Ground Balance Theory

[Davor]

I can’t believe that it is still not evident to some of you that the flyback pulse is the one that kicks the target.

That's nicely put

[Davor]

TX pulse is ONLY thing that energize metallic target, flyback is just side effect, needed to quickly remove energy stored in circuit.

This is a bit misleading. The flyback pulse is certainly a consequence of coil charging, but it can be produced by other means that are no less effective. Step pulse for example. Works the same.

So it is not a tail wagging the dog, but the other way around. There are also fully functional dogs without tails.

Extremely simplified view: consider a PI machine to be a simple transformer - what phenomenon transformers convey - voltage or current?

[PiTec]

Tepco, just build the CC circuit linked on Eric’s PI Technology Forum 11 years ago (or a similar circuit), and separate the driving pulse from the flyback pulse. You can see live on the scope how the target response increases when the flyback pulse is not weakened by the driving pulse. Or just simulate it with LTspice.

http://www.findmall.com/read.php?34,129617
www.tb-electronic.de/pi_tech/pulse_cc_circuit.pdf

Anyway, I will no longer spend time to convince anyone. Find out yourself or keep believing in the power of the driving pulse

[Tepco]

There is nothing about “conviction” or something, just experience gained in some tests. Somewhere in this thread I described simple CC (but this time, constant peak current control) test circuit, capable to produce variable pulses, ranging from few hundred uS, down to less than 500nS, and keeping total energy per pulse (1\2xLxIxI) about constant. As a result, flyback amplitude and shape is fairly constant over very wide range of TX pulse width. Observed on scope, any difference is indistinguishable, until DC coil voltage + flyback voltage combined is below avalanche voltage. In some conditions, using 1200V transistor and bit smaller peak current, up to 800V coil voltage and less than 500nS TX pulses are produced. Only deviation from “constant flyback” shape and energy is actually at very wide TX, circuit operating with only few volts col voltage, then nonlinear transistor output capacitance may have bit more influence. Otherwise, flyback time and shape can be considered quite constant and independent of actual TX time. Now, only varying TX time, I recorded response of different metallic objects, and find that it is variable, depending on object TC and applied pulse width. As theory predict, pulses shorter than TC produce less and less response, longer pulse, slightly longer than TC will not increase response above some point. Considering flyback conditions are identical to any practical meaning in this test, result can be attributed to TX drive pulse only. Note that soil, or ferrite behave bit differently.


Circuit I used is based on constant peak coil current control (not constant current source!!!) so linear current ramp-up time is controlled by applying variable coil voltage. I will comment more on your CC version tomorrow (sorry, have to go offline now) and results, your approach is quite different. My intention is not to advocate some “theory” or to “wage some war” here, ( I have to admit his is interesting part of forum life too, actually sometimes I missing it) but rather to learn something and exchange experience. Will be interesting, for example, to see how we, attempting about same thing, using different means, ended up with completely opposite conclusions. Also, soon I will be back to my hardware again, so building setup for any proposed “acid proof” test or something similar will be no problem anymore.

[Tepco]

This is a bit misleading. The flyback pulse is certainly a consequence of coil charging, but it can be produced by other means that are no less effective. Step pulse for example. Works the same.

So it is not a tail wagging the dog, but the other way around. There are also fully functional dogs without tails.

Extremely simplified view: consider a PI machine to be a simple transformer - what phenomenon transformers convey - voltage or current?

Lovely, so let's consider 1x1m loop detector running at few hundred uS pulse width, detecting some pile of rusted junk underneath. Now, if I fire some 2uS pulse in that same coil (by means of suitable hardware), containing exactly same energy, number of miliJoules, as original pulse, will I get same target response? Never going to happen.

[deemon]

In reality , we need two things if we want to get a proper PI target response :

1. Magnetize the target properly . I mean that we need to "turn on" a magnetic field and wait enough time for the field to permeate the target completely . This time depends on the target size , its conductivity , shape , etc .

2. Change the field rapidly . When we turn the field off , we'll start the eddy currents in the target . It looks like the target tries to "hold" the magnetic field level has been established in the target before . Those currents will decay by the exponential law , and this exponential decay is what we call "target response" . Of course we may reverse the field instead of removing it , just quickly change its polarity ... This would double the target response , but the principle is just the same .

So we can easily understand that only the fast change of the field is the real factor that produces the target response , but the voltage spike ( flyback pulse ) is the only a "side effect" of the transmitter coil , and nothing more .

[Alexismex]

Hello Deemon,always good and clear informacion thank you very much,
Now
I want to tell you another factor that i think exist :
I have a good test field (5 years now) where i test my PI detector the target go from 60 cm to 3 meter , i have test more that 50 detectors always of my design, i have noted a strange phenomena
( in the test to detect a 200litters tank at 3meters) THAT sometimes i will detect perfectly moving the coil East / West above the target and more difficult moving the coil North / south
in one day but another day is the contrary , i detect the target very well North/ South and with more difficulty East /west, this i have repeated many times from years to years , and i Know very well my detector for make confirmation that the phenomena exist ...i understand it is in relation with the electromagnetic field of the earth but why sometimes it reversed ???

[Aziz]

Hi all,

I have continued my numerical best fit simulations and found interesting facts, we can apply to our latest measurements. I have even added some noise to the numerical simulations.

The formula 17 in the mentioned scientific paper can be extended to detect different exponents b. But not phase shifts or time base corrections p. So the time code in the measurement data must be correct.

The formula 17 gets into the form, when adding exponent b:
G(t) = a*( t^b - 1/(t+w) ), (p set 0 and simplified the equation)

The Excel Solver works quite stable to find the parameters a, b and w. It seems, that in this case only one optimal solution is possible (no local minima). That's a good news.

But if your time base isn't correct (if you set your time code t0 at 0 µs but in reality it is time shifted like t0 = +1 µs) then we get totally inconsistent & not reliable model parameters. (see first pic).


If we set exponent at b=-1 (1/t term gets valid in formula 17), then we can detect phase shifts and time base correction offsets (see second pic).


But this wouldn't make sense. Better we eliminate p with correct time base and look at the exponent b variation. And we can test, whether the VRM formula fits well into the measurments.


Cheers,
Aziz


PS: The exponent b calculation is quite more stable and reliable than other parameters. We can compare the b=-1 best fit against the calculated b in the measurement data. And this is going to be very interesting, which one wins the run (with less fitting errors).

[Aziz]

Ok guys,

Thomas measurement data is indicating a severe time base error (not consistent). We won't get reliable VRM model parameters. No chance.

I suggest, he makes his measurements once again, where t0 = t=0 = switch-off time = trigger time position for data acquisition.
The f$in switch-off time t0 is (time code = 0 µs), when the f$in mosfets has been commanded to switch-off.
Not when the flyback period ends. That is very very important.

Don't make any changes to the original time code. Never ever!!! If you sample later, the time code will be greater of course. It should be 0 at trigger point = switch-off time = t0 = t=0. You can even sample earlier where the earlier samples will have a negative time code. At at switch-off time the time code should be 0 µs.

Now I'm looking forward to more accurate measurements and results.
Cheers,
Aziz

[PiTec]

Hi Aziz,

no need to repeat the measurements. If you want t0 to be at the end of the driving pulse, simply add the durations of the flyback pulses to the first time codes:

Dacay_curves_03: Flyback pulse = 2 µs -> first time code = 7+2 = 9 µs
Dacay_curves_04: Flyback pulse = 2 µs -> first time code = 8+2 = 10 µs
Dacay_curves_05: Flyback pulse = 10 µs -> first time code = 9+10= 19 µs

Of course the curves are no longer straight then …

[Aziz]

Hi Aziz,

no need to repeat the measurements. If you want t0 to be at the end of the driving pulse, simply add the durations of the flyback pulses to the first time codes:

Dacay_curves_03: Flyback pulse = 2 µs -> first time code = 7+2 = 9µs
Dacay_curves_04: Flyback pulse = 2 µs -> first time code = 8+2 = 10µs
Dacay_curves_03: Flyback pulse = 10 µs -> first time code = 9+10= 19µs

Of course the curves are no longer straight then …

Thomas, do you recognize now, why I got totally confused?

How do you exactly know, when the flyback has been really finished?
Please, please, please, make your measurements once again at the right trigger point t0 this time.
We will see much clearer, accurate and consistent results this time.
Cheers,
Aziz

[PiTec]

I do not understand your problem … I can set the duration of the flyback pulse to any value by simply adjusting the flyback voltage while measuring the flyback pulse width with the scope. I described this in posts #545 and #549:

TX timing: 50 µs TXon (10 volts), 2 µs TXoff (flyback, determined by 250 volts avalanche breakdown of the MOSFET).
10µs TXoff are achieved by clamping the flyback voltage to 50 volts. 500µs TXon consist of 50µs current ramp followed by 450µs constant current to keep dI/dt at zero.

With the scope timebase set to 1 µs / div, I measure the time between trigger (2) and the end of the flyback pulse (3) directly at the TX coil. The end of the flyback pulse is t=0 (t0). The error should be quite small, maybe +/- 0.25 µs. Any delay in the driver and MOSFET would not be added.


Before we continue we should find an agreement concerning the role of the flyback pulse …

[Aziz]

Thomas,

do you have a multi-channel oscilloscope with seperate trigger input?
Do the following please:
Trigger input: TX mosfet switch logic wave form, triggered at switch-off (where the data acquisition begins, has the first data sample the time code 0?)
Ch-1 input: pre-amp output (the response signal, which we interested in)
Ch-2 input: TX coil voltage (clipped to +10/-10V or less, we don't want to see the high flyback voltage, we want to see the time lag between switch-off command and the real TX coil switch-off, but don't screw your oscilloscope!!!)

I can at least see using ch-2 data, when the f*in TX coil will be switched off and correct the acquisition time code t0 by myself.

The VRM decay calculations rely on the correct time base/time code. Unless they aren't reffered to t0=switch-off time, we can't make accurate VRM modelling anymore.

Aziz

[Davor]

...Never going to happen.

As Einstein said ... Problems cannot be solved by the same level of thinking that created them.
If you can't separate a pulse from the charging in your head, you are bound to thinking that charging is somehow essential for target response. Think again.

You can produce a simple experiment that will include charging (so that you are happy) but make it completely irrelevant for the outcome (WTF!) and see that a flyback pulse devoid of the charging period works the same. It goes like this:
- place a resistor to limit current to your coil at some tolerable value,
- make a charging period abnormally long, so that it can't influence the outcome,
- fire a flyback and behold - a target response.

By forcing a constant current to your coil, it behaves like a permanent magnet, and to any target it is not unlike Earth field, but a flyback does produce a response.

[Tepco]

As Einstein said ... Problems cannot be solved by the same level of thinking that created them.

Works for problems created by humans, not nature.


What you propose is insanely long charging time, so it will make any effect irrelevant to outcome. What I want to say is, if you compare response from, say, 1, 10 and 100uS “charging time” containing same amount of energy, applied to same object, and with same flyback release, you will see difference, very clearly.