minimum sample delay

[green]

You are not even close with the impulse simulation. It's design relies heavily of the coil discharge energy being routed to some serious capacitance clamped to the input supply voltages. Here are simulation files for the basic TX concept. This one is using transistors that are available in LtSpice. I had models for the actual transistors used in the Impulse, but they are on a dead computer. You can play with the large capacitance and the coil values to change the coil current waveform anywhere from the triangular to a half sine. You can vary the coil however you like and adjust the supply voltage to get the desired coil current for test. If you over damp you will destroy short TC response. Base your total damp on the coil capacitance using the function R= SQRT(L/C)/2.

The impulse was introduced here, not because it provided the best current waveform for fast TC detection (I think the limit for it is probably about a 3 usec target) , but to show that the reference to TX switch-off was not really a valid measure to judge sampling. In actuality the best current waveform for your fast TC test is a trapezoidal current waveform with the rise time, the duration of the steady current, and discharge time tailored to your desired TC of detection. DaveJ, I believe, has designed a bi-polar TX that has roughly the following characteristics: fast ramp up (~5 usec) of current to about 1A (roughly 1/4 sine shape), near steady current for 40 usec, 5 usec coil disicharge (roughly 1/4 sine shape). If you take steps to insure all residual coil current is stopped at the zero point during discharge, you could theoretically sample at that time... but in the real world, theory does not equal reality. So you try to achieve as close as possible.

I think the discharge current is going thru the zener diodes like in my simulation, maybe I'm wrong.

[green]

One time where it might make sense to trigger on Tx off. Ground should chart a straight line on a log log chart. Does if trigger is Tx off, doesn't if zero time is after Tx off.

Toying around with the 3 states (increasing field, constant field, collapsing field, the rate of change, and the magnitude of change), you can come up with the optimum coil current waveform envelope to achieve maximum detection of your desired target. Then you can determine the optimum time within practical limits to sample the received target response. I have found constant rate vs constant current to give a higher signal if average coil current and Tx time is the same. Have you found different?

My error. Tx off probably wouldn't the place for zero time with the Impulse.

maybe not the best for any PI when charting data

[KingJL]

Tried am Impulse simulation with spice. Not the same but maybe close enough. With R3 1ohm(TC=1usec) Impulse has less signal. With R3 .005 ohms(TC=200usec) Impulse has more signal.

You are not even close with the impulse simulation. It's design relies heavily of the coil discharge energy being routed to some serious capacitance clamped to the input supply voltages. Here are simulation files for the basic TX concept. This one is using transistors that are available in LtSpice. I had models for the actual transistors used in the Impulse, but they are on a dead computer. You can play with the large capacitance and the coil values to change the coil current waveform anywhere from the triangular to a half sine. You can vary the coil however you like and adjust the supply voltage to get the desired coil current for test. If you over damp you will destroy short TC response. Base your total damp on the coil capacitance using the function R= SQRT(L/C)/2.

The impulse was introduced here, not because it provided the best current waveform for fast TC detection (I think the limit for it is probably about a 3 usec target) , but to show that the reference to TX switch-off was not really a valid measure to judge sampling. In actuality the best current waveform for your fast TC test is a trapezoidal current waveform with the rise time, the duration of the steady current, and discharge time tailored to your desired TC of detection. DaveJ, I believe, has designed a bi-polar TX that has roughly the following characteristics: fast ramp up (~5 usec) of current to about 1A (roughly 1/4 sine shape), near steady current for 40 usec, 5 usec coil disicharge (roughly 1/4 sine shape). If you take steps to insure all residual coil current is stopped at the zero point during discharge, you could theoretically sample at that time... but in the real world, theory does not equal reality. So you try to achieve as close as possible.

[green]

Tried am Impulse simulation with spice. Not the same but maybe close enough. With R3 1ohm(TC=1usec) Impulse has less signal. With R3 .005 ohms(TC=200usec) Impulse has more signal.

[green]

in anything I build, I always include a 0.1 ohm resistor on the low side of the coil circuit that is accessible for measurement across it. Once you can determine that point, you can make decisions about the excitation and detection of the target that will minimize the things that detract from detection. A couple of things to consider (take away all the math and use a simplified explanation): (1) eddy currents are produced in one direction when the the magnetic field is expanding and the opposite direction when the magnetic field is collapsing (eddy currents subside in a static magnetic field), (2) induced eddy currents, in effect, have an inertia due to and somewhat proportional to the inductive qualities of the target. This "inertia" manifests itself as the time constant of the target. Having "inertia", the eddy currents lag the excitation force of the changing magnetic field... the amount of lag again manifests itself as the time constant. (3) The more (up to a point) you can separate the increasing field from the collapsing field (in other word separate these by a period of static or near-static field strength), you can maximize the target signal that is left for detection by minimizing the cancellation effect of the induced eddy currents (the eddy currents subside between the state change). Toying around with the 3 states (increasing field, constant field, collapsing field, the rate of change, and the magnitude of change), you can come up with the optimum coil current waveform envelope to achieve maximum detection of your desired target. Then you can determine the optimum time within practical limits to sample the received target response.

I've been thinking we are discussing what we call zero time when we compare or present data. Do you sample your data with an operating PI based on coil current reaching 1ma instead of Tx off. I would have thought that wouldn't work,
if that is what you are doing and it works better I will give it a try.

[green]

One time where it might make sense to trigger on Tx off. Ground should chart a straight line on a log log chart. Does if trigger is Tx off, doesn't if zero time is after Tx off.

Toying around with the 3 states (increasing field, constant field, collapsing field, the rate of change, and the magnitude of change), you can come up with the optimum coil current waveform envelope to achieve maximum detection of your desired target. Then you can determine the optimum time within practical limits to sample the received target response. I have found constant rate vs constant current to give a higher signal if average coil current and Tx time is the same. Have you found different?

[KingJL]

How do you determine when coil current has reduced to 1mA?

in anything I build, I always include a 0.1 ohm resistor on the low side of the coil circuit that is accessible for measurement across it. Once you can determine that point, you can make decisions about the excitation and detection of the target that will minimize the things that detract from detection. A couple of things to consider (take away all the math and use a simplified explanation): (1) eddy currents are produced in one direction when the the magnetic field is expanding and the opposite direction when the magnetic field is collapsing (eddy currents subside in a static magnetic field), (2) induced eddy currents, in effect, have an inertia due to and somewhat proportional to the inductive qualities of the target. This "inertia" manifests itself as the time constant of the target. Having "inertia", the eddy currents lag the excitation force of the changing magnetic field... the amount of lag again manifests itself as the time constant. (3) The more (up to a point) you can separate the increasing field from the collapsing field (in other word separate these by a period of static or near-static field strength), you can maximize the target signal that is left for detection by minimizing the cancellation effect of the induced eddy currents (the eddy currents subside between the state change). Toying around with the 3 states (increasing field, constant field, collapsing field, the rate of change, and the magnitude of change), you can come up with the optimum coil current waveform envelope to achieve maximum detection of your desired target. Then you can determine the optimum time within practical limits to sample the received target response.

[green]

I have wrestled with the refernce point for well on to 2 years now. For it to have any meaning, "sample time" must have the same reference from circuit to circuit. Personally, I have come to using the point to where coil current has decreased to 1 mA, for my personal comparisons. That way, it takes into account the actual removal of excitation of the target and takes away some of the mystery of why one particular detector appears so sensitive to fast tau targets and another is not.

How do you determine when coil current has reduced to 1mA?

[KingJL]

... I've always referenced to coil off command thinking that made the most sense, maybe not. If there is a better way, I'm all for it.

I have wrestled with the refernce point for well on to 2 years now. For it to have any meaning, "sample time" must have the same reference from circuit to circuit. Personally, I have come to using the point to where coil current has decreased to 1 mA, for my personal comparisons. That way, it takes into account the actual removal of excitation of the target and takes away some of the mystery of why one particular detector appears so sensitive to fast tau targets and another is not.

[green]

The focus on the impulse is probably dragging Green's thread a little off track. The reason I introduced it was to emphasize two points. (1) "minimum sample delay" to what reference, and (2) a fast coil circuit is not the only way to achieve detection of small resistive targets... there are other methods. I think the discussion of small fast tau target detection and the focus on the religeous need for ultra fast coils actually deserves it's own thread. There are a lot of moving parts to the physics that need to viewed and understood as an integral and not in isolation. There is no "holy grail"!!

If the objective is to measure the TC of AWG28 solid wire and it can be done without a fast coil circuit no problem. I've always referenced to coil off command thinking that made the most sense, maybe not. If there is a better way, I'm all for it.

[rickb]

Have a read of this article. It changed my thinking on using expensive Litz wire.

https://engineering.dartmouth.edu/in...s/stranded.pdf

If the bottom line is cost, and you're making thousands... Given the time required to make a quality coil, it makes no sense to use ersatz litz wire. If the product is a cheap switching supply made in china, then oxidized stranded copper may be good enough.

[KingJL]

The focus on the impulse is probably dragging Green's thread a little off track. The reason I introduced it was to emphasize two points. (1) "minimum sample delay" to what reference, and (2) a fast coil circuit is not the only way to achieve detection of small resistive targets... there are other methods. I think the discussion of small fast tau target detection and the focus on the religeous need for ultra fast coils actually deserves it's own thread. There are a lot of moving parts to the physics that need to viewed and understood as an integral and not in isolation. There is no "holy grail"!!

[KingJL]

Thanks Kingjl, is it this circuit ? always been curious about it, DaveJ has said a few times its worth investigating.

That is the circuit! Note in the bottom right in the info block "FISHER - IMPULSE".

[6666]

Thanks Kingjl, is it this circuit ? always been curious about it, DaveJ has said a few times its worth investigating.

[KingJL]

The reason is the TX/coil current waveform... The detector is the Fisher Impulse which I believe was designed by Dave Johnson (Dave J). That is the problem I have with referencing sample time relative to TX-off. In my opinion, sampling needs to be referenced to the trailing boundary of the coil current waveform envelope. If you reference the sample timing to the trailing boundary of the coil current waveform enveope, the sample time for the Impulse becomes ~15 usec.

A view of the Impulse coil current in relation to TX pulse.



The impulse is a bipolar TX. By referencing the sample period to the trailing edge of the coil current envelope, the sample delay is < 15 usec (closer to 5).