PopularFX
Home Help Search Login Register
Welcome,Guest. Please login or register.
2024-05-05, 17:54:26
News: Check out the Benches; a place for people to moderate their own thread and document their builds and data.
If you would like your own Bench, please PM an Admin.
Most Benches are visible only to members.

Pages: 1 2 [3] 4
Author Topic: Magnetic Compression Study" by Cyril Smith  (Read 32114 times)

Group: Professor
Hero Member
*****

Posts: 1863
The 4E2 core from Ferroxcube may not be military, but is a special used in large scale experiments such as particle accelerators.  It is available on special order but likely to be expensive.  Searching Google Images for mu v. temperature to look for other possibilities I came across the use of meta-materials.   The "swiss-roll" material used in the imaging experiment described in the attached pdf could be interesting.  Its complex permeability curve is given and although the mu's are low the ratio from peak mu to low mu is quite high, like 8:1.  Making your own resonant core could offer possibilities.

Smudge
   
Group: Experimentalist
Hero Member
*****

Posts: 1671
Using the circuit of post #47, I have the following results from testing an Arnold TSF7070 ferrite toroid core. I have attached the data sheet below.

The core is wound with 7 turns evenly spread over the surface of the toroid. The linear reference coil is an air coil with 101 turns of 25ga  wound on a 4.84cm/1.90" pvc tube with no spacing between turns. The measured inductance using a current ramp method is 362uh and the self resonant frequency SRF=2.6MHz.

The test results are shown below and if the numbers are correct, this core appears to have a Hi/Lo permeability change of 5153/2736 = 1.88.

I have purchased these on Ebay and they are still available but before purchasing any, let's be sure these test results are reasonably correct!

I plan to test this core in the circuit I previously posted and will post the results when finished.

pm

Edit: Please note that the ur's are reversed on the original schematic posted. This post is now correct.
« Last Edit: 2016-10-04, 20:51:28 by partzman »
   

Group: Administrator
Hero Member
*****

Posts: 3867


Buy me some coffee
Hi Partsman

I just tried your circuit and a large unknown core, there seems to be a constant difference in amplitude between the air core and the 7 turn coil on the core to test, then when i get up to about 780khz the core under test's amplitude increases to match my input amplitude and starts going out of phase.

Yellow chan is across 50 Ohm sig gen source, cyan chan is across coil under test.

Air coil is on a 20cm tube not sure of turns but they are bifilar wound but i am only using 1 winding of the bifilar Inductance measures 164uH

The large core under test has 7 turns wound evenly around the circumference and measures 161uH which by coincidence is very close to the air core inductance.
I measured the SRF of the air core coil at 2.64MHz by scoping and using a grid dip meter to excite it.

I suspect this core has a flat resonance frequency, previous inductance tests did not show much, i will try to did those tests out when i get back from a short trip tonight.
   
Group: Experimentalist
Hero Member
*****

Posts: 1671
Hi Peterae,

OK, look forward to your results. In the meantime I've had a chance to test the TSF7070 core using the high/low and low/high circuits and here is the problem I've found. The required value of capacitance to resonate with 452uH at 810kHz is 85pfd which is less than the drain to source capacitance of nearly all mosfets making any measurement attempts totally inaccurate IMO.  The high/low test for example yielded a COP ~.5 which I really don't have any confidence is accurate. One thing to note is that there was a slight increase in the clamped inductor current which would indicate that the permeability was decreasing over time.

Hi Partsman

I just tried your circuit and a large unknown core, there seems to be a constant difference in amplitude between the air core and the 7 turn coil on the core to test, then when i get up to about 780khz the core under test's amplitude increases to match my input amplitude and starts going out of phase.

Yellow chan is across 50 Ohm sig gen source, cyan chan is across coil under test.

Air coil is on a 20cm tube not sure of turns but they are bifilar wound but i am only using 1 winding of the bifilar Inductance measures 164uH

The large core under test has 7 turns wound evenly around the circumference and measures 161uH which by coincidence is very close to the air core inductance.
I measured the SRF of the air core coil at 2.64MHz by scoping and using a grid dip meter to excite it.

I suspect this core has a flat resonance frequency, previous inductance tests did not show much, i will try to did those tests out when i get back from a short trip tonight.

Hi Peterae,

OK, look forward to your results. In the meantime I've had a chance to test the TSF7070 core using the high/low and low/high circuits and here is the problem I've found. The required value of capacitance to resonate with 452uH at 810kHz is 85pfd which is less than the drain to source capacitance of nearly all mosfets making any measurement attempts totally inaccurate IMO.  The high/low test for example yielded a COP ~.5 which I really don't have any confidence is accurate. One thing to note is that there was a slight increase in the clamped inductor current which would indicate that the permeability was decreasing over time.

pm
   

Group: Professor
Hero Member
*****

Posts: 3366
Just found this paper hidden among my files.  It may help in understanding just what complex permeability means and how it effects circuits.  It
You should make an appendix to this paper instructing how to measure it with VNAs.
VNA is a standard lab instrument that provides phase vs. frequency information as well as S parameters in arbitrary 2-port networks.

I have never seen complex permeability graphed on a Smith chart, did you?  (Cole-Cole chart notwithstanding)
« Last Edit: 2016-10-05, 00:01:28 by verpies »
   
Group: Tech Wizard
Hero Member
*****

Posts: 1123
Hi Folks,

I went through the 'in stock' offer of some core distributors on ferrite cores they sell that may have a "promising" complex permeability peak change vs frequency as Smudge explained in his paper. I found two core materials, N87 and N97 from Epcos/TDK, these are available in toroidal shape (for the N87 material) and in not toroidal but various core shapes (for the N97 material).

Here are the choices for the N87 ring cores with prices:
http://www.mouser.co.uk/Passive-Components/EMI-RFI-Components/EMI-RFI-Suppressors-Ferrites/Ferrite-Toroids-Ferrite-Rings/_/N-bw7t9Zscv7?Keyword=b64290+n87&FS=True&Ns=Pricing|1

Here are the choices for the N97 material for various core shapes with prices:
http://www.digikey.com/product-search/en/magnetics-transformer-inductor-components/ferrite-cores/5113149?k=&pkeyword=&pv70=744&FV=fff4004e%2Cfff8053d&mnonly=0&newproducts=0&ColumnSort=0&page=1&stock=1&quantity=0&ptm=0&fid=0&pageSize=25
   

Here are the choices for the PE22 material, core shapes are EC and EE:
http://www.mouser.co.uk/Passive-Components/EMI-RFI-Components/EMI-RFI-Suppressors-Ferrites/Ferrite-Toroids-Ferrite-Rings/_/N-bw7t9Zscv7?Keyword=pe22&Ns=Pricing%7c1&FS=True   

I included the curves of complex permeability change vs frequency for the N87 and N97 core materials, for the N87 I estimated 40% gain and for the N97 I estimated 52% gain if I am not mistaken.

Gyula
   

Group: Professor
Hero Member
*****

Posts: 1863
You should make an appendix to this paper instructing how to measure it with VNAs.
VNA is a standard lab instrument that provides phase vs. frequency information as well as S parameters in arbitrary 2-port networks.

I've used network analyzers in the past (HP and Rhode & Schwartz) for L band measurements.  But that was a lifetime ago, I'll leave it to the younger generation to do what you suggest.

Quote
I have never seen complex permeability graphed on a Smith chart, did you?
No, and I can't claim any connection to the Smith who invented it.   Used the Smith chart a lot to impedance match antenna over a wide frequency band (again at L band).
Quote
(Cole-Cole chart notwithstanding)
WTH is a Cole-Cole chart?

Smudge
   

Group: Professor
Hero Member
*****

Posts: 3366
It is a 2D chart, on which the real part of complex permittivity is graphed on the X axis and the imaginary part is graphed on the Y axis. ( see here )
   

Group: Administrator
Hero Member
*****

Posts: 3867


Buy me some coffee
Thanks gyula hopefully that will give us some more to look at.

partsman

unfortunately i am not ready to test this yet, i will need to order some bits and build the drive electronics.

Is the Drain Source capacitance going to be a problem?.

For an IRF840 for instance it looks to be CDS = COSS−CRSS or 200pf which is not so bad

but a logic level fet IRLZ44 for instance is 1200-200 or 1nF  :o

i need to do some more research on this i think  O0

EDIT
Added logic fet IRLB8743PBF = 300-105 = 195pf
IRFI530NPBF = 160-88=72pf
IRLI530NPBF= 160-90=70pf

trouble is that when Vds is low 0v then COSS is larger
« Last Edit: 2016-10-05, 21:40:32 by Peterae »
   
Group: Experimentalist
Hero Member
*****

Posts: 1671
Thanks gyula hopefully that will give us some more to look at.

partsman

unfortunately i am not ready to test this yet, i will need to order some bits and build the drive electronics.

Is the Drain Source capacitance going to be a problem?.

For an IRF840 for instance it looks to be CDS = COSS−CRSS or 200pf which is not so bad

but a logic level fet IRLZ44 for instance is 1200-200 or 1nF  :o

i need to do some more research on this i think  O0

Peterae,

Yes, in my tests using the IRF636 with a Cds~130pfd it is a problem as I require an 83pfd cap for the high frequency storage. IOW, i can't reach the 810kHz peak permeability frequency.  The other problem with the mosfet Cds is non linearity of C0ss. It is greater than the spec sheet data at low voltages and decreases with drain voltage increase.  As I see it, we need a core material like Smudge spec'd that has a low overall permeability and a reasonable high frequency permeability peak. This would allow the use of higher capacitance values for the high frequency peak.

There are possible solutions with certain mosfets. I have some IRF5802 with a Cds ~20pfd but it is smd and a bugger to handle, at least for me!

So, all things considered, I'm now focusing on altering core permeability with PM biasing which is where this is all headed anyway IMO. Nothing to report at the moment however.

pm
 
   
Group: Tech Wizard
Hero Member
*****

Posts: 1123
....

There are possible solutions with certain mosfets. I have some IRF5802 with a Cds ~20pfd but it is smd and a bugger to handle, at least for me!

...
 

Hi partzman,

You could make or buy a simple adapter that would host the smd MOSFET and "convert" it to a bigger size.  It would still make you use a lupe to mount the TSOP-6 package but it needs to do once... and then you can handle it easier any time.
Perhaps it could be plugged into a TSOP socket or mount onto such board like this link shows (ebay also has many smd adapters):
http://www.newark.com/capital-advanced/33206/smd-adapter-6-sc-59-6-sot-23-6/dp/10M5369    
or you could make such from a small pcb piece.

Gyula
   
Group: Experimentalist
Hero Member
*****

Posts: 1671
Hi partzman,

You could make or buy a simple adapter that would host the smd MOSFET and "convert" it to a bigger size.  It would still make you use a lupe to mount the TSOP-6 package but it needs to do once... and then you can handle it easier any time.
Perhaps it could be plugged into a TSOP socket or mount onto such board like this link shows (ebay also has many smd adapters):
http://www.newark.com/capital-advanced/33206/smd-adapter-6-sc-59-6-sot-23-6/dp/10M5369    
or you could make such from a small pcb piece.

Gyula

Hi Gyula,

Thanks for the suggestions. I have some adapters that I used to mount the TSOP-6 devices but they aren't an exact fit as they are SOT-6 as I recall. After soldering a few, I'm a bit cross-eyed for a while! :o

pm
   

Group: Administrator
Hero Member
*****

Posts: 3867


Buy me some coffee
I managed to get the fet drive electronics built today, i used a pcb i made some time ago that has a 18V supply on board to drive a dual fet driver, originally built to be driven from 2 onboard pic micros, having found i had some 74HC123 monostable chips in my box i used this dual mono and built the mono circuit on vero to plug into the fet driver board.

Just need to order some FETs now and i will be ready to test partsmans circuit.

the scope shot shows
mono 1 output yellow
mono 2 output cyan
fet driver output 1 purple
fet driver output 2 blue

The fet driver is a dual TC4427

I was amazed how good these monostables are i could get under 100ns pulse set using the pot.

PS the soldering looks bloody terrible on the pcb LOL, looked all right with my glasses on but looks like i need to resolder it using a magnifying glass. Oh the joy of getting old.
   

Group: Administrator
Hero Member
*****

Posts: 3867


Buy me some coffee
   

Group: Administrator
Hero Member
*****

Posts: 3867


Buy me some coffee
Just testing the 3C95 or N95 core, i wound 6 turns on the core and measured 121uH, my air core is the same as before at 165uH

loose coupled oscillator shows SRF at 1.94MHz

Using Partzman's circuit i tested at 100KHz and the found the frequency at which the Ch1-CH2 difference was smallest, strangely this appears to be above the SRF at 2.744MHz see scope shots.

   

Group: Administrator
Hero Member
*****

Posts: 3867


Buy me some coffee
I am setup as per post 37

So core is N95 or in this case equivalent 3C95 as per previous post

i used a 120pf cap for my high cap silver mica and a 20nf for my low cap.

I used 2 x IRF640NLPBF Fets

First scope shot is to show overall picture of whats happening and triggered from CH1 yellow trace
Second scope shot is triggered of CH2 cyan and zoomed in.
Third scope shot zoomed in on time-base more
Fourth scope shot is on the gates
   
Group: Experimentalist
Hero Member
*****

Posts: 1671
I am setup as per post 37

So core is N95 or in this case equivalent 3C95 as per previous post

i used a 120pf cap for my high cap silver mica and a 20nf for my low cap.

I used 2 x IRF640NLPBF Fets

First scope shot is to show overall picture of whats happening and triggered from CH1 yellow trace
Second scope shot is triggered of CH2 cyan and zoomed in.
Third scope shot zoomed in on time-base more
Fourth scope shot is on the gates

Hi Peterae,

Something doesn't look quite right to me but perhaps I don't understand your scope probe connections so it would help if you identify where your probes are attached in the schematic.

In looking at the 3C95 data sheet, I see the permeability resonance peak is at ~750kHz and this is approximately the frequency the CH1-CH2 minimum test should have indicated.

I also can't justify your value of 120pfd for high frequency resonance at 2.744MHz with 121uH if I understand correctly. Chigh should be ~28pfd in this case.

The problem I experienced in this Hi to Lo test is that the capacitance of L1 seemed to create a divider with Chigh that made the high frequency half sine impossible to evaluate with my setup.

pm
   

Group: Administrator
Hero Member
*****

Posts: 3867


Buy me some coffee
Hi partzman
Sorry i was really short of time this weekend so it was a fast setup before going out, each scope probe is across each capacitor, so the yellow trace is across the higher value (19.56nF) cap and the Cyan across the lower value cap (120pf)

yes you are right about the 120pf, problem i have is that i have no decent lower value caps that are rated over 100V, i was hoping that from the above test i could get an idea of what the fet capacitance is and see what value i needed to aim for and then buy some caps, i had not realized the value would be so low though, only just started looking at the test results myself, now i need to try and understand what's going on.

The high voltage peak base looks to be about 800ns is this my high frequency tuned period 1/800ns  = 1.25MHz half cycle so my resonant frequency is = 625KHz, if thats correct for my 121uH then that equates to a capacitance of 535pf, which means my fet is having a massive effect.



EDIT
Quote
In looking at the 3C95 data sheet, I see the permeability resonance peak is at ~750kHz and this is approximately the frequency the CH1-CH2 minimum test should have indicated.
I cannot explain my result either.

I just found some blue 3kv ceramic caps in a box with varying values, would they be any good for my high frequency capacitance, i believe that silver mica are probably best but maybe i can give a 47pf 3kv ceramic cap a go and see what the difference is on the pulse.
« Last Edit: 2016-10-16, 21:57:03 by Peterae »
   

Group: Administrator
Hero Member
*****

Posts: 3867


Buy me some coffee
Could we use 2 fets in series to half our fet capacitance as per a bidirectional fet switch.


http://electronics.stackexchange.com/questions/79028/understanding-two-mosfet-with-sources-connected

ah i knew Matt whats built one of these, here is his link
http://www.overunityresearch.com/index.php?topic=2594.msg41116#msg41116

   
Group: Tech Wizard
Hero Member
*****

Posts: 1123
Hi Peter,

Yes, for the bidirectional MOSFET switch the resulting drain-source capacitance may get lower than any of the single devices has. However, to be precise, they would not be halved because the capacitances involved are voltage dependent.  The total voltage may be halved across each device but the device capacitances may not, due to the lower (halved) voltages across each. We know that this capacitance increases as the drain-source voltage is lowered, and the dependence is nonlinear.
Would like to show a simpler circuit than what Matt used. His circuit is enhanced with the fast diodes in parallel with the body diodes and also with the fast driver ICs.  Of course, these are application specific, I uploaded this schematic to Doug Konzen at the EVGray yahoo group years ago, when he tinkered with coil shorting in motors if I recall correctly, this is why the reed switch with the series 10k (or lower) was included at the driver chip input.  A function generator may also give the input there instead, and of course other circuit variations are possible.

Gyula
   
Group: Experimentalist
Hero Member
*****

Posts: 1671
Hi partzman
Sorry i was really short of time this weekend so it was a fast setup before going out, each scope probe is across each capacitor, so the yellow trace is across the higher value (19.56nF) cap and the Cyan across the lower value cap (120pf)

yes you are right about the 120pf, problem i have is that i have no decent lower value caps that are rated over 100V, i was hoping that from the above test i could get an idea of what the fet capacitance is and see what value i needed to aim for and then buy some caps, i had not realized the value would be so low though, only just started looking at the test results myself, now i need to try and understand what's going on.

The high voltage peak base looks to be about 800ns is this my high frequency tuned period 1/800ns  = 1.25MHz half cycle so my resonant frequency is = 625KHz, if thats correct for my 121uH then that equates to a capacitance of 535pf, which means my fet is having a massive effect.



EDITI cannot explain my result either.

I just found some blue 3kv ceramic caps in a box with varying values, would they be any good for my high frequency capacitance, i believe that silver mica are probably best but maybe i can give a 47pf 3kv ceramic cap a go and see what the difference is on the pulse.

Hi Peterae,

I had responded to your post earlier but I see it didn't make it so I'll try again.  You might try to remove your 120pfd cap which would reduce the actual capacitance to 415pfd. This should make your hi frequency resonance ~710kHz which is close to the 3C93 data sheet.  If you have the device socketed, swap out the 640's and see if you can get closer to the 750kHz target.

However, let's look at your results of the numbers you have posted. It appears that the 19.56nfd cap has been charged to 17.9 volts from the dc supply so this equates to an energy of 3.13uJ. The output cap at 535pfd has reached a peak of 123 volts and this equates to 4.05uJ.  If I read your scope traces correctly, this equates to a gain of 129%.  To be at 100%, Chigh would be 414pfd.

pm
   

Group: Administrator
Hero Member
*****

Posts: 3867


Buy me some coffee
Thanks for your replies gyula and partzman.

ok so first then i can try to totally remove the 120pf and rely on the fet capacitance and see what happens.

Not sure why the post did not get posted, it can do that if someone posts while you posted and warns you that, the only other way is a timeout when the page does blank.

yes i can also try different fets at some point, it maybe interesting to try the fet switch as well.
   

Group: Administrator
Hero Member
*****

Posts: 3867


Buy me some coffee
I cut the high frequency cap out of circuit now so there is only the fet capacitance.

Quote
The high voltage peak base looks to be about 800ns is this my high frequency tuned period 1/800ns  = 1.25MHz half cycle so my resonant frequency is = 625KHz, if thats correct for my 121uH then that equates to a capacitance of 535pf, which means my fet is having a massive effect.
The above was based on the fact that the inductance of the coil under test was 121uH but having thought about that the whole idea is that the inductance will change, so i cannot see how we can do any calculations to work out the fet capacitance because at high frequency we have an unknown fet capacitance and an unknown inductor inductance.

EDIT
Its quiet interesting the capacitance changes when the pulse goes from rising to falling, when rising it takes 260ns and when falling it takes 340ns which corresponds to a frequency of 961KHz and 735KHz respectively.

So if the inductance was the same 121uH at 961KHz the we have a capacitance off 226pf when rising and 387pf when falling.

I suppose really what i need to do is now try caps to see what cap value i need to use to get resonance of 961KHz and then calculate the inductance from that and then we would also know for sure the fets capacitance which should give us a more accurate estimation of the energy in the fets capacitance.

So in the above experiment i have an initial voltage of 18v and a cap of 19.56nF = 3.17uJ
and the output voltage of 178V and a estimated cap value of 226pf = 3.58uJ or a gain of 113%

so if the theory that the inductance increases at resonance then we can say that the capacitance must be smaller than our above calculations, lets say that the inductance increases by 5% then we can assume we have a inductance of  127.05uH and therefore out fet's capacitance would calculate out to be 215.88pf and a total stored energy at 178v of 3.4199uJ so we are still > unity at 5% increase in inductance which is interesting.

How about a 30% increase in inductance, 157.3uH giving a capacitance of 174.37pf and energy at 2.762uJ, so now we are under unity.

So it looks like it really comes down to how much our inductance increases at resonance.
« Last Edit: 2016-10-17, 19:05:43 by Peterae »
   
Group: Experimentalist
Hero Member
*****

Posts: 1671
I cut the high frequency cap out of circuit now so there is only the fet capacitance.
The above was based on the fact that the inductance of the coil under test was 121uH but having thought about that the whole idea is that the inductance will change, so i cannot see how we can do any calculations to work out the fet capacitance because at high frequency we have an unknown fet capacitance and an unknown inductor inductance.

EDIT
Its quiet interesting the capacitance changes when the pulse goes from rising to falling, when rising it takes 260ns and when falling it takes 340ns which corresponds to a frequency of 961KHz and 735KHz respectively.

So if the inductance was the same 121uH at 961KHz the we have a capacitance off 226pf when rising and 387pf when falling.

I suppose really what i need to do is now try caps to see what cap value i need to use to get resonance of 961KHz and then calculate the inductance from that and then we would also know for sure the fets capacitance which should give us a more accurate estimation of the energy in the fets capacitance.

So in the above experiment i have an initial voltage of 18v and a cap of 19.56nF = 3.17uJ
and the output voltage of 178V and a estimated cap value of 226pf = 3.58uJ or a gain of 113%

so if the theory that the inductance increases at resonance then we can say that the capacitance must be smaller than our above calculations, lets say that the inductance increases by 5% then we can assume we have a inductance of  127.05uH and therefore out fet's capacitance would calculate out to be 215.88pf and a total stored energy at 178v of 3.4199uJ so we are still > unity at 5% increase in inductance which is interesting.

How about a 30% increase in inductance, 157.3uH giving a capacitance of 174.37pf and energy at 2.762uJ, so now we are under unity.

So it looks like it really comes down to how much our inductance increases at resonance.

Peterae,

A suggestion, it appears that your on time for Vpulse2 might be a little short and if so, it won't hurt your measurements to increase it some. This might be creating the apparent difference in the parametric capacitance of the fet from rising to falling edge but then again maybe not.

pm

   

Group: Administrator
Hero Member
*****

Posts: 3867


Buy me some coffee
Not had a chance to try increasing the time period for M2 yet but definitely worth trying.

So tonight i have measured the cap value required to resonate at 961KHz & 100KHz to see what the inductance calculates out to

for 961KHz i had to dial a cap value of 156pf and for 100KHz 17.7nf, the inductance calculates to 175.82uH for 961KHz & 143.12uH for 100KHz

So at 961KHz we have a fet capacitance value of 156pf which was at 178V so our energy calculates to 2.471u Joules

and using the same process to work out our start energy we have a cap value of 17.7nf at 18V which calculates out at 2.867u Joules

so unless i have something wrong i calculate an efficiency of 86.2%

anyone agree or disagree with my findings?

It is interesting that the efficiency seemed higher with the 120pf cap, it will be interesting to try another core type, i ordered a high frequency type which looked good on the permeability chart i think it was 4C65 and will try this at the weekend.

EDIT
Actually in hindsight i should have used the measured value of the low capacitance for input energy and then this is as previously posted, i want to remeasure this value at some point just to confirm it measures 19.56nF
Quote
So in the above experiment i have an initial voltage of 18v and a cap of 19.56nF = 3.17uJ

so with an input of 3.17uJ and an output of 2.471u Joules this is worse still at 78% efficiency.


« Last Edit: 2016-10-18, 21:21:14 by Peterae »
   
Pages: 1 2 [3] 4
« previous next »


 

Home Help Search Login Register
Theme © PopularFX | Based on PFX Ideas! | Scripts from iScript4u 2024-05-05, 17:54:26