PI holes

[Teleno]

The soil model is defined as:
which demonstrates the effect that different TX on-times have on the response.

is your gronud signal at the coil (variation of the magnetization).

The authors examine the following differential measurements:

For this study we considered the following four differential measurements:
1. Receiver height;
2. Transmitter eld direction;
3. Transmitter loop size; and
4. Transmitter waveform.

The latter is the technique Davour and green talk about in this thread:

Candy's system uses this equation to develop linear combinations of different integration windows that
cause the ferrite response to sum to zero. This methodology has been applied in the development of the
Minelabs F1A4 detector, which has been used successfully in the demining community.

This is also relevant:

Of all the differential measurements, we believe that modications to the transmitter waveform have
the most promise...
...
we have shown that differential measurements based on varying the transmitter waveform
are promising because the form of the time-domain response does not depend on the spatial distribution of
the magnetic soil. This is in contrast to the other differential measurements proposed in this chapter (receiver
height, transmitter eld direction) as they make (or rely on) assumptions regarding the spatial distribution of
the soil material.

It's an excellent paper, isn't it?

[Qiaozhi]

It's an excellent paper, isn't it?

Yes ... I hadn't seen that one before.

[Teleno]

Is the XL sheet green was playing with downloadable somewhere? I'm not able to locate it. Would like to try the "A" thing with his data.

[Davor]

Long pulses maximize ground effect. Is this desirable?

The property that is improved with long charging pulses is viscous ground response linearity in log-log scale. It enables unique GB solution for any target sample duration, where GB sample duration becomes a constant ratio against it.
There is no problem achieving a GB solution for a mutually subtracting pair of samples that have a perfect GB even in a curved log-log ground response, but the setting is unique for a specific set of sample duration that has to be readjusted for a different set of sample durations. You'd set a GB once, and by virtue of sample duration ratio, it should be valid for any and all sample durations.

Now imagine a setup where you have two different sets of sample durations. Of course they'll both have their holes, but at different taus. If you toggle between these quickly, and apply some clever play of sound pitch, you'd have 3 distinct tonal situations:
- deeeeeeee for targets with short taus and above a first hole
- dee-dah-dee-dah for targets with taus in between holes
- daaaaaaah for targets with long taus below the second hole

In effect you'd lose holes in detection, and instead you'd have a clever means of discrimination, and GB on top of everything.

[green]

If you pay a close attention to Figure 3.2: Effect of pulse length on the ferrite decay, you'll see precisely why you need long constant charge pulse to achieve perfect ground balance. Most probably this effect is responsible for different values (other than 1/t) for ground response reported in various sources. This effect is also seen in my spice model of viscous soil.

I've plotted ground response in the past. I think most of the time the slope has been around -1.28. I plotted some clay from the yard today with 50 usec and 100 usec on times. Record no target, record target(ground), subtract no target signal from target signal and chart with Excel. Got a -1.3 slope each time. Any suggestions why I get -1.3 instead of -1 for the slope or what I might try next? I tried the ferrite core that is easy to see with a IB coil during on time, but couldn't see a change today plotting the decay. The x axis is usec with external trigger, fet gate off (time zero).

[green]

I've plotted ground response in the past. I think most of the time the slope has been around -1.28. I plotted some clay from the yard today with 50 usec and 100 usec on times. Record no target, record target(ground), subtract no target signal from target signal and chart with Excel. Got a -1.3 slope each time. Any suggestions why I get -1.3 instead of -1 for the slope or what I might try next? I tried the ferrite core that is easy to see with a IB coil during on time, but couldn't see a change today plotting the decay. The x axis is usec with external trigger, fet gate off (time zero).

Another chart. Landscaping lava rock and a rerun of the clay at a lower scope full scale. Not quite the same slope, but still not near a slope of -1.

[Davor]

You may try limiting coil current with a resistor to any current you are comfortable with, and extend the charging period as the sampling period, and see what effective slope you get. I expect 1/t.

[green]

You may try limiting coil current with a resistor to any current you are comfortable with, and extend the charging period as the sampling period, and see what effective slope you get. I expect 1/t.

The data has more noise than I would like. The slope looks close to -1.3, same as the other runs, not-1. Things get hot and there is time between the two recordings. I think causing some of the problem.

[green]

Generated a ground simulation with a slope of -1. Time scale zero to 100 usec (.1 usec resolution). The sum of the samples 10 to 20 usec, 20 to 40 usec and 40 to 80 usec add up to the same value which they should. Summing the values in 10 usec steps and plotting don't plot a -1 slope. If I change the generated slope to -1.28 and sum they plot closer to a -1 slope. My math is rusty, maybe someone could check my results. Maybe how the data is sampled effects the result.

[Davor]

If there is any memory effect in a process, e.g. integration, yes, you'll experience that. Point is that integration when done in linear time affects more the initial samples.

[green]

:

My math isn't what it needs to be. I measured soil response at 50, 100 and 300 usec coil on times. What does the formula predict for those coil on times?

[Qiaozhi]

:

My math isn't what it needs to be. I measured soil response at 50, 100 and 300 usec coil on times. What does the formula predict for those coil on times?

According to the final report "UX-1355-FR-01", the magnetic soil model is 1/t (see page 18 ). But this is a theoretical soil model for homogeneous magnetic soil. Therefore, to reduce the number of geologic anomalies in non-homogeneous soil, they performed a fit of the median-filtered data set to the theoretical soil model and recorded the misfit results. This enabled them to ignore many of the magnetic anomalies while still allowing detection (and even sometimes enhancing) any metallic targets. In some cases the reverse was true, and certain metallic targets were less evident from the results. The equation in your quote is a modified magnetic soil model. This is a partial differential equation that quantifies the rate-of-change of the magnetic field response [H] with time, assuming all other variables remain constant. The constant A is an empirically defined fudge factor which allows the data measured for 4 different TX on-times to match the predicted response.

Basically, I don't think this equation is going to help you that much. Magnetic soil is supposed to have a response of 1/t; so [I suspect] either you have a measurement error, or your test soil too aggressive for the model.

[Teleno]

:

My math isn't what it needs to be. I measured soil response at 50, 100 and 300 usec coil on times. What does the formula predict for those coil on times?

As Qiaozhi said, it's the rate of change of soil magnetization. In practice it's the voltage you see at the detection coil.

The 1/t model only applies when previous magnetization of the soil is complete. Theoretically you'd have to turn on a pulse for a very long time and you'd see 1/t after turn off. You make t=0 at turn off.

In real life pulses have a finite duration and the signal is the subtraction of two "" curves shifted in time: and


An alternative is to use a Constant Current pulse and measure the response at turn-on, which is valid because the formula applies to both magnetization and de-magnetization At turn-on you have a ground that has been an "infinite time" at the same magnetization state, so the transition to the new state has a perfect 1/t rate of change. Here your t=0 is the moment of tuning the pulse on.

Can you post one of your XL files for me to download and try some maths on your values?

[green]

Thanks for the replies on the formula. I'll try to figure out how to post the XL files. Would like to find out why I keep getting a slope other than -1.

[Teleno]

Thanks for the replies on the formula. I'll try to figure out how to post the XL files. Would like to find out why I keep getting a slope other than -1.

If you want you can send them to my email: [email protected]