Magnetic Compression Study" by Cyril Smith

[Smudge]

After setting up the variables in Mathcad and solving for the low frequency resonance capacitor Cl in equation (4) in Smudge's paper "On using Complex Permeability Resonance to get OU", it becomes obvious that the lower the ui, Fl, and Lair, the more realistic the value of Cl will be.

In my example, with Lair = 250uh, Fl = 100kHz, and ui = 2500, Cl = 4.053pfd. Does this sound right to you Smudge?

pm

Yes.  The Lair is the inductance of your coil wound on air.  With the core in place at permeability 2500 the actual inductance is 625mH, and that resonates at 100Khz with a 4pF capacitor.  That's a small value for the LF capacitor (if 100KHz is your LF charging frequency), the HF capacitor for the discharge phase would be even smaller.  Problems then with scope probes.  I would use fewer turns and use larger capacitor values.

Smudge

[partzman]

Yes.  The Lair is the inductance of your coil wound on air.  With the core in place at permeability 2500 the actual inductance is 625mH, and that resonates at 100Khz with a 4pF capacitor.  That's a small value for the LF capacitor (if 100KHz is your LF charging frequency), the HF capacitor for the discharge phase would be even smaller.  Problems then with scope probes.  I would use fewer turns and use larger capacitor values.

Smudge

OK I understand. What I don't seem to be able to work out is how you arrived at the 625mH inductance on the ui=2500 core for this coil? The actual measured Lcore ~50-55mH. Using the core dims of le=4.82*10e-2 and Ae=3.91*10e-5 (in meters) I calculate ~76mH so the ui is somewhat lower than 2500. Magnetics specs the Al for this core at 1430mH/1000 turns. Using this I calculate ~44mH for the 175 turn coil.

pm

[Smudge]

OK I understand. What I don't seem to be able to work out is how you arrived at the 625mH inductance on the ui=2500 core for this coil?

The only information you gave was the value of Lair=250uH and since the actual inductance is mu times that I got 2500*250uH = 625mH.  Using the core dimensions you now quote the value of Lair should be 31.22uH and actual inductance 2500 times that = 78mH.  As you say the actual mu is somewhat lower than 2500 and that is reflected in their Al value.

Smudge.

[K4zep]

Here is a paper written some time ago on using the slow charge fast discharge scheme.  This looks at resonant charging at 100KHz by switching a charged capacitor onto the inductor, that capacitor being two capacitors in parallel.  At the point where all the energy is now stored in the inductor, where the capaciitor voltage has fallen to zero, one of the capacitors is removed leaving only one there, and that one resonates with the changed inductance value at the ferromagnetic resonant frequency.  The energy stored in the inductor now swings back into the reduced capacitance, and when the current reaches zero the capacitor is disconnected to then be switched onto a load.   The formula for COP is derived using published complex permeability data.

Smudge

Morning Smudge,

Why use a Cap. to charge the inductor/core initially?  Just charge it from a power supply till the core is close to or saturated, then disconnect from power supply, connect resonating Cap., rectify the burst and Oscillatory current flow from it till the burst is essentially zero, dump/integrate rectified charge into output circuit and repeat?  Much simpler circuitry relatively speaking.  With the fixed time to charge the inductor/core known, the resonating frequency of the core known, the length of the burst at resonating frequency known, the circuitry is pretty simple in concept!  There are some very interest variables there to play with too!  Or.....are you doing all of your suggested sequence/cycle time during one cycle and not allowing the inductor/core to Osc. during the resonating part of the sequence and just discharge into Res. Cap.?

Respectfully,
Ben

[Smudge]

Good Day K4,

I chose resonant charging for efficiency but of course you could charge directly.  In that case obtaining the quantity of energy actually supplied to the inductor involves an accurate measure of the voltage and the current and an integration, whereas with the capacitor method you merely need to know the voltage and the capacitor value.

You can't disconnect from the power supply without first having the discharge capacitor connected, else there will be an almighty ring into stray capacity at the disconnect time.  If the intention is to create something useful I would not go for a multi-cycle burst of output frequency as that will lose energy, a quarter cycle charge is the best.  But if you only want to prove the concept then the peak voltage reached on the output capacitor tells you everything.

Smudge

[Peterae]

Smudge
I found an unknown ferroxcube core in my box, i have done some tests on it this afternoon like we did some time ago during the simulation negative resistance tests, i have switched capacitors across a 10 turn coil on the core and recorded the results and calculated inductance versus frequency, any chance you could pass the data through your permeability spread sheet to produce a graph to see if it's suitable for this experiment when you get a chance.

Note my cap box has a variable cap inside which read 220pf and is in parallel with the switched caps so i had to add 220pf to all the switched in cap values in the spread sheet.

PS i worked out i can use Zeners as voltage controlled vari caps to change the capacitance, not yet sure if it fast enough though to do it on a cycle basis.

Cheers
Peter

[K4zep]

Good Day K4,

I chose resonant charging for efficiency but of course you could charge directly.  In that case obtaining the quantity of energy actually supplied to the inductor involves an accurate measure of the voltage and the current and an integration, whereas with the capacitor method you merely need to know the voltage and the capacitor value.

You can't disconnect from the power supply without first having the discharge capacitor connected, else there will be an almighty ring into stray capacity at the disconnect time.  If the intention is to create something useful I would not go for a multi-cycle burst of output frequency as that will lose energy, a quarter cycle charge is the best.  But if you only want to prove the concept then the peak voltage reached on the output capacitor tells you everything.

Smudge

Excellent, a good explanation of what you would be doing.  Some mighty fast precise switching there but nothing that can't be done!

Ben

[Smudge]

Smudge
I found an unknown ferroxcube core in my box, i have done some tests on it this afternoon like we did some time ago during the simulation negative resistance tests, i have switched capacitors across a 10 turn coil on the core and recorded the results and calculated inductance versus frequency, any chance you could pass the data through your permeability spread sheet to produce a graph to see if it's suitable for this experiment when you get a chance.

Note my cap box has a variable cap inside which read 220pf and is in parallel with the switched caps so i had to add 220pf to all the switched in cap values in the spread sheet.

PS i worked out i can use Zeners as voltage controlled vari caps to change the capacitance, not yet sure if it fast enough though to do it on a cycle basis.

Cheers
Peter

Can't get permeability without core size but plot of L against frequency will have same shape as plot of mu against frequency.  No signs of a resonance there

Smudge

[Peterae]

OK thanks Smudge was worth a try.

If i can work out a circuit to do what we need then i will order a core 

Anyone else got any ideas on how to circuit this up would be appreciated.

[Smudge]

Here is a spread sheet that takes in core dimensions, complex permeability data at the two different frequencies, number of turns, input voltage to the first capacitor then calculates the inductances, the capacitor values, the peak current when the first cap is fully discharged and the peak voltage on the second cap at the end of its charging period.  Also energy in, energy out and the COP.

Smudge

[partzman]

OK thanks Smudge was worth a try.

If i can work out a circuit to do what we need then i will order a core 

Anyone else got any ideas on how to circuit this up would be appreciated.

Peterae,

Attached is a simple circuit to test for permeability resonance  as Smudge describes.

M1 is basically "on" for most of the period which allows a desired constant current to be generated in L1 by means of V1, a variable dc supply. M1 is then switched "off" allowing the energy in L1 to be dumped into the parallel combo of the fet's drain to source capacitance Cdss and C2 thus generating a 1/2 sine across the total output capacitance.

The peak of the 1/2 cycle resonant voltage can easily be measured to calculate the output energy as compared to the input energy calculated from the constant  current thru L1. No integration is necessary on the input.

There is a problem that must be considered however and that is even though both the dcr of L1 and the Rdson of M1 are low, there will still be a small amount of dc voltage required to produce the inductor current needed. The energy produced by this voltage and the decaying current ramp during the charging of the resonant output capacitance must be considered in the overall COP calculation.

pm 

[Smudge]

I think the problem is getting the value of L1 to use in the calculation.  If there is a resonance in the core then L1 will vary with the frequency used to measure it.  So if you are just looking for the resonance to check that it is there you might as well just look for that change of inductance.

If you are going for COP measurement then you must measure L1 at a low frequency, but are you then sure that your Li²/2 calculation of energy is good enough? My feeling is that you won't be sure you have got COP>1 without a cross check on energy going in and coming out, which is why I suggested resonant charging of L1.  Putting energy into L1 direct from a DC supply doesn't allow you to get a good handle on the energy flow.

Smudge

[partzman]

I think the problem is getting the value of L1 to use in the calculation.  If there is a resonance in the core then L1 will vary with the frequency used to measure it.  So if you are just looking for the resonance to check that it is there you might as well just look for that change of inductance.

If you are going for COP measurement then you must measure L1 at a low frequency, but are you then sure that your Li²/2 calculation of energy is good enough? My feeling is that you won't be sure you have got COP>1 without a cross check on energy going in and coming out, which is why I suggested resonant charging of L1.  Putting energy into L1 direct from a DC supply doesn't allow you to get a good handle on the energy flow.

Smudge

OK, here is a circuit which does include both input and output resonance and is still reasonably simple. The only caveat is that C2 is also charges thru Rchg and L1 along with C1 but it will be rapidly discharged by M1 and should have no material effect on the charging of L1 by C1.

Basic operation is as follows- M1 and M2 are off for a period of time allowing C1 or Clow to be charged to a peak voltage as determined by the dc supply V1. Rchg should be at least several magnitudes larger than the impedance of the the L/C circuit. M1 is then switched on with Vpulse1 starting at T0 and turned off at T1 when C1 is fully discharged and the current in L1 is at it's peak. The time period from T0 to T1 should be 1/4 of the period of the low resonance frequency desired.

With C1 at zero volts and M1 turning off, M2 is now switched on at T1 which will produce a reverse conduction mode that will clamp L1 at ground with very little loss. C2 or Chi will now charge at the high resonance frequency desired from the energy stored in L1. M2 must remain on from T1 to T2 lnog enough to allow a peak voltage to be produced in C2. The starting voltage on C1 and the ending peak voltage on C2 along with their values provide an easy means to calculate the COP.

This operation may be periodic as long as the circuit conditions are met.

pm   

[Peterae]

Hi partzman

Thanks looks good, and easy as well

Very ingenious

Regards
Peter

[Kator01]

Partzman,

if M2 is turned on C1 is still parallell to L1 and the desired high frequency will be shorted by C1.
Solution could be another MOSFET which seperates C1 and Rchg from L1 at the instant you turn on M2.
But even then I will suspect that the Miller-Capacity will be a hf-bypass disturbing the build-up of the hf-resonance

So this problem is not so easy to be solved. The capacity of the supply may be another problem but could be neglegted if Rchg is large enough.

At the present I also have no idea how to solve this, have to think about it...

Mike

[forest]

Hi

I think you make things too complicated using electronic knowledge in part when it is not useful.
Here is my idea for consideration: use ordinary push-pull circuit first with a proper HF choke on input with a additional electronic snubber to suppress any high frequency oscillations. Second circuit is a disruptive discharge on the same ferrite core which push-pull operates on (or on secondary core as you wish if you direct output of push-pull there) - with a almost non-inductive winding at 90 degrees (i non-inductive relation) to the push-pull windings . The output of push-pull must have hf diode bridge.
The essence of non-inductive winding and disruptive discharge circuit in non-inductive relation to the original power source winding (push-pull winding) is to create very fast pulses at push-pull winding by a special mode of operation. This winding should create big magnetic field in one direction followed by equal opposite big magnetic field in counter direction , with nanoseconds delay  between. This will cause saturation of core for nanoseconds - breaking push-pull magnetic induction action for this period. Along with normal push-pull output will be also HF oscillation of additional energy - it will be on secondary and primary also - so here is the trouble.

I believe all the Akula, Ruslan, Dally,SR913  devices with ferrite core work that way (or trying to do so).

Electrons are valves to the cosmic energy from external magnetic field (Earth field), every generator in reality condense the external field, input energy is just wasted. When magnetic flux is passed through material using electrons momentum it cannot flow in two directions at the same time.So if you saturate core in 90 degrees then the original flux is broken too 

[Verpies]

Here is something to think about regarding soft materials.

I enjoyed your take on soft magnetic materials and I mostly agree with it.

However I make an exception for this passage:

And magnets swinging about inside coils induce voltage. Thus charging a coil with current at low frequency
creates a small voltage to "load" the current generator, while discharging itat the resonant frequency creates a greater output voltage, hence greater power out than power in.

First of all, voltage is not power nor energy, so it would be wrong to imply, that more induced voltage means more electric energy coming out of a coil.
It is almost as if you are saying, that the electric energy output of an ideal coil is proportional to dΦ/dt, where Φ is the instantaneous magnetic flux penetrating the inside of the coil (caused by these swinging magnets).

[partzman]

Partzman,

if M2 is turned on C1 is still parallell to L1 and the desired high frequency will be shorted by C1.
Solution could be another MOSFET which seperates C1 and Rchg from L1 at the instant you turn on M2.
But even then I will suspect that the Miller-Capacity will be a hf-bypass disturbing the build-up of the hf-resonance

So this problem is not so easy to be solved. The capacity of the supply may be another problem but could be neglegted if Rchg is large enough.

At the present I also have no idea how to solve this, have to think about it...

Mike

Mike,

I've attached a general equivalent circuit showing the conduction periods of M1 and M2. 

During the conduction of M1, C2 is grounded through the Rdson of M1 and the energy in C1 is transferred to L1.

During the conduction of M2, C1 is grounded through the Rdson of M2 (reverse conduction) and the energy in L1 is transferred to C2.

It is desirable that the on resistance of M1 and M2 be as low as possible to reduce any effect on the COP calculation.

pm

edit-

[Verpies]

The attached pdf could help in turning published complex permeability data into the frequency variations of chi.

I appreciate your paper on the basics of complex permeability but I can't help to notice that you did not even mention, that the contribution of domain walls is strongly dependent on the ratio of magnetization, since for total magnetization these domain walls disappear.

This domain wall behavior is nicely illustrated in the attached paper.

[Smudge]

I appreciate your paper on the basics of complex permeability but I can't help to notice that you did not even mention, that the contribution of domain walls is strongly dependent on the ratio of magnetization, since for total magnetization these domain walls disappear.

And of course the permeability also disappears (or rather drops down to unity) so there is no resonance to be seen.  This will affect the power you can get from a given size of ferrite but it doesn't alter the principle.

Smudge

[Smudge]

Here is another paper I wrote in 2007.  This offers a slightly different approach in that the coil is charged with current quickly (at the resonant frequency), then held at that current while the core permeability relaxes down to its low frequency (or DC) value, then discharged slowly.  The loop is again traversed CW indicating an energy gain.  So you circuit gurus now have an alternative to consider.

Smudge

[Peterae]

So here's a circuit to drive partsman's circuit.

Zener Diode 1 starts conducting at a pre determined voltage, pulling up the voltage divider R3/R4,  the voltage when above 1.5V will cause the positive triggering Schmitt input of monostable IC1 to trigger, mono ICI output Q goes high for a determined amount of time set by R1/C3, thus turning the logic fet on M1.

After Mono 1C1 has timed out the output Q will go low triggering IC2's active low trigger pin causing IC2's output pin Q to go high which in turn drives logic fet  M2 on.
when IC2 times out both fets are off and the power supply starts charging again until we get a trigger again on the Zener diode and the cycle completes again.

[partzman]

Here is a circuit that will allow one to find the resonance of any unknown junk box ferrite core in relatively quick fashion. The magnitude can be calculated from the results if the reference coil and measuring tools are accurate.

Basically, a differential measurement is taken between the output of a sine signal generator (CH1) and an output voltage (CH2) created by an inductive divider made up of an air coil reference and the inductor to be measured as seen in the attached schematic. Since XL = 𝜔L, the divider voltage will remain in a constant ratio to the input thru a given frequency range until the inductance of L2 begins to change. If the inductance in L2 increases, the difference (that is CH1-CH2) will decrease and reach a peak dip at some frequency. This can be measured on a scope or with two wideband AC voltmeters that would preferably be identical.

The reference coil L1 ideally should be linear and with low self capacitance and away from any material that would alter it's u=1.  The reference coil should be equal or close to the inductance of the coil/core to be tested but is not necessary but would provide the best sensitivity.

I have used an Agilent 33120A generator set with 50 ohm output and prefer to do a manual sweep for finding the peak rise in inductance. For most ferrites, the frequency range seems to be from 100kHz to 1.5mHz or thereabout.

pm 

[partzman]

Here is another paper I wrote in 2007.  This offers a slightly different approach in that the coil is charged with current quickly (at the resonant frequency), then held at that current while the core permeability relaxes down to its low frequency (or DC) value, then discharged slowly.  The loop is again traversed CW indicating an energy gain.  So you circuit gurus now have an alternative to consider.

Smudge

If I understand this paper correctly, the previous circuit I proposed for your low to high permeability change would work for the high to low with a change in timing and cap values.

The attached schematic shows the required changes. C1 is now the high frequency resonating cap with inductor L1 and is brought to a predetermined energy level by V1 and R1. Then, Vpulse1 is applied to M1 at T0 and held until T2. When the energy in C1 has completely transferred to L1, Vpulse2 is applied to M2 at T1 thus clamping and holding the current in L1. The time interval from T1 to T2 is the relaxation period for the core permeability to settle.

M1 is then switched off at T2 dumping the energy in L1 into C2 at a lower determined frequency rate. M2 must remain on long enough for this energy transfer to occur.  The output energy calculated from the peak voltage in C2 as compared to the starting input energy in C1 determines the COP.

pm

[Peterae]

Thanks for the circuits partsman, i can adapt my circuit also for circuit 2 by adding another mono.

I think one of the most important things now would be to try and find a core that would give at least 50% gain that is possible to purchase, unless we can find a source of the military spec core that was in Smudges list, i dont think this thing will be loop-able without that sort of figure.