DEEPER PI DETECTION DEPTH

[simonbaker]

-SB,

here is one for you. This is a sine wave simulation.

The green trace is the coil current.

The 3 other traces are the response signals of 3 different targets.

Tinkerer

Looks interesting. Can you attach the LTSpice file?

Regards,

-SB

[Tinkerer]

Looks interesting. Can you attach the LTSpice file?

Regards,

-SB

Here is the same simulation, but the target response is summed to the TX current (green trace) wave form.
Note the change in signal amplitude as well as the phase shift.

Tinkerer

[Aziz]

To test the effect of turn-on settling, I ran a 100us pulse width with and without R=15 ohms in series with the coil. Without the R the VCC was 10V and with the R it was cranked to 38V to give the same peak current. Ergo, in both cases I had a 100us pulse width and about 2A peak current, the only difference was the slope of the current: flat top vs saw tooth.

I got about what I expected, a 10-15% increase in preamp Vout deflection with the flat-top current, both with low and high conductance targets. This sounds good, until you realize that a 15% voltage increase is about a 2% depth increase. Ergo, in a std PI depth is not greatly affected by the di/dt just before turn-off.

- Carl

Thanks for sharing your test results Carl.

Indeed, it doesn't make sense with the flat-top current. What a waste of battery power to maintain the flat-top current for a negligible depth increase.

Can we say now, that the switch-off time period kicks the target in a PI detector? More precisely, the flyback period and provided that, that the flyback period is shorter than the pulse-on period. (High frequency stimulation matters.)

Cheers,
Aziz

[Midas]

So Carl, no sign of the effect that shows up in Tinkerers post #93, which seemed to indicate a detrimental effect of a flat top for one of the targets?

Thats kind of what I was getting at Tinkerer when I asked for real scope shots, something that could confirm or not whether that the strange effects in post #93 isn't just an error in the sim..

[Tinkerer]

So Carl, no sign of the effect that shows up in Tinkerers post #93, which seemed to indicate a detrimental effect of a flat top for one of the targets?

Thats kind of what I was getting at Tinkerer when I asked for real scope shots, something that could confirm or not whether that the strange effects in post #93 isn't just an error in the sim..

Error in the Simulation?
I wonder.
So I re-did the simulation more precisely to Carl's numbers.
100us TX for both, 300uH for both 15 Ohm and 31V for flattop and 600 mOhm and 7V for sawtooth. Same damping at 500 Ohm. 2A coil current.

I attached the zipped LTSpice simulation files so you all can help me find the error.

3 different targets with TC's of 15us, 100us and 200us.

The 15us target comes out close with 13% more response for the flattop.
The 100us target has 41% more response.
The 200us target has 46% more response.

Where is the error?

Tinkerer

[simonbaker]

Error in the Simulation?
I wonder.
So I re-did the simulation more precisely to Carl's numbers.
100us TX for both, 300uH for both 15 Ohm and 31V for flattop and 600 mOhm and 7V for sawtooth. Same damping at 500 Ohm. 2A coil current.

I attached the zipped LTSpice simulation files so you all can help me find the error.

3 different targets with TC's of 15us, 100us and 200us.

The 15us target comes out close with 13% more response for the flattop.
The 100us target has 41% more response.
The 200us target has 46% more response.

Where is the error?

Tinkerer

Hi Tinkerer:

If you zoom in on the TX currents during discharge, you can see something went wrong in the sim -- there is an oscillation in the "flattop" case that is probably skewing your results. Didn't analyze any further as to why...

Regards,

-SB

[Tinkerer]

Hi Tinkerer:

If you zoom in on the TX currents during discharge, you can see something went wrong in the sim -- there is an oscillation in the "flattop" case that is probably skewing your results. Didn't analyze any further as to why...

Regards,

-SB

Thanks Simon,

the oscillation on the Flattop current wave form is due to the parasitic coil capacitance. This is a well known problem with traditional PI. This is the inter wire capacitance, the wire to shield capacitance and the cable capacitance. I have put it down as 200pf, but this is a very conservative figure. 500pf would not be exaggerated.

But why does it not appear on the saw-tooth? This seems to be something with the wire resistance that is not equal between the two sims, but I need to follow up on that.

Tinkerer

Edit on the above.
I just run it again, and it has now the same oscillation in both sims. The oscillation disappears when the Mosfet gate resistor is changed from 10R to 100R, or, in other words slowing down the switching a little bit.

[Qiaozhi]

Thanks Simon,

the oscillation on the Flattop current wave form is due to the parasitic coil capacitance. This is a well known problem with traditional PI. This is the inter wire capacitance, the wire to shield capacitance and the cable capacitance. I have put it down as 200pf, but this is a very conservative figure. 500pf would not be exaggerated.

But why does it not appear on the saw-tooth? This seems to be something with the wire resistance that is not equal between the two sims, but I need to follow up on that.

Tinkerer

Edit on the above.
I just run it again, and it has now the same oscillation in both sims. The oscillation disappears when the Mosfet gate resistor is changed from 10R to 100R, or, in other words slowing down the switching a little bit.

It appears to be connected with the flyback voltage. Try plotting the coil voltage and coil current in different plot panes. You'll see that the kink in the current waveform follows the peak of the flyback. It might be a problem with the mosfet model, and that could be confirmed by replacing it with an ideal switch. Also, the mosfet model does not include the breakdown voltage. This can easily be fudged into the simulation by putting a large voltage zener diode in parallel with the mosfet.

[Qiaozhi]

Simon - I took your simulation and hacked it around to create a comparison between a PI without a resistor in series with the coil (case 1), and another with a 15 ohm resistor in series (case 2).

In case 1, the charging period is 18us.
In case 2, the charging period is 100us.
At the time the mosfet is ready to be switched off, they both have the same current flowing in the coil.
Plot "No-res" and "With_res" in one plot pane, and "Target1" and Target2" in a second pane.
As you will see in case1, the eddy currents are fairly well established in Target1 when the mosfet switches off. These eddy currents need to be overcome, resulting in a lower target response when compared to case2, where the eddy currents in Target2 have essentially decayed almost to zero.

These results appear to agree nicely with the practical test performed by Carl.

[simonbaker]

Thanks Simon,

the oscillation on the Flattop current wave form is due to the parasitic coil capacitance. This is a well known problem with traditional PI. This is the inter wire capacitance, the wire to shield capacitance and the cable capacitance. I have put it down as 200pf, but this is a very conservative figure. 500pf would not be exaggerated.

But why does it not appear on the saw-tooth? This seems to be something with the wire resistance that is not equal between the two sims, but I need to follow up on that.

Tinkerer

Edit on the above.
I just run it again, and it has now the same oscillation in both sims. The oscillation disappears when the Mosfet gate resistor is changed from 10R to 100R, or, in other words slowing down the switching a little bit.

How did that change your target response comparison?

-SB

[simonbaker]

Simon - I took your simulation and hacked it around to create a comparison between a PI without a resistor in series with the coil (case 1), and another with a 15 ohm resistor in series (case 2).

In case 1, the charging period is 18us.
In case 2, the charging period is 100us.
At the time the mosfet is ready to be switched off, they both have the same current flowing in the coil.
Plot "No-res" and "With_res" in one plot pane, and "Target1" and Target2" in a second pane.
As you will see in case1, the eddy currents are fairly well established in Target1 when the mosfet switches off. These eddy currents need to be overcome, resulting in a lower target response when compared to case2, where the eddy currents in Target2 have essentially decayed almost to zero.

These results appear to agree nicely with the practical test performed by Carl.

Good rearrangement for comparing the two.

BTW - That's Tinkerer's nice LTSpice simulation.

-SB

[Tinkerer]

How did that change your target response comparison?

-SB

OK, the fact of having R100 on the saw-tooth and R10 on the flattop, was a slip-up. Should be the same for comparison.
However, after correcting, the difference in signal response is insignificant.

To resume, for a short TC target, like TC=15us the result is similar to Carl's.

For longer TC's, the difference increases. This coincides with the opinion of many PI design aficionados and the theory of Eric Foster.

I am not happy with this result. Tests that I did in the past, were more in line with Carl's results.

So what is the cause of the difference?

Can we find something in the wave form of the eddy currents during TX ON, that could explain the result? Let's take a closer look.

Tinkerer

[Qiaozhi]

Good rearrangement for comparing the two.

BTW - That's Tinkerer's nice LTSpice simulation.

-SB

Oh yes - sorry, got lost in all the replies.

[simonbaker]

Simon - I took your simulation and hacked it around to create a comparison between a PI without a resistor in series with the coil (case 1), and another with a 15 ohm resistor in series (case 2).

In case 1, the charging period is 18us.
In case 2, the charging period is 100us.
At the time the mosfet is ready to be switched off, they both have the same current flowing in the coil.
Plot "No-res" and "With_res" in one plot pane, and "Target1" and Target2" in a second pane.
As you will see in case1, the eddy currents are fairly well established in Target1 when the mosfet switches off. These eddy currents need to be overcome, resulting in a lower target response when compared to case2, where the eddy currents in Target2 have essentially decayed almost to zero.

These results appear to agree nicely with the practical test performed by Carl.

Actually, case 1 has a larger delta I (change in current) in the target. Since magnetic field follows current, could this not actually create a larger signal in an RX coil? Or at least virtually as much? In other words, no penalty for established current?

-SB

[Tinkerer]

Actually, case 1 has a larger delta I (change in current) in the target. Since magnetic field follows current, could this not actually create a larger signal in an RX coil? Or at least virtually as much? In other words, no penalty for established current?

-SB

I think we need to stick with Carl's numbers, 100us TX. Otherwise we compare apples and potatoes.

Unfortunately, Carl has not given the target TC's. The target TC is what makes the difference.

On the sawtooth, when we look at the eddy currents increasing during TX ON, we see that the 15us target eddy currents increase until about 50us, or roughly 3TC, then they start decaying, in spite of the current ramp still raising nearly linear. Does this suggest that the target is "saturated"?

The longer TC eddy currents are still rising at switch OFF.

When we look at the flattop, we can see the same behavior, but the decay is enhanced as the delta I diminishes and goes flat.
There is a current spike at switch ON, that distorts the behavior a little.

Could it be that it makes a difference if the eddy currents are still raising at switch OFF, or that they are already decaying?

Tinkerer