Analysis of PI Detector Main Sample Delay

[Qiaozhi]

KingJL has previously stated that referencing the main sample time delay, from the point where the TX is turned off, is not a valid measurement for comparison between different PI detectors. This has the side-effect that some designs can apparently sample later than others but still have the ability to detect low conductivity targets. So what’s really going on?

Intrigued by the claim that the TX wave shape affects sampling time (potentially allowing the main sample to be taken later than normal, and still allow detection of low conductivity targets) I ran some simulations on both the Vallon and MPP circuits for comparison.

For both cases I used a 3us TC target with a coupling coefficient of 10m.

I used the Vallon simulation created by KingJL, but with some small changes. C2 was changed to 0.01uF, the capacitance of L1 reduced to 400pF, the initial condition associated with C5 reduced to 250V, the transient analysis modified to “.tran 0 5m 6m 1u” (to speed up the simulation), 3us TC target added, and preamp circuit included (but not used in this analysis). The first three changes were suggested by KingJL.

With the Vallon the maximum current in the target was reached @ 5.1us after TX-off with a value 61.2mA.
Value @ 10us = 21.1mA
Value @ 15us = 4.2mA
Value @ 20us = 808uA

With the MPP the maximum current in the target was reached @ 1.5us after TX-off with a value of 42.9mA.
Value @ 10us = 3.1mA
Value @ 15us = 604uA
Value @ 20us = 114uA

As you can readily see, the Vallon had considerably more current still flowing in the target at the specified sample times (referenced to TX-off) than the MPP, even bearing in mind that the energy stored in the Vallon TX coil is 750uJ, and the MPP stores 547uJ.

The only obvious difference I could detect in the two TX waveforms was that the Vallon MOSFETs [apparently] remain in the breakdown region for longer. However, KingJL pointed out that “Actually the Vallon never reaches MOSFET breakdown (the sim I sent had the IC for the 0.65uF cap (C5) between the High and Low side set way too high... now that the C2 is the right value, the IC for C5 should be 250 V which matches the real world Vallon. The high sink of the current is not the MOSFET breakdown but the charging of C5 and C1 & C15 or C2 & C15 (depending on the cycle) during flyback. The effect on coil current amounts to the same thing though. But the flyback never reaches beyond 250V in a working Vallon, well below the 400 V breakdown of an IRFR320.”

Clearly there is no magic here that allows low conductivity targets to be detected at later sampling times. But it does confirm the original statement that simply referencing the main sample time delay, from the point where the TX is turned off, does not allow a direct comparison between different detectors. The only true reference is the point where the target reaches maximum stimulation.

For example:
10us after maximum target stimulation [Vallon] = 3.9mA
10us after maximum target stimulation [MPP] = 1.9mA

And now we have a different picture.

Although the Vallon still has more current flowing in the target, the difference is nowhere near as dramatic as before. At the end of the day, there is no twisting of the laws of physics.

The problem seems to be that everyone assumes the reference point is TX-off. But defining the main sample delay at [say] 10us this way is not the same for all PI detectors, as the eddy currents in the target need to rise to a maximum value first before decaying, and the rise time depends on the shape of the TX waveform.

Discuss...

[waltr]

I have been reading the previous discussions by KingJL and others.
Your simulations/analysis is very interesting but haven't yet got my head around all the details.

If I remember the simulations the big difference in the Vallon (and doesn't the old Impulse do this) is the TX pulse 'falling' edge is much slower verse the sharp falling edge of the MPP (and other PI's).

[green]

I've been trying a spice simulation to compare different discharge times. Change V3 volts to change discharge time. Change R3 to adjust target TC. Maybe there is a better simulation? Maybe zero time could be when coil volts is 1/2 supply volts. With short TC targets the sooner you sample the better, but like KingJL posted Tx off doesn't define zero time.

[KingJL]

If I remember the simulations the big difference in the Vallon (and doesn't the old Impulse do this) is the TX pulse 'falling' edge is much slower verse the sharp falling edge of the MPP (and other PI's).

Remember, we are dealing with coil current, not TX pulse (although the TX pulse starts everything off)! The Vallon coil current falls off fairly linearly over a period of 5.5 usec - 6 usec. The impulse coil current falls off fairly linearly over a period of ~52 usec.

Vallon Coil Current:


Fisher Impulse Coil Current:

[KingJL]

Qiaozhi,
Playing around with a half sine current waveform, it would appear that the peak target voltage is moved out to the outer limit of the coil waveform (closer to the area that can be detected by the pre-amp. What would your analysis that was applied to the Vallon vs the MPP show with the half sine?

[dbanner]

From your analysis, it appears that the vallon coil current drops off sharply within about 6 uS, allowing sampling of low TC targets. But how far is this from the end of tx pulse?
Just reading along.
So the effective zero point should be where? Maximum stimulation(with target) or zero coil current(no target)?
Granted, the Vallon isn't your average p.i machine.

[Qiaozhi]

Qiaozhi,
Playing around with a half sine current waveform, it would appear that the peak target voltage is moved out to the outer limit of the coil waveform (closer to the area that can be detected by the pre-amp. What would your analysis that was applied to the Vallon vs the MPP show with the half sine?

I've been thinking about this, and my conclusion is that playing around with the TX wave shape does not provide any benefits with regards to detection of low conductivity targets. The target is maximally stimulated when the TX-off transition is as short as possible. Using slower transition times (or altering the shape of the waveform) only delays the time when eddy currents in the target reach their peak, but that peak will be a lower amplitude than for the short transition, and the associated TX coil current delay is also delayed by the same amount.

This is a difficult one to analyse, as detection by the preamp may appear to be improved, when in fact this is soley due to the preamp spending less time in saturation because of the lower target signal. So there's no free lunch unfortunately.

[bbsailor]

I've been thinking about this, and my conclusion is that playing around with the TX wave shape does not provide any benefits with regards to detection of low conductivity targets. The target is maximally stimulated when the TX-off transition is as short as possible. Using slower transition times (or altering the shape of the waveform) only delays the time when eddy currents in the target reach their peak, but that peak will be a lower amplitude than for the short transition, and the associated TX coil current delay is also delayed by the same amount.

This is a difficult one to analyse, as detection by the preamp may appear to be improved, when in fact this is soley due to the preamp spending less time in saturation because of the lower target signal. So there's no free lunch unfortunately.

According to theory, the maximum target stimulus comes from a TX pulse discharge time constant that is 5 times faster than the target time constant. The TX discharge time constant is determined by dividing the damping resistor value into the coil inductance. Here is where reducing total coil and TX circuit capacitance allows higher values of damping resistors to be used with the resultant effect of stimulating smaller targets better.

Joseph J. Rogowski

[Qiaozhi]

According to theory, the maximum target stimulus comes from a TX pulse discharge time constant that is 5 times faster than the target time constant.

Which is exactly why playing around with slower transition times, or altering the waveshape, simply makes matters worse. In fact, the Vallon PI conforms to this theory, as the damping resistor is not connected until later in the discharge curve. This allows the current to initially fall at a faster rate and provide more stimulus to the target.