a resistive damper.

moodz.

PS nevertheless your article on fast coils is very well written/informative and I enjoyed reading it.

[moodz]

But the diode is a diode first and a capacitor second and it conducts while the mosfet
capacitance is charging so the in-series capacitance rule can't help us in this case.

Hi Crane ... you are right but consider that the diode only conducts during the TX ON and most of the flyback then turns hard off by voltage trapped between mosfet drain and diode.

The main drawback of this capacitance was not sampling time impact but noise injection to frontend in my experience.

[Davor]

Hi BB ... the diode capacitance is also minimal at higher reverse voltages ( see attached pic for fast recovery diode 1N4937 ) so the high voltage trapped between the diode and the drain of the mosfet is actually beneficial to a certain degree. No need to drain it off...

Actually... The Mosfet leaks, and a diode leaks as well, hence the high reverse voltage diminishes over time, and introduces a decaying current into the Rx, mimicking a static target. OK, it gets ironed out in subsequent circuitry, but it is there. In a way it resembles the "air signal" of VLF

[crane]

The high reverse voltage on the mosfet drain and diode cathode does diminish over time but it is still at a high value during sampling time and up until the next pulse begins so it shouldn't introduce a decaying current in the RX.

A SF VLF's air signal alters with the GB control and depends on if the demod window sees more of the positive or negative portion of the sine wave, and the filtered receive signal can also swing wildly negative when the GB is set to a particular setting and the coil is swept over some rocks (ferrite).

Ideally a PI air signal is always the same (0v) and doesn't vary with the GB setting and the audio can't swing negative in normal use.

[moodz]

Actually... The Mosfet leaks, and a diode leaks as well, hence the high reverse voltage diminishes over time, and introduces a decaying current into the Rx, mimicking a static target. OK, it gets ironed out in subsequent circuitry, but it is there. In a way it resembles the "air signal" of VLF

True ... the mosfet leaks anyway just with the TX supply voltage across it and the leakage is practically flat in most cases at any voltage .... however variation of leakage with temperature of the mosfet is a very different matter .. the leakage varies quite markedly with temp ... not that this should matter but you can hear it in a sensitive PI as it warms up.

[bbsailor]

Hi BB ... the diode capacitance is also minimal at higher reverse voltages ( see attached pic for fast recovery diode 1N4937 ) so the high voltage trapped between the diode and the drain of the mosfet is actually beneficial to a certain degree. No need to drain it off.
[ATTACH]33289[/ATTACH]

My patent AU2013101058



shows you how to damp a coil in 1 TC or better ...improvement over 5 TC approx with a resistive damper.

moodz.

PS nevertheless your article on fast coils is very well written/informative and I enjoyed reading it.

Moodz is talking about the coil discharge TC which is measured by dividing the damping resistor (or the effective damping value for an active circuit) into the coil inductance. Thus, a 300uH mono coil that has a 700 ohm damping resistor will have an effective damping value of 700 ohms in parallel with 1000 ohms (input resistor value) while the clamping diodes conduct above about 0.6 volts. This value is 411.76 ohms or 300/411.76 for a TC of .7285 uS down to 0.6 Volts and then a steeper curve of only the damping resistor or 300/700 or .4285 uS down to zero. Theory states that from a target TC perspective a pulse that turns off 5 times faster than the TC of the target generates the maximum amount of eddy currents in that target. When trying to find very small gold nuggets with very low TCs switching to the RX mode as fast as possible will catch the quickly decaying eddy currents in these fast decaying targets.

Once you get your PI machine to sample in this low uS delay range you face another problem. If the coil wire holds lingering eddy currents the small targets will be missed so special wire needs to be used to avoid eddy currents being detected in the coil wire itself. This is why Litz wire or tin-plated stranded wire is used to only detect the decaying eddy currents from the desired small gold nugget targets. For those looking for coins at the beach or occasional gold jewelry having a minimum delay of 10uS is OK as below about 15 uS you start detecting the salt wet beach sand. Different beaches cause this to start in different delay ranges but hunting on the dry portions of the beach you can usually use a delay setting near 10 uS but need to back off the delay as you approach the wetter sand.

Those hunting for gold in the outback of Australia face the problems of highly mineralized soil that tends to mask the small gold targets sought. That is why Moodz patented his unique and creative approach to quench the flyback pulse as fast as possible using an active MOSFET to suck up this flyback energy and allow the coil to turn on the RX mode as fast as possible and allow a wider flexibility in the type of coils that can be used to offset the mineralization problem.

The bottom line: sampling fast requires a coil that can have a lower eddy current decay time than the targets sought. Low TX circuit capacitance and low noise in the front end is a common place to start optimizing before you look elsewhere.

Joseph J. Rogowski

[Davor]

... almost.
moodz' patent is about replacing the damping resistor with a controlled passive current source. So for high voltages at such synthetic damping resistor, it behaves as a current source with equivalent resistance approaching infinity. As the voltage drops, also the resistance drops, so for small voltages it appears as a normal damping resistor. Because the initial resistance values are high, the flyback drops fast, but it also overshoots, and quenching of the flyback is still faster than with a normal resistor. It also rises the flyback voltage, and may prove snubber an useful addition.
moodz have shown it can work well with actively controlled MOSFETs, but when that principle is applied with just about any kind of constant current sink/source the very same effect is observed, as I've seen in simulations. Unlike moodz' solution that is digitally controlled and very predictable, other components with self-regulating behaviour are seldom that much predictable, so moodz' solution is a much better candidate for steady production.
Of course, a fast coil is a prerequisite for proper operation