A
Anonymous
Guest
Since nobody's paying me to do PI these days, and since I'm not rich enough to be messing around with patenting stuff, I may as well post this in a public place where people who may be interested will see it.
DESIGN FOR A PULSE INDUCTION METAL DETECTOR WITH POWER EFFICIENCY COMPARABLE TO VLF
This design requires an induction balance searchcoil. Values of voltage, timing, inductance, etc. are merely examples for purposes of discussion. The transmitter coil is described as being single-ended, but it could be driven double-ended.
This is not the system described in the "Fisher patent". However, the "Fisher Patent" provides good background for understanding this system.
TRANSMITTER
The "hot" side of the transmitter coil can be switched to +50 volts, to -50 volts, or to ground (shorting it out). The timing sequence of the transmitter comprises four phases: 10us to +50v, 90us to ground, 10us to -50v, and 90us to ground. Thus, alternating polarity pulses are transmitted with a fundamental period of 200us (= 5kHz).
Suppose that the coil has a Q of 4 at 5 kHz, and that its inductance is 500 uH. The current waveform will be as follows:
Phase I: 50 volts across 500 uH for 10us causes a change in current of one ampere in the positive direction.
Phase II: because the coil is (nearly) shorted out, the current continues to flow, decaying somewhat because of the finite Q of the coil.
Phase III: there is a change in current of one ampere in the negative direction.
Phase IV: the current continues to flow, decaying somewhat.
Note that there is a reversal of current during the transmit pulses (Phases I and III). During the 4 microseconds or so prior to the current reversal, the energy stored in the coil is returned to the power supply. Also note that the mean current is approximately +/- 0.5 ampere, since there is only a 1 ampere change between Phase II and Phase IV (and vice versa).
Also note that what's going on in this transmitter coil is completely different from conventional (even Fisher Impulse) practice. The flyback period is not a short pulse intended to kill the coil current-- it's a long period intended to sustain the current.
Note also that the current waveform is a distorted 5 kHz square wave.
In conventional PI practice the transmit pulse is a waste of energy from the target's point of view-- it's the flyback pulse that kicks the target. In this new system, the transmit pulse is what kicks the target.
Since the transmitter coil always has current flowing in it and is always shorted out to something, it cannot do double duty as a receiver coil. Therefore a separate receiver coil, preferably in a condition of induction balance relative to the transmitter, is necessary.
POWER CONSUMPTION
The transmit duty cycle is 10%. If the Q of the transmitter coil were low, say less than unity, then during each pulse the current would ramp up linearly from zero to 1 ampere, with an average current of 0.5 ampere. Multiplied by the 10% duty cycle, this is a current drain of 50 milliamperes. However, since the Q of the coil is 4, the direction of current reverses during the transmit pulse, with the net current being only 1/4 the full delta. Therefore the current drain from the 50 volt supply is 12.5 mA. That's 625 mW.
To put this in perspective, consider a 500 uH coil driven by a 5 kHz sinusoid with an RMS current of 0.5 ampere. The reactance is 16 ohms, so the RMS voltage must be 8 volts. That's 23 volts P-P, pretty darn hefty drive. Given a Q of 4, the resistance would be 4 ohms, so the power consumption in the coil is I-squared-R, or 1 watt. The newfangled PI system didn't violate the laws of thermodynamics-- the sinusoidal component of its current at 5 kHz is actually somewhat less than 0.5 A RMS.
THE RECEIVED WAVEFORM
In the presence of a reactive offset, such as that produced by mechanical misalignment of the searchcoil or by the presence of low-loss magnetic minerals such as magnetite, the voltage induced into the receiver soil will approximately replicate the transmit voltage waveform. If we normalize the voltage induced during Phases I and III to 1 volt (induced by 100 milliamperes per microsecond) then the slope of the current during Phases II and IV (very approx. 1 mA/us) will induce a voltage on the order of 10 millivolts during Phases II and IV.
Resistive losses in low-conductivity targets will extend short "tails" from Phases I and III into Phases II and IV. In the case of high-conductivity targets and iron metal, the tails will be much longer, being truncated by the next transmit pulse.
Since the reactive signals are "piled up" in Phases I and III, gating the preamp off during these phases eliminates most of the reactive signal component, especially those of the higher frequency components. The suppression of reactive at the fundamental is roughly proportional to the Q of the transmitter at the fundamental.
In actual practice, the preamp would normally not be gated on until several microseconds past the ends of the transmit pulses. This allows time for the spikes induced into the receiver coil to decay, and helps eliminate response to salt water and conductive soils. ......If a very fast preamp is used, which recovers quickly and gracefully from saturation, then it may be operated ungated, with everything getting sorted out at the demodulators, in accordance with conventional PI practice.
USING THE REACTIVE SIGNALS
Since this system has residual reactive signal components esp. at the lower frequencies, it will be desirable to demodulate the primary reactive signal so it can be used for ground balancing. If the preamp is both fast and linear, the reactive signal can be demodulated in its respective time-slots, corresponding approximately to Phases I and III. Otherwise, a separate low-gain amplifier can be furnished for feeding the reactive demodulator. It may be desirable to gate this amplifier on during the transmit pulses and off at other times.
DEMODULATORS
As is well known in PI practice, it is necessary to do full wave demodulation in order to cancel earth field pickup, even if the desired signals are unipolar.
This new system produces a bipolar symmetrical signal which is intrinsically well-suited for full wave demodulation.
Timing of the demodulator switches would presumably follow conventional PI practice-- long delays or wide gating periods for high conductivity targets, short delays with short gating periods for low conductivity targets.
RECEIVER COIL RESONANCE
In order to provide high sensitivity, the receiver coil will ordinarily have more turns than the transmitter, resulting in higher inductance and lower self-resonant frequency. Self-resonance is undesirable, but its effects can be minimized.
The conventional method of dealing with self-resonance is to provide a parallel damping resistor which is always in the circuit. In many applications, this method will suffice.
Where the period of self-resonance is of the same order of magnitude as the transmit pulse duration, self-resonance can be put to good use.
Consider the case where the receiver coil is self-resonant at 100 kHz, having a period of 10 microseconds. When kicked by a step function, it will ring with a period of 10 us. Now suppose that 10 us later we kick it again with another step function that goes in the opposite direction. (This sequence is equivalent to our 10 microsecond transmit pulse.) By linear superposition, the ringdown induced by the second step function will cancel the ringdown induced by the first step function. .....Then, and only then (several microseconds later, giving the receiver circuit time to know it's been kicked again), do we switch the damping resistor into the circuit. Since the resonance of the receiver was synchronized to cancel the reactive signals induced by the transmit pulse, the damping resistor doesn't have much left to dampen.
The foregoing explanation is a gross oversimplification. To get optimum cancellation it will be necessary to fiddle with resonant frequency, damping during the transmit pulse, damping after the transmit pulse, and delay before kicking in the main damping resistor.
It's possible this fancy trick may interfere with effective demodulation of reactive signals.
SOMETHING THAT MAY NOT BE INTUITIVELY OBVIOUS
In a conventional (including Fisher) PI, the transmit current is brought to zero, and then the receiver looks at the signals induced in the coil. This is true even if the coil is of the induction balance type.
In this new design, during the period that the receiver is turned on, large currents are flowing in the transmitter coil. Intuitively it might seem that large voltages would therefore be induced in the receiver, but this is not the case.
The voltage induced in the receiver by the transmitter via loop imbalance or magnetite is proportional to the rate of change of transmitter current. If the current is large, but its rate of change is zero, no voltage is induced. In fact the reason that conventional PI's don't couple transmitter into the receiver is not because the transmit current is zero (although it is), but because its rate of change is zero.
If the Q of the transmitter coil were infinite, the current during Phases II and IV would be constant, and there would be nothing induced into the receiver by imbalance. However, at the fundamental frequency, the Q will usually fall within the range of 4 to 15 in order to take advantage of the power efficiency made possible by this scheme, while at the same time keeping the weight of the searchcoil reasonable. So, although some reactive signal will be induced at the fundamental, it will be several times less than what would have been induced in a comparable VLF design.
QUESTIONS------------
1. Has it already been done?
2. Has it already been patented?
3. Is there some fatal flaw in the scheme?
-------------------------------
David E. Johnson
Prescott, Arizona
29 December 01
DESIGN FOR A PULSE INDUCTION METAL DETECTOR WITH POWER EFFICIENCY COMPARABLE TO VLF
This design requires an induction balance searchcoil. Values of voltage, timing, inductance, etc. are merely examples for purposes of discussion. The transmitter coil is described as being single-ended, but it could be driven double-ended.
This is not the system described in the "Fisher patent". However, the "Fisher Patent" provides good background for understanding this system.
TRANSMITTER
The "hot" side of the transmitter coil can be switched to +50 volts, to -50 volts, or to ground (shorting it out). The timing sequence of the transmitter comprises four phases: 10us to +50v, 90us to ground, 10us to -50v, and 90us to ground. Thus, alternating polarity pulses are transmitted with a fundamental period of 200us (= 5kHz).
Suppose that the coil has a Q of 4 at 5 kHz, and that its inductance is 500 uH. The current waveform will be as follows:
Phase I: 50 volts across 500 uH for 10us causes a change in current of one ampere in the positive direction.
Phase II: because the coil is (nearly) shorted out, the current continues to flow, decaying somewhat because of the finite Q of the coil.
Phase III: there is a change in current of one ampere in the negative direction.
Phase IV: the current continues to flow, decaying somewhat.
Note that there is a reversal of current during the transmit pulses (Phases I and III). During the 4 microseconds or so prior to the current reversal, the energy stored in the coil is returned to the power supply. Also note that the mean current is approximately +/- 0.5 ampere, since there is only a 1 ampere change between Phase II and Phase IV (and vice versa).
Also note that what's going on in this transmitter coil is completely different from conventional (even Fisher Impulse) practice. The flyback period is not a short pulse intended to kill the coil current-- it's a long period intended to sustain the current.
Note also that the current waveform is a distorted 5 kHz square wave.
In conventional PI practice the transmit pulse is a waste of energy from the target's point of view-- it's the flyback pulse that kicks the target. In this new system, the transmit pulse is what kicks the target.
Since the transmitter coil always has current flowing in it and is always shorted out to something, it cannot do double duty as a receiver coil. Therefore a separate receiver coil, preferably in a condition of induction balance relative to the transmitter, is necessary.
POWER CONSUMPTION
The transmit duty cycle is 10%. If the Q of the transmitter coil were low, say less than unity, then during each pulse the current would ramp up linearly from zero to 1 ampere, with an average current of 0.5 ampere. Multiplied by the 10% duty cycle, this is a current drain of 50 milliamperes. However, since the Q of the coil is 4, the direction of current reverses during the transmit pulse, with the net current being only 1/4 the full delta. Therefore the current drain from the 50 volt supply is 12.5 mA. That's 625 mW.
To put this in perspective, consider a 500 uH coil driven by a 5 kHz sinusoid with an RMS current of 0.5 ampere. The reactance is 16 ohms, so the RMS voltage must be 8 volts. That's 23 volts P-P, pretty darn hefty drive. Given a Q of 4, the resistance would be 4 ohms, so the power consumption in the coil is I-squared-R, or 1 watt. The newfangled PI system didn't violate the laws of thermodynamics-- the sinusoidal component of its current at 5 kHz is actually somewhat less than 0.5 A RMS.
THE RECEIVED WAVEFORM
In the presence of a reactive offset, such as that produced by mechanical misalignment of the searchcoil or by the presence of low-loss magnetic minerals such as magnetite, the voltage induced into the receiver soil will approximately replicate the transmit voltage waveform. If we normalize the voltage induced during Phases I and III to 1 volt (induced by 100 milliamperes per microsecond) then the slope of the current during Phases II and IV (very approx. 1 mA/us) will induce a voltage on the order of 10 millivolts during Phases II and IV.
Resistive losses in low-conductivity targets will extend short "tails" from Phases I and III into Phases II and IV. In the case of high-conductivity targets and iron metal, the tails will be much longer, being truncated by the next transmit pulse.
Since the reactive signals are "piled up" in Phases I and III, gating the preamp off during these phases eliminates most of the reactive signal component, especially those of the higher frequency components. The suppression of reactive at the fundamental is roughly proportional to the Q of the transmitter at the fundamental.
In actual practice, the preamp would normally not be gated on until several microseconds past the ends of the transmit pulses. This allows time for the spikes induced into the receiver coil to decay, and helps eliminate response to salt water and conductive soils. ......If a very fast preamp is used, which recovers quickly and gracefully from saturation, then it may be operated ungated, with everything getting sorted out at the demodulators, in accordance with conventional PI practice.
USING THE REACTIVE SIGNALS
Since this system has residual reactive signal components esp. at the lower frequencies, it will be desirable to demodulate the primary reactive signal so it can be used for ground balancing. If the preamp is both fast and linear, the reactive signal can be demodulated in its respective time-slots, corresponding approximately to Phases I and III. Otherwise, a separate low-gain amplifier can be furnished for feeding the reactive demodulator. It may be desirable to gate this amplifier on during the transmit pulses and off at other times.
DEMODULATORS
As is well known in PI practice, it is necessary to do full wave demodulation in order to cancel earth field pickup, even if the desired signals are unipolar.
This new system produces a bipolar symmetrical signal which is intrinsically well-suited for full wave demodulation.
Timing of the demodulator switches would presumably follow conventional PI practice-- long delays or wide gating periods for high conductivity targets, short delays with short gating periods for low conductivity targets.
RECEIVER COIL RESONANCE
In order to provide high sensitivity, the receiver coil will ordinarily have more turns than the transmitter, resulting in higher inductance and lower self-resonant frequency. Self-resonance is undesirable, but its effects can be minimized.
The conventional method of dealing with self-resonance is to provide a parallel damping resistor which is always in the circuit. In many applications, this method will suffice.
Where the period of self-resonance is of the same order of magnitude as the transmit pulse duration, self-resonance can be put to good use.
Consider the case where the receiver coil is self-resonant at 100 kHz, having a period of 10 microseconds. When kicked by a step function, it will ring with a period of 10 us. Now suppose that 10 us later we kick it again with another step function that goes in the opposite direction. (This sequence is equivalent to our 10 microsecond transmit pulse.) By linear superposition, the ringdown induced by the second step function will cancel the ringdown induced by the first step function. .....Then, and only then (several microseconds later, giving the receiver circuit time to know it's been kicked again), do we switch the damping resistor into the circuit. Since the resonance of the receiver was synchronized to cancel the reactive signals induced by the transmit pulse, the damping resistor doesn't have much left to dampen.
The foregoing explanation is a gross oversimplification. To get optimum cancellation it will be necessary to fiddle with resonant frequency, damping during the transmit pulse, damping after the transmit pulse, and delay before kicking in the main damping resistor.
It's possible this fancy trick may interfere with effective demodulation of reactive signals.
SOMETHING THAT MAY NOT BE INTUITIVELY OBVIOUS
In a conventional (including Fisher) PI, the transmit current is brought to zero, and then the receiver looks at the signals induced in the coil. This is true even if the coil is of the induction balance type.
In this new design, during the period that the receiver is turned on, large currents are flowing in the transmitter coil. Intuitively it might seem that large voltages would therefore be induced in the receiver, but this is not the case.
The voltage induced in the receiver by the transmitter via loop imbalance or magnetite is proportional to the rate of change of transmitter current. If the current is large, but its rate of change is zero, no voltage is induced. In fact the reason that conventional PI's don't couple transmitter into the receiver is not because the transmit current is zero (although it is), but because its rate of change is zero.
If the Q of the transmitter coil were infinite, the current during Phases II and IV would be constant, and there would be nothing induced into the receiver by imbalance. However, at the fundamental frequency, the Q will usually fall within the range of 4 to 15 in order to take advantage of the power efficiency made possible by this scheme, while at the same time keeping the weight of the searchcoil reasonable. So, although some reactive signal will be induced at the fundamental, it will be several times less than what would have been induced in a comparable VLF design.
QUESTIONS------------
1. Has it already been done?
2. Has it already been patented?
3. Is there some fatal flaw in the scheme?
-------------------------------
David E. Johnson
Prescott, Arizona
29 December 01
"> Seriously, I bought my first detector 40 years ago, but I'm sorry to say your name didn't "ring a bell." Of course, I never heard of Eric until a few years ago. Other than the Impulse, what other detectors did you design? Welcome to the forum!... Jeff