PI SAMPLE DELAY QUESTION

[Qiaozhi]

Thanks Joe,

That is indeed another good method to measure true SRF of a coil and /or coil /feed combination. From a practical standpoint I would say that the measurement made with the coil in the detector circuit gives the best opportunity to properly damp the coil with all variables in play. I would recommend that a variable resistance network like you outline in your monocoil paper be placed in circuit and adjusted to give the least ringing with maybe a bit of residual overshoot. This would be the shortcut to the critical damping that is the real objective here. It will be interesting to see what damping resistance actually is needed and how much it varies from the original 390 ohm damping.

Thanks again,

Dan

Thanks Joe,

That's a good tip.

I've just done a quick test with 3 different coils:
1) Original Pulse Technology MiniPulse Coil = 2.64us (SRF = 378.8kHz)
2) Coil wound with standard hookup wire = 1.62us (SRF = 617.3kHz)
3) Coil wound with teflon wire = 1.48us (SRF = 675.7kHz)

[deemon]

Have you confirmed this with a practical experiment?
If so, was the target ferrous or non-ferrous?

Of course , I did it many times .... at first , I predicted this theoretically , and then tested in real experiments . It's not a problem to show it , although I use another technology in my projects ( bipolar square wave ) , I have a little experimental rig that contains a classic PI power chain and front-end . In this setup I have a NE555 pulse gen ( with manual duration and period setting ) , power switch ( BU1508AX ) , little test coil ( 420 uH , 1,6 Ohm wire , 390 Ohm parallel ) with serial resistor limiting the current , and a simple bipolar switch ( flyback pulse blanking ) to prevent oscilloscope overload by a strong hi-volt pulse just after a power switch shutdown , power supply is 24 V ....

So I took some pictures on the real scope . First picture is an idle signal , without any target - . Upper trace is the coil current , lower trace is the signal after switch ( coil flyback pulse cancelled ) , 100 uS/div . Then we can see the signal from the massive metal target with a large TC ( thick copper ashtray ) with the ON pulse duration 50 uS . On the next picture ON pulse is 100 uS , then 200 and 300 uS . So we can see how the signal magnitude depends on power pulse length - before "current saturation" and after it . And after 100 uS ( where the current is steady ) we can see what I call a "target charging"

By the way , a similar effect can be observed on ferrous targets too , but I haven't a ferrous target with large enough TC - so to watch the effect I need more "fast" coil . I took 2 pictures on the ferrous target ( big iron hammer ) too , with the noticeable effect , where the signal height is 10% more on 300uS point than that on 150 uS ( with the same coil current ) ... but of course , if I make a coil with a less inductance and thus more rapid current rise - it must be the same pictures that I had on my copper target , but with shorter timing of course .

[deemon]

And here are these 2 pictures of a ferrous target . As we can see - the "target charging" effect is still present here , but obviously more weak - because the target has lower TC and being "saturated" on less ON pulse duration , so when we use longer pulse ( 300 uS instead of 150 ) we can watch only 10% signal growth . I think that if I use a coil with 100 uH inductance ( and several times faster current rise ) - it might be possible to see the same drastical effect , like that in the previous case ....

[Qiaozhi]

Of course , I did it many times .... at first , I predicted this theoretically , and then tested in real experiments . It's not a problem to show it , although I use another technology in my projects ( bipolar square wave ) , I have a little experimental rig that contains a classic PI power chain and front-end . In this setup I have a NE555 pulse gen ( with manual duration and period setting ) , power switch ( BU1508AX ) , little test coil ( 420 uH , 1,6 Ohm wire , 390 Ohm parallel ) with serial resistor limiting the current , and a simple bipolar switch ( flyback pulse blanking ) to prevent oscilloscope overload by a strong hi-volt pulse just after a power switch shutdown , power supply is 24 V ....

So I took some pictures on the real scope . First picture is an idle signal , without any target - [ATTACH]30821[/ATTACH] . Upper trace is the coil current , lower trace is the signal after switch ( coil flyback pulse cancelled ) , 100 uS/div . Then we can see the signal from the massive metal target with a large TC ( thick copper ashtray ) with the ON pulse duration 50 uS . On the next picture ON pulse is 100 uS , then 200 and 300 uS . So we can see how the signal magnitude depends on power pulse length - before "current saturation" and after it . And after 100 uS ( where the current is steady ) we can see what I call a "target charging"

By the way , a similar effect can be observed on ferrous targets too , but I haven't a ferrous target with large enough TC - so to watch the effect I need more "fast" coil . I took 2 pictures on the ferrous target ( big iron hammer ) too , with the noticeable effect , where the signal height is 10% more on 300uS point than that on 150 uS ( with the same coil current ) ... but of course , if I make a coil with a less inductance and thus more rapid current rise - it must be the same pictures that I had on my copper target , but with shorter timing of course .

Thank you deemon for your comprehensive and very interesting results. I have combined all the scope shots, plus comments, into a single PDF document for easier viewing, and converted the images to monochrome and inverted. Please find document attached.

If you look carefully at the coil current ON-time, you will see that the current amplitude is still increasing in all the examples. There is even a visually detectable increase between the 200us and 300us tests. I suspect that this charging effect will disappear if you were to increase your ON-time period even longer. However, you might then argue the target is now fully charged, and is therefore saturated, so increasing the ON-time will have no effect. My conclusion (from your scope images) is that the RX signal is increasing in an exponential manner consistent with the ON-time of the coil, which suggests this increase is the result of a stronger collapsing magnetic field due to the higher current.

Sorry, but I'm still not convinced by your target charging theory.

[Qiaozhi]

And here are these 2 pictures of a ferrous target . As we can see - the "target charging" effect is still present here , but obviously more weak - because the target has lower TC and being "saturated" on less ON pulse duration , so when we use longer pulse ( 300 uS instead of 150 ) we can watch only 10% signal growth . I think that if I use a coil with 100 uH inductance ( and several times faster current rise ) - it might be possible to see the same drastical effect , like that in the previous case ....

What ferrous target did you use?
I assume it is physically much smaller than the massive copper plate?

[green]

I might be wrong but I think Qiaozhi and deemon would predict the same amplifier signal change when varying coil on time with a coil TC of 5usec and a target TC of 50usec. The difference being is the target charging or discharging during coil on time. I did the spice simulation in reply #56 because I didn't know how to show if the target is charging during coil on time with hardware. Spice shows discharging. Maybe simulation is not done correctly. I have looked at target signals during coil on time with an IB coil. If I remember right ferrous don't decay, non-ferrous targets do. Maybe with above target and coil TC the result might be different. Maybe someone has a suggestion on how to prove if target charges or discharges with an actual test. Would do a test with IB coil if it would answer the question. Please tell me if Qiaozhi and deemon are predicting a different outcome. Maybe I'm just confused.

[deemon]

I might be wrong but I think Qiaozhi and deemon would predict the same amplifier signal change when varying coil on time with a coil TC of 5usec and a target TC of 50usec. The difference being is the target charging or discharging during coil on time. I did the spice simulation in reply #56 because I didn't know how to show if the target is charging during coil on time with hardware. Spice shows discharging. Maybe simulation is not done correctly. I have looked at target signals during coil on time with an IB coil. If I remember right ferrous don't decay, non-ferrous targets do. Maybe with above target and coil TC the result might be different. Maybe someone has a suggestion on how to prove if target charges or discharges with an actual test. Would do a test with IB coil if it would answer the question. Please tell me if Qiaozhi and deemon are predicting a different outcome. Maybe I'm just confused.

Maybe I didn't understand correctly your question , but what I think - the green trace on your picture looks correct . In reality one "spike" ( positive one on the left side ) must be more "smooth" - because of less voltage transient on the coil at that moment . I can check it in the real setup , but I predict that it must look similar to your simulation . But what about your blue trace - it looks weird .... in reality the current must grow exponentially from a start point , moving to a horizontal line - something similar to a real trace on my photos in the previous comment . But why you have a sharp spike at the 0 uS point on the graph - I don't know

[deemon]

Thank you deemon for your comprehensive and very interesting results. I have combined all the scope shots, plus comments, into a single PDF document for easier viewing, and converted the images to monochrome and inverted. Please find document attached.

If you look carefully at the coil current ON-time, you will see that the current amplitude is still increasing in all the examples. There is even a visually detectable increase between the 200us and 300us tests. I suspect that this charging effect will disappear if you were to increase your ON-time period even longer. However, you might then argue the target is now fully charged, and is therefore saturated, so increasing the ON-time will have no effect. My conclusion (from your scope images) is that the RX signal is increasing in an exponential manner consistent with the ON-time of the coil, which suggests this increase is the result of a stronger collapsing magnetic field due to the higher current.

Sorry, but I'm still not convinced by your target charging theory.

You see , Qiaozhi , in reality the current stops to grow at least after 150uS point , so you cannot explain the signal growth in that way ... it's a pity that I made a photos too dark , so we cannot see the CRT scale . Tomorrow I'll make another photo with the scale illumination and higher current trace resolution ...

[deemon]

What ferrous target did you use?
I assume it is physically much smaller than the massive copper plate?

It really smaller in length , but heavier - it's a simple iron hammer with a wooden handle . I can make a photo of both targets ( and the coil ) - it isn't a problem .

[green]

Maybe I didn't understand correctly your question , but what I think - the green trace on your picture looks correct . In reality one "spike" ( positive one on the left side ) must be more "smooth" - because of less voltage transient on the coil at that moment . I can check it in the real setup , but I predict that it must look similar to your simulation . But what about your blue trace - it looks weird .... in reality the current must grow exponentially from a start point , moving to a horizontal line - something similar to a real trace on my photos in the previous comment . But why you have a sharp spike at the 0 uS point on the graph - I don't know

The coil TC is 5 usec so the blue trace (current trace) should flatten out at 25usec. It grows exponentially, just hard to see with the large scale showing the full signal decay to zero. The current scale is on the right side. Not a spike, current starts at zero.

[mschmahl]

This looks more like a MOSFET sustaining some kind of relaxation oscillation. Could you see what happens at a MOSFET gate?

Hi Davor,
this is a picture from my MOSFET gate, 5V/Div 20uS/div.
And a part of the schematic.

Ive made a big mistake with the coil im making !!! to get the coil into the shell i had to wind the turns side by side and also reduce them to 12 turns...i now have a coil which is only 140uH and i feel that ive made a big mistake....can this coil still be used for beach work for rings and coins without problems ?????

Hi daverave, i have used a 199uH coil with the Surf and it was very good. So try your coil.

Some traces I posted in another thread awhile back. The first is without a diode in series with the coil. The next three are with different diodes in series. I think the coil was 280uh. The 344 khz without the diode is close to what you got. Looking at the upper trace it looks kike I had the scope probe on the coil, 450 volt flyback. The lower traces are are the ringing at a lower volt setting.

Thank´s green. So i´m on the right track.

It is difficult but looking at the ringing portion of the scope screen shot at 20us /division, and assuming 7 cycles in 20us I get a ringing frequency of about 350 kHz. Your construction details of 24 awg wire with PVC insulation and a copper shield, plus the capacitance of the detector electronics i.e. mosfet etc. make it very possible that 350kHz is your true IN CIRCUIT ringing frequency. 24 awg wire will have more capacitance than smaller gauges, PVC has more capacitance than PTFE/TEFLON insulation and a solid copper shield contributes more capacitance than no shield or a more sparse shield. This all adds up to a slow coil. Another question is how much space is there between the coil and the shield and what is the spacer made of? Also if the coil has been flooded with epoxy between the windings the capacitance will be even higher, lowering the resonant frequency even more. I believe the coil is operating at about 350kHz.

Assuming the damping resistor was disconnected for this test...With 350kHz X PI X 295uh damping resistor should be about 324 ohms. In other words the coil appears to be under damped using a 390 ohm damping resistor.

Regards,

Dan

Hi Dan!
Thank´s for your explanation. So i know how to make a slow coil. :-) The space between wire and shield is two layers of PVC spiralwrap (around 2mm). The coil is not flooded between the windings. I think the mainpoint for slowing it down is the shield, it is to massive.

[mschmahl]

Dan, and all interested

There is another way to measure the Self Resonant Frequency (SRF) of a coil. Place a coil to measured (coil under test) on top of a fully functioning PI machine's coil. Attach the scope to the leads of the coil under test. Turn on the PI machine and through inductance the coil under test will be stimulated by the square wave pulse of the TX pulse. In the coil under test will be a ringing that tapers off. Measure the space between the peaks of the two highest rining pulses. That distance represents the time of the resonant frequency unloaded by the MOSFET COSS and any other circuit capacitance. Make sure that there is no damping resistor in the coil as the SRF ringing will not be easily seen.

This test will reveal the result of the PI coil under test inductance in parallel with:

1. The coil turn-to-turn capacitance
2. Capacitance based onthe Dielectric constant of the wire insulation and space between windings
3. Capacitance based on the dielectric constant of the spacer material and thickness of the space between the coil wire bundle and the shield
4. Capacitance of the length of coax wire between the coil and the PI control box.
5. Scope probe capacitance

Joseph Rogowski

Hello Joseph!

Thanks for this great tip. I have meassured my coil now with your instruction.
Now i get 1.6 uS~625 kHz.
Scope-Settings are 50V/Div and 1uS/Div.
I will test some other coil´s to see differences with shield and without.

[Davor]

OK, so gate is clean, and those are only coil oscillations.

[mschmahl]

For calculation of the dumping resistor Dan´s (Baum7154) variant is correct. Attached is a picture with the calculated 324 Ohm dumping resistor.

But for quick finding if a coil is a good one, i think Joseph´s (bbsailor) variant is the best.

So we should meassure some coil´s in both variants to see a ratio from Joseph´s to Dan´s variant to find quick (and dirty) the right dumping resistor.

The ratio for my 10" Coil is: 625 kHz : 377 kHz = 1,78

So for calculation the dumping with joseph´s variant: 625:1,78 x PI x 295uH coil = 325 Ohm (Dan´s variant: 377 kHz x PI x 295uH = 349 Ohm)
I have to test it with other coil´s, but think it is close enough.

Had one big mistake in my conclusion. The ratio depending on how you built your pi. So this factor is only right for my pi. So we should use joseph´s variant to compare coils without the factor of different pi-builds.

[Qiaozhi]

deemon - What value resistor do you have in series with the coil?