ferrite simulation

[green]

How to simulate ferrite with spice? Can I use parallel inductor, resistor? If yes, what ratio?

[Skippy]

That's a question I've pondered, but haven't come up with a good answer to. Ferrite has a frequency response that's basically unaffected by detector operating frequency. So any electrical model would also have to behave like this. If you're using a L & R or L & C combination, for example, the time-constant / corner frequency would have to be way above the detectors frequency. Example: if you want the phase lag of your target model to be less than 0.5 degrees, you would need the target corner freq to be more than 115 times the detector freq.

F _detector = F _target x tan phi

so if F _detector = 13KHz, and phi = 0.5 degrees, then F _target = 1.49 MHz

I was also thinking about simulation of ferrous targets in general, and the ground signal specifically.

[Qiaozhi]

How to simulate ferrite with spice? Can I use parallel inductor, resistor? If yes, what ratio?

I've never tried to simulate a ferrite target, but pure ferrite is essentially purely resistive. That's why you can use a "ferrite wand" when setting up the ground balance in an air test on a VLF. The reason is that it only affects the signal amplitude, and not the phase. So when the ground balance sample pulse is centered over the zero-crossing of the RX signal, there is no response. In practice, the ferrite slug consists of iron oxide, which can produce a response from the detector if you get it too close to the coil. Real (neutral) ground has a phase shift of around -5 degrees.

For a PI detector, you would be hard pushed to get a response from a ferrite slug.

[green]

I recorded amplifier out with a ferrite bead and some California ground using an IB coil and my bench PI awhile back. Tx 160usec constant rate(6250 amps/sec). Copied scope signal to Excel to see slope better, both looked flat on the scope trace. Been playing with spice today, constant rate pulse. 10uH parallel with .001 ohms(10,000usec time constant)gives a fairly flat signal during Tx. 1000usec TC has some decay slope. Something over 10,000usec should be flat. The ground signal is higher at the end of Tx. Ferrite doesn't show a R signal after Tx off delay, ground does.

Added cr.zip, Rx(c)out
Don't know how to simulate ground with spice but I think the Rx signal during Tx on time should be fairly flat with constant rate and an IB coil.

[Davor]

If you find any slope, it should be from either viscosity or conductivity.
Ferrite response is instantaneous in amplitude and phase. Since phase is obscured in PI, Skippy's solution is sadly not a working one. In frequency domain, a worst possible conductor still has 90° phase.
The effect ferrite produces is adding some extra inductance to coils. If a target is ferromagnetic like a nail, it means the conducting part (tau) is superimposed with increase in inductance on the coil(s) side. Such increase may be produced either through mutual inductance in a K statement, or directly as minute increase in inductance of a monocoil.
That said, it is easy to implement such increase in Spice in .step param manner, but not so much in a continuous way. I developed a "varicoil" model for that purpose.

[pustareka]

How to simulate ferrite with spice? Can I use parallel inductor, resistor? If yes, what ratio?

imposible simulate.ferit isheart of earth.how heart simulate? imposible.

[green]

imposible simulate.ferit isheart of earth.how heart simulate? imposible.

The parallel resistor, inductor(time constant>10,000usec) target simulation matches what I get with my PI with ferrite beads, not ground(earth). 100,000use time constant is better. Seems to work with a VLF simulation if I did it right. Do you see something wrong with the simulations?

Added VLF simulation. click on Tx and Rx, try 5kHz and 50kHz with V1. Get a small shift at 5kHz, more shift at 2kHz. Probably shouldn't see any shift. Is the problem with the spice simulation or ferrite simulation?

[Ferric Toes]

The parallel resistor, inductor(time constant>10,000usec) target simulation matches what I get with my PI with ferrite beads, not ground(earth). 100,000use time constant is better. Seems to work with a VLF simulation if I did it right. Do you see something wrong with the simulations?

Added VLF simulation. click on Tx and Rx, try 5kHz and 50kHz with V1. Get a small shift at 5kHz, more shift at 2kHz. Probably shouldn't see any shift. Is the problem with the spice simulation or ferrite simulation?

Magnetic viscosity is 'frequency dependent magnetic susceptibility', so you would expect to see a shift between a LF and HF measurements. I have a Bartington MS2 susceptibility meter which measures a sample at 0.47kHz and 4.7kHz. The greater the difference between the LF reading and the HF reading, the higher the magnetic viscosity (Xlf - Xhf = Xd). The basic reasoning is that at higher frequencies of an a.c. field the magnetic domains progressively cannot keep up with the field reversals. Commercial ferrites are not a good substitute to simulate ground response as some have very low levels of viscosity while others show higher levels; it all depends on the application.

In Spice, can you simulate a magnetic core with different levels of susceptibility?

Eric.

[Davor]

I've put a viscous soil spice model before. You may search here on geotech.
As for ferrite, try this modified file from few posts back (one step is a non-ferrous response, and the other is ferrous):

[green]

Magnetic viscosity is 'frequency dependent magnetic susceptibility', so you would expect to see a shift between a LF and HF measurements. I have a Bartington MS2 susceptibility meter which measures a sample at 0.47kHz and 4.7kHz. The greater the difference between the LF reading and the HF reading, the higher the magnetic viscosity (Xlf - Xhf = Xd). The basic reasoning is that at higher frequencies of an a.c. field the magnetic domains progressively cannot keep up with the field reversals. Commercial ferrites are not a good substitute to simulate ground response as some have very low levels of viscosity while others show higher levels; it all depends on the application.

In Spice, can you simulate a magnetic core with different levels of susceptibility?

Eric.

First, I am just trying to learn something. When I recorded a ferrite bead with a IB coil and a constant rate Tx the Rx signal was constant amplitude during Tx. I was thinking VLF Rx signal amplitude should be constant with frequency if the Tx amplitude is constant. Don't have a VLF detector so I was wondering if I could simulate with spice. (How to simulate a ferrite bead with spice?) Looked at VLF simulation again this morning, change L1 resistance to .1 ohms and the phase and amplitude stay constant with frequency so I'm thinking the parallel L R might be a good ferrite simulation. Don't know how to simulate the California ground with a positive slope during Tx. Any suggestions would be appreciated.

[green]

I've put a viscous soil spice model before. You may search here on geotech.
As for ferrite, try this modified file from few posts back (one step is a non-ferrous response, and the other is ferrous):

I changed L1 to .1 ohms with your simulation and the Rx amplitude is constant with frequency if I'm reading it correctly. I'm just learning spice so now I need to understand what you are doing.
I'll try to search the viscous soil spice model. Thanks

[Qiaozhi]

I changed L1 to .1 ohms with your simulation and the Rx amplitude is constant with frequency if I'm reading it correctly. I'm just learning spice so now I need to understand what you are doing.
I'll try to search the viscous soil spice model. Thanks

Unfortunately I don't have the time to investigate, but this document might help you ->
http://www.spectrum-soft.com/news/wi...rritebead.shtm
It talks about Micro-cap, which is also a SPICE simulator, so you can probably work it out.

There's some other stuff here that might give you a clue ->
http://www.pspice.com/resources/appl...mi-suppression
https://www.altera.com/content/dam/a...e/an/an583.pdf
http://www.analog.com/media/en/techn...es/AN-1368.pdf

[Davor]

Sensing ferrite in free space is entirely different from observing it in a circuit.
Its response has virtually no delay, and therefore a classical PI can't see it.
My circuit modification gives frequency response in 2 steps. The first step is a pure non-ferrous target, with normal amplitude and phase response. The other step introduces mutual inductance between Tx and Rx coils, and together with a target is a static simulation of ferromagnetic target. Both phase and amplitude response are affected.

I'm missing descriptions of real-life targets to tune such models properly. Besides the ferromagnetic behaviour of a real life target is orientation dependent, and differently so than the non-ferrous targets.

If someone has at hand measured frequency and phase responses, I'll be able to create a model of such target.

[green]

Unfortunately I don't have the time to investigate, but this document might help you ->
http://www.spectrum-soft.com/news/wi...rritebead.shtm
It talks about Micro-cap, which is also a SPICE simulator, so you can probably work it out.

There's some other stuff here that might give you a clue ->
http://www.pspice.com/resources/appl...mi-suppression
https://www.altera.com/content/dam/a...e/an/an583.pdf
http://www.analog.com/media/en/techn...es/AN-1368.pdf

Thanks for the links. http://www.analog.com/media/en/techn...es/AN-1368.pdf has some graphs showing ferrite is inductive at lower frequency. The metal detector covers maybe 1 to 100kHz so probably inductive. Maybe the L R ferrite simulation makes sense?

[Qiaozhi]

Thanks for the links. http://www.analog.com/media/en/techn...es/AN-1368.pdf has some graphs showing ferrite is inductive at lower frequency. The metal detector covers maybe 1 to 100kHz so probably inductive. Maybe the L R ferrite simulation makes sense?

That seems to be the conclusion.
At low frequencies (below crossover) the ferrite bead is inductive. However, at the frequencies used by VLF detectors, the impedance is effectively zero, which is why a ferrite bead can be used to ground balance the detector, since it doesn't produce any phase-shift, only an amplitude change. In the case of a PI detector, the low impedance almost instantly kills off any eddy currents that are generated in the bead, and (as Davor says) "a classical PI can't see it".

If we assume that the impedance is actually 1mOhm (i.e. not zero, but very small) then at 10kHz the inductance calculates as 15.9nH. Which is too small to generate any appreciable phase-shift.