PopularFX
Home Help Search Login Register
Welcome,Guest. Please login or register.
2026-01-29, 10:09:31
News: Forum TIP:
The SHOUT BOX deletes messages after 3 hours. It is NOT meant to have lengthy conversations in. Use the Chat feature instead.

Pages: 1 2 3 4 5 6 7 8 9 10 11 [12] 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 ... 31
Author Topic: partzmans board ATL  (Read 36294 times)
Topic Options img
img  Chet K
Re: partzmans board ATL « Reply #275 on: 2022-01-21, 19:22:44 »
Group: Ambassador
Hero Member
*****

Posts: 4725
Quote from: partzman on 2022-01-21, 16:18:57
Well, I hope nobody has spent any time of any of the previous RLE posts because there is an error.

The ".ic" or initial conditions statement should have the voltages at VL2 and VL3 set to 300v instead of zero.  This had the effect of creating a slight offset current in L1,L2 and L3 which favored the OU measurements.

I'm sorry to have unintentionally mislead anyone so I apologize.  I will be taking a break from this all research for a time.

Regards,
Jon
////—////——-////—-//////////


Jon
Your tenacity and brutal honesty inspires!

Thanks
Chet
   
img  partzman
Re: partzmans board ATL « Reply #276 on: 2022-06-05, 16:24:15 »
Group: Moderator
Hero Member
*****

Posts: 2232
All,

My research into the Holcomb device has lead to the following development in generation of extra gain with current driven devices.  Holcomb, as most know, uses a rotating electromagnetic field that induces to a stationary stator.  His architecture that interested me the most was his planar device that was intended for cell phone charging and similar applications.  Attempts to simulate or build a simplified device to prove his concept is what lead to this discovery shown below.

The simulation which uses gyrator-capacitor models for the coil-core characteristics is built around an EC-52 ferrite core with windings an all legs.  The center leg coil could be considered the stator and the outside leg coils combine for the rotor.  The stator or center coil is biased with 200ma from a constant current inductor L2.  One outer leg is then ramped to 200ma via a current source and after a 2us delay, the other outer leg is ramped to 200ma.  As can be seen from the plot, each rotor ramp produces an increase in the S1 stator current and also L2. 

This device is considered asymmetrical for two reasons.  First, when considering the coil polarities and the condition when all windings are charged to ~200ma, the apparent total inductance of the windings is P1 buck P2 buck S1 = 890uH.  This is not intuitive but is apparent if one considers the individual flux paths and direction for each coil. 

Second, we stop the sim at the 6us time mark after the first outer leg is discharged to zero but the second outer leg is still fully charged.  This leaves the core in a condition where the apparent total inductance of the windings is now P1 buck S1 = 2.96mH.  Also, L2 retains a current of 207.8ma from a starting current of 200ma.  Because of these conditions, a gain can be realized via classical electrodynamics.

The source of energy for this device are the two current sources I1 and I2.  Different methods could be used to charge P1 and P2 but this method was chosen for clarity and simplicity.  As can be seen, 68.134uJ is consumed by I2 and 33.555uJ is generated in L1 for a net energy input of 34.58uJ.  The interactions producing these results is complex and will be explained later along with options.

From the increase in L2, we have (.20783^2-.200^2)*.025/2=39.9uJ of gain.  We are already slightly OU but we have more stored energy in the core.  We have 207.ma in S1 and 200ma in P2.  This is the condition of P2 buck S1 which results in an energy of (~.200^2)*.00296/2 = 59.2uJ for a total gain of 99.1uJ .  This results is an apparent COP=2.86 .

Regards,
Pm

Please note: I did not account for the starting energy in S1 in the above which if done lowers the COP<1!
« Last Edit: 2022-06-05, 22:16:34 by partzman »
   
img  partzman
Re: partzmans board ATL « Reply #277 on: 2022-06-07, 22:58:26 »
Group: Moderator
Hero Member
*****

Posts: 2232
Well, here is another asymmetrical transformer that I've bench tested that has some interesting characteristics and performance.  Of course I don't expect anyone to really get excited about it due to my past failures plus we all know OU isn't possible.  Right?

Anyway, the xfmr assembly is unique in that the secondary "Sec" consists of two flat coils that are stacked and placed in the center leg gap of an EC-52 ferrite core as seen in the pix.  This coil is induced by the dB in both the gap and the outer core leg it surrounds.

The next pix is the schematic of the circuitry.  Initially, S2 and S3 are turned on and a current ramp begins in both P1 and P2.  P2 begins to ramp due to diode D1 and the voltage polarity of P2.  IOW, P1 and P2 are in a buck mode and their combined apparent inductance is 8.2mH.  During this same time period, a positive voltage appears on the junction of "Sec" and "Lcc".  This has an effect of reducing the 434.6ma starting bias current in both Sec and Lcc.

After 20.16us,  S2 and S3 turn off and S1 turns on.  P1 then begins to discharge through D1 to P2 and also to C1 creating a half sine voltage across C1.  At ~48us in time, the voltage across C1 reaches zero after discharging it's energy into P2 and the entire cycle is considered finished.  During this time, the voltage on the junction of Sec and Lcc goes negative and the current both Sec and Lcc increases as a result.  This is key and will be explained later.

Another key factor in this transformer is the very low coupling factor of k=.049 between when the primaries are in bucking mode relative to Sec.  The result of this is that when Lcc is clamped by S4 at the end of the cycle, the ending currents in both Lcc and Sec have very little difference.  IOW, the secondary is pretty much independent of P1 and P2 under these conditions.

Scope pix #1 shows the Pin as 1.542w over 20.16us for a Uin = 31.1uJ.

Pix #2 and #3 show the starting and ending currents in Lcc and Sec of 434.6ma and 440.1ma respectively.  With the net inductance of Lcc and Sec being 26.6mH, the increase in energy is 63.9uJ .

Pix #4 and #5 show the ending currents in P1 and P2 at the end of the cycle to be 55.76ma and 83.94ma respectively.  The average therefore is 69.85ma and with the P1 P1 buck inductance at 8.2mH, the ending energy level in P1 and P2 combined is 20uJ .  There is also a gain from the average of the differential currents in the P1 aid P2 mode but is ignored here.

The resultant apparent COP = 84/31.1 = 2.7 .

Pix #6 shows the level of current in the secondary to be nearly equal to the current in Lcc after clamping.

Pix #7 and #8 show an important aspect of the topology and that is the asymmetry of the voltage area uVs across Sec and Lcc for the charging and discharging phases of the device.  IOW, the dI in Sec and Lcc is greater for the discharge phase and is one means of gain for the device.

Regards,
Pm   

   
img  PhysicsProf
Re: partzmans board ATL « Reply #278 on: 2022-06-08, 00:11:54 »
Group: Professor
Hero Member
*****

Posts: 3230
  Thank you for sharing and for all this work, Jon!!   
   God bless you!
   
img  partzman
Re: partzmans board ATL « Reply #279 on: 2022-06-08, 14:57:13 »
Group: Moderator
Hero Member
*****

Posts: 2232
Quote from: PhysicsProf on 2022-06-08, 00:11:54
  Thank you for sharing and for all this work, Jon!!   
   God bless you!

Thanks Steve!

Jon
   
img  partzman
Re: partzmans board ATL « Reply #280 on: 2022-06-10, 18:26:27 »
Group: Moderator
Hero Member
*****

Posts: 2232
This is a version of the device posted previously that has been cleaned up and slightly modified in operation.  The flat coils fit tightly in the gap in the original so as to create a slight gapping in the outer legs.  This version has the gap opened up so the flat coils have a slip fit resulting in no gaps in the outer legs.  Therefore P1, P2 and S1 all have slightly higher inductances.

A new schematic is shown that has the updated inductance changes and all the coupling factors.

In the scope pix, CH1(yel) is the switch drive voltage V2, Ch2(grn) is the current probe, CH3(pnk) is the voltage across C1 unless otherwise specified, and CH4(blu) is the supply voltage.

The 1st pix is Pin at 2.016 watts for 20.16us for a Uin = 40.64uJ.

The 2nd and 3rd pix show the starting and ending currents in ILcc of 434.9ma and 443.3ma respectively for an energy gain of 91.1uJ .

The 4th, 5th, and 6th pix show the ending currents in P1, P2, and S1 of 56.73ma,94.52ma, and 441.8ma respectively.

The 7th pix shows the accounting of the stored energy from the ending currents in P1, P2, and S1 from simulation.  The sim parameters closely match the bench parameters and is an easy way to display the results without creating the complex switching on the bench device.

The starting current in S1 is 435ma and therefore the starting energy in S1 is 492uJ.  In the sim analysis, P1, P2, and S1 are each assigned their respective ending currents and then allowed to discharge into C1 until the currents in all windings are equal thus allowing for an accurate accounting of the network's stored energy.  We first see that C1 reaches a voltage level of 258.4 volts which equates to 333.8uJ of energy.  Next we see that the winding currents have an equal 184ma and the network inductance in the config shown has an inductance of 10.8mH .   The results in a stored energy of 182.8uJ for a total network stored energy level of 516.6uJ .

These levels result in a net gain of 516.6uJ-492uJ = 24.6uJ .

We can now account for all energies in the device and calculate the apparent COP as (91.1+24.6)/40.64 = 2.85 which is relatively close to the previous post analysis.

This was created in a limited time span as I have to visit my wife in the nursing home so I will check it thoroughly later and make any corrections necessary.

I already see that I missed pointing out that in the 6th pix, CH3 shows the clamp drive voltage which is turned on shortly after start.  The reason for this is that is prevents the positive voltage across S1 from decreasing the current in Lcc thus slightly improving performance.  One may also notice the non-linearity of the voltage across C1.  More later.

Regards,
Pm

   
img  partzman
Re: partzmans board ATL « Reply #281 on: 2022-06-13, 20:42:21 »
Group: Moderator
Hero Member
*****

Posts: 2232
For those who are interested, this is an operational equivalent to the flat coil placed in the gap in the previous posts.  As hopefully can be seen , the coil surrounds the inner and outer leg on the bobbin.  This design is much easier to implement and gives the opportunity to change the center leg gap.  The induction principles are the same as previous or IOW, the circuit operation is identical.

The previous examples of this device suffered from saturation in the outer legs and was corrected in present versions with a small gap that allows currents in the 1000ma range and greater.  The highest COP achieved to date is ~3.2 .  These numbers are created on the bench completely that is, the device is run and the resulting remainder currents in P1, P2, and S2, are set up on the bench with current DC sources and then discharged into a storage capacitor to determine with fair accuracy what the core's stored energy level really is.

Regards,
Pm
   
img  PhysicsProf
Re: partzmans board ATL « Reply #282 on: 2022-06-14, 03:26:38 »
Group: Professor
Hero Member
*****

Posts: 3230
  Good work, Jon!  exciting...    O0
   
img  partzman
Re: partzmans board ATL « Reply #283 on: 2022-06-15, 15:17:03 »
Group: Moderator
Hero Member
*****

Posts: 2232
Some interesting comparisons.

The core assembly with the 72 turn flat coil secondary placed in the center gap with .010" gaps in the outer legs to linearize the core, produced an apparent COP = 1.3 .

The same core assembly using the 70 turn bobbin wound secondary with an S1/Lcc starting current of 434ma, produced an apparent COP = 2.18 .

The input power/energy stays the sames no matter the starting current level in the S1 secondary and the  constant current inductor Lcc.  Therefore, if the starting current is lowered to a certain level, the COP will be <1.

OTOH, if the starting current is increased, the COP will increase.  For example again with the same core assembly, a starting current in S1/Lcc of 637ma produces an apparent COP = 2.70 .

Regards,
Pm
   
img  partzman
Re: partzmans board ATL « Reply #284 on: 2022-06-16, 01:11:16 »
Group: Moderator
Hero Member
*****

Posts: 2232
This is a revised version of the previous design.  This Rev2 utilizes a scheme that allows the removal of the Lcc clamp thus yielding a slightly lower Pin while increasing efficiency.  The P1 and P2 primaries are charged in a buck mode and then P2 is clamped during discharge creating asymmetrical coil currents that increase the currents in Lcc/S1.

Currently, one such version operates at an apparent COP=3.15 .

Regards,
Pm 
   
img  partzman
Re: partzmans board ATL « Reply #285 on: 2022-06-16, 15:11:08 »
Group: Moderator
Hero Member
*****

Posts: 2232
This is a visual of the statement made above regarding the asymmetrical currents in the secondary.  In this case, it is the asymmetrical voltage created in the open secondary during the charge and discharge phases of the primary windings.  This large ratio of the voltage areas is what creates the asymmetrical current increase in Ls/Lcc.  IOW, there is very little constant current decrease during the charge phase but a large increase during the discharge phase.

Now one might say "well, I can create a secondary with zero voltage on one phase and then a large voltage on the other phase with a simple diode".  Yes that's true, but there is no bias current in the windings.  Try the same circuit with a bias current in the secondary and view the results.  This is what makes this secondary coil winding special.

Regards,
Pm
   
img  partzman
Re: partzmans board ATL « Reply #286 on: 2022-06-16, 19:00:51 »
Group: Moderator
Hero Member
*****

Posts: 2232
It appears the ratio of the area of the center leg verses the outer leg does play an important part in the device gain.  Normally, in a standard ferrite E core, the outer leg is half the area of the center leg.  This is done to maintain and equal flux density throughout the core.  The EC series cores however have an outer leg that has ~81% of the area of the center leg.  I think ideally for this concept, all legs should have equal area but this is TBD.

The following scope traces show the differences between a standard 1/4" E core and a modified core of the same type with equal legs.

Regards,
Pm
   
img  Itsu
Re: partzmans board ATL « Reply #287 on: 2022-06-16, 20:01:25 »

Group: Elite Experimentalist
Hero Member
*****

Posts: 4695
PM,

nice project you are working on, also the presented COP figures are promising.

I understand you use EC52 E-cores, not sure what you mean in your latest post with "a standard ¼" E-core", as ¼" would be rather small (0.6cm).

Anyway, found some EC52 cores online which are designated as EC52/24/14 and either made of 3C91 or 3C94 ferrite and seems to be ungapped while your cores are obviously gapped.

What about the used wires, i see some multi stranded wire, is that litz wire or several magnet wires combined?

I have some E-cores here of similar size, but they are gapped, but i am not sure what ferrite is used.

Thanks,  regard Itsu
   
img  partzman
Re: partzmans board ATL « Reply #288 on: 2022-06-16, 21:27:54 »
Group: Moderator
Hero Member
*****

Posts: 2232
Quote from: Itsu on 2022-06-16, 20:01:25
PM,

nice project you are working on, also the presented COP figures are promising.

I understand you use EC52 E-cores, not sure what you mean in your latest post with "a standard ¼" E-core", as ¼" would be rather small (0.6cm).

Anyway, found some EC52 cores online which are designated as EC52/24/14 and either made of 3C91 or 3C94 ferrite and seems to be ungapped while your cores are obviously gapped.

What about the used wires, i see some multi stranded wire, is that litz wire or several magnet wires combined?

I have some E-cores here of similar size, but they are gapped, but i am not sure what ferrite is used.

Thanks,  regard Itsu

Itsu,

The last post with the 1/4" cores was a comparison test for the secondary output voltages and I used the small cores as I had them both on hand from previous work.  So, it is good that you found some EC-52 cores and the material type is not that important to test the concept and the larger than normal outside legs improves performance.

Yes, the cores I've been using have the center leg gapped anywhere from .030"-.060" (.75mm-1.5mm) and the outside legs have a .010" (.254mm) gap on each leg.  I will make a bench test with un-gapped cores but an overall gap of .010" or greter for your reference.  The general idea is to try to maintain as low a secondary inductance as possible for a given number of turns and that is controlled by the center and outside gaps.  Eventually I hope to have some math to optimize some of these variables.

I used 8-34 litz on the primaries that are 162T each but you could use 26-24ga awg and should work fine.  The secondary is 70T of 26ga awg but again any gauge that is close that will fit on the core.  I 3D printed a special bobbin that fits over the center and outside leg but you could insulate the core with tape and wind directly on the core legs as an option.  Should work the same.

I still learning the device myself so I'm sure there will be some improvements along the way.

Regards,
Pm   
   
img  partzman
Re: partzmans board ATL « Reply #289 on: 2022-06-16, 23:18:16 »
Group: Moderator
Hero Member
*****

Posts: 2232
I ran a bench test with the EC52 cores gapped .02"(.5mm) on all legs with all winds as previously specified, a power supply of 32v DC, and a frequency of 25kHz.  You should see an apparent COP~2.5 .  I think larger gaps would provide a better COP but not sure until tested.

Also, the starting current in Lcc/Ls is ~430ma.  If you increase the starting current, the COP will also increase.

I should also clarify that P1 is under S1 and is diode clamped.

Regards,

Pm
   
img  partzman
Re: partzmans board ATL « Reply #290 on: 2022-06-16, 23:41:24 »
Group: Moderator
Hero Member
*****

Posts: 2232
This test was done to see if what I suspected was true and it is.  To offset the fact that the outer leg of most E cores has a smaller area than the center leg, to maintain a constant flux density, a simple turns adjustment seemed logical to compensate.

For example, the 1st and 2nd scope pix below shows the EC52 core assembly with P1 and P2 = 162T.  The areas for the open circuit voltage [OCV] for the charge and discharge phases can be seen for reference.

The 3rd and 4th scope pix shows the same assembly with P1 = 140T and P2 = 162T.  Note that the OCV ratio is worse for this arrangement.

The 5th and 6th pix show the same assembly only now with P1 =162T and P2 = 140T.  Now we see that both OCV's are positive meaning that Lcc/Ls will show an increase in current on both phases.  This also tells us that we can compensate for the smaller area in the outer leg by increasing the turns to increase the H field and thus the flux density.

A test was run with this last configuration and the apparent COP~3.55 .

Regards,
Pm
   
img  Itsu
Re: partzmans board ATL « Reply #291 on: 2022-06-17, 09:39:08 »

Group: Elite Experimentalist
Hero Member
*****

Posts: 4695
Quote from: partzman on 2022-06-16, 21:27:54
Itsu,

The last post with the 1/4" cores was a comparison test for the secondary output voltages and I used the small cores as I had them both on hand from previous work.  So, it is good that you found some EC-52 cores and the material type is not that important to test the concept and the larger than normal outside legs improves performance.

Yes, the cores I've been using have the center leg gapped anywhere from .030"-.060" (.75mm-1.5mm) and the outside legs have a .010" (.254mm) gap on each leg.  I will make a bench test with un-gapped cores but an overall gap of .010" or greter for your reference.  The general idea is to try to maintain as low a secondary inductance as possible for a given number of turns and that is controlled by the center and outside gaps.  Eventually I hope to have some math to optimize some of these variables.

I used 8-34 litz on the primaries that are 162T each but you could use 26-24ga awg and should work fine.  The secondary is 70T of 26ga awg but again any gauge that is close that will fit on the core.  I 3D printed a special bobbin that fits over the center and outside leg but you could insulate the core with tape and wind directly on the core legs as an option.  Should work the same.

I still learning the device myself so I'm sure there will be some improvements along the way.

Regards,
Pm

Thanks PM,

i was wondering why your EC52 cores looks gapped in the middle leg on your photo's, as the specification for EC52 cores show they are all ungapped.

Anyway, i will order some of these EC52 ungapped cores and for now use my gapped (1mm) middle leg EDT49 cores.

I have some litz wire (0.04mm x 200) i could use or else as suggested normal 26-24ga awg.

I have just finished building my 3D printer (Prusa), so if you have any files for the former you want to share please do.

Itsu
   
img  partzman
Re: partzmans board ATL « Reply #292 on: 2022-06-17, 13:47:48 »
Group: Moderator
Hero Member
*****

Posts: 2232
Quote from: Itsu on 2022-06-17, 09:39:08
Thanks PM,

i was wondering why your EC52 cores looks gapped in the middle leg on your photo's, as the specification for EC52 cores show they are all ungapped.

Anyway, i will order some of these EC52 ungapped cores and for now use my gapped (1mm) middle leg EDT49 cores.

I have some litz wire (0.04mm x 200) i could use or else as suggested normal 26-24ga awg.

I have just finished building my 3D printer (Prusa), so if you have any files for the former you want to share please do.

Itsu

Attached are the .stl and g-code files if they will pass thru the forum software, we'll see.  I use PLA+ for the filament in most bobbins.

OK, the .gcode didn't go thru so change the filetype from .png to .gcode.
   
img  PhysicsProf
Re: partzmans board ATL « Reply #293 on: 2022-06-17, 14:49:31 »
Group: Professor
Hero Member
*****

Posts: 3230
Quote from: partzman on 2022-06-17, 13:47:48
Attached are the .stl and g-code files if they will pass thru the forum software, we'll see.  I use PLA+ for the filament in most bobbins.

OK, the .gcode didn't go thru so change the filetype from .png to .gcode.

Does one need to have a 3D printer in order to replicate your device, PM? 
   
img  partzman
Re: partzmans board ATL « Reply #294 on: 2022-06-17, 15:19:55 »
Group: Moderator
Hero Member
*****

Posts: 2232
Quote from: PhysicsProf on 2022-06-17, 14:49:31
Does one need to have a 3D printer in order to replicate your device, PM?

No I wouldn't say so but it would make it easier.  The reason I say this is that both bobbins are not common but I'm using bobbins for the primary that aren't the right size and my original test xfmr used a coil for the secondary that was wound on a suitable form, removed, and then shaped to fit the core.

Pm
   
img  partzman
Re: partzmans board ATL « Reply #295 on: 2022-06-17, 20:27:38 »
Group: Moderator
Hero Member
*****

Posts: 2232
There seems to be a problem with the energy recovery process.

Pm
   
img  partzman
Re: partzmans board ATL « Reply #296 on: 2022-06-18, 15:45:15 »
Group: Moderator
Hero Member
*****

Posts: 2232
Quote from: partzman on 2022-06-17, 20:27:38
There seems to be a problem with the energy recovery process.

Pm

After considerable effort on the recovery of the energies built up in the asymmetrical xfmr assembly, the only variation that appears to produce an apparent COP ~1.95 is the Ver1 schematic as shown in post #280.  All other variations fall short as far as I can tell at present.

This core has the large center gap of ~.070" [1.78mm] and outside gaps at .010" [.254mm] with  162T primaries and 70T secondary.

The recovery is tricky and there may be better solutions that what I'm presently using.

Regards,
Pm
   
img  partzman
Re: partzmans board ATL « Reply #297 on: 2022-06-19, 13:58:33 »
Group: Moderator
Hero Member
*****

Posts: 2232
Quote from: partzman on 2022-06-18, 15:45:15
After considerable effort on the recovery of the energies built up in the asymmetrical xfmr assembly, the only variation that appears to produce an apparent COP ~1.95 is the Ver1 schematic as shown in post #280.  All other variations fall short as far as I can tell at present.

This core has the large center gap of ~.070" [1.78mm] and outside gaps at .010" [.254mm] with  162T primaries and 70T secondary.

The recovery is tricky and there may be better solutions that what I'm presently using.

Regards,
Pm

Well, after more strenuous bench work, it now appears that even this Ver1 has no apparent gain.  I'm still going to say that I have some more ideas to try before absolutely giving up on this topology.

One of my assumptions that got me into trouble here is that when you have identical coils with identical inductances on a common  core,  dissimilar currents bucking or aiding, can be averaged to arrive at an equivalent current for energy calculations.  This is absolutely correct however, when a third coil is introduced on the same core, this does not necessarily hold true which makes logical sense.  In this topology this is exactly what happened and I just missed it!

Pm
   
img  Itsu
Re: partzmans board ATL « Reply #298 on: 2022-06-19, 17:24:46 »

Group: Elite Experimentalist
Hero Member
*****

Posts: 4695

Thanks PM,

no pain, no gain, its all in the game.

Let us know when you think there is an improvement.

Regards Itsu
   
img  partzman
Re: partzmans board ATL « Reply #299 on: 2022-06-20, 13:57:55 »
Group: Moderator
Hero Member
*****

Posts: 2232
Quote from: Itsu on 2022-06-19, 17:24:46
Thanks PM,

no pain, no gain, its all in the game.

Let us know when you think there is an improvement.

Regards Itsu

Thank you Itsu and I will let you know on any improvements.

Regards,
Pm
   
Pages: 1 2 3 4 5 6 7 8 9 10 11 [12] 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 ... 31
« previous next »


 

Home Help Search Login Register
Theme © PopularFX | Based on PFX Ideas! | Scripts from iScript4u 2026-01-29, 10:09:31