Hi Barry, I did rework the transformer a bit. Exactly as you said, increasing the primary turns from 2 to 4, reduced the frequency to 56kHz to get the maximum penetration due to skin effect. Reduced the magnetic flux from 1800 to 1050 to avoid core saturation, probably due to flux walking. I did rework the load also a bit from 290W to about 100W. The results do look good. The MOSFET's are now warm, instead of being very hot. But still, I do see that the transformer still draws about 3A wit= h no load, which seems a bit on the higher side to me ? Maybe this is due to the leakage flux ? Or is it the magnetization current = ? I have would Primary1 closest to the core, the Secondary on top of it and Primary2 on top of the secondary. Would winding the primaries in a bifiliar fashion, reduce this no-load curr= ent drawn ? Any pointers would be helpful. Thanks, Manu On Thu, Mar 21, 2013 at 1:31 AM, Barry Gershenfeld wro= te: > Vasile's right--I didn't read the whole email. I skipped down to the > bottom to see what the question is. I sure don't know much about > transformer design maths, but "drawing too much current", especially with > no load, sounds like core saturation to me. One thing to do is put a > 'scope on it and watch for the voltage to clip or current to spike partwa= y > though the pulse cycle. The other thing you can do is reduce the duty > cycle until it comes out of saturation. If that's what the problem reall= y > is. Winding too few turns on the primary will also reduce the capacity o= f > the transformer--that's why you have to use a lot more turns even though > you could keep the turns ratio correct using fewer turns. > > > On Tue, Mar 19, 2013 at 1:55 AM, Manu Abraham wro= te: > >> Hi Vasile, >> >> It took some time for me to find the issue. Faradays law has been verifi= ed >> and >> it looks applicable. >> >> As it looks like, the issue seems to be Rac/Rdc related, where Rac is >> optimal but Rdc isn't. ie Rac/Rdc < 1. >> >> I have another question, but will address it in another thread/post. >> >> Best Regards, >> Manu >> >> >> On Sat, Mar 2, 2013 at 12:28 AM, Vasile Surducan >> wrote: >> > Hi Manu, >> > >> > Your email is far too long and maybe a few people will read it to the >> end. >> > As far I've understood you have less wire as it should be on the prima= ry >> coil. >> > Have you verify your coil design using the Faraday law? >> > >> > Vasile >> > >> > On Tue, Feb 26, 2013 at 11:34 AM, Manu Abraham >> wrote: >> >> Hi, >> >> >> >> I worked out a transformer based on these calculations: >> >> >> >> http://www.icecomponents.com/images/document/PDF/SMPS.pdf >> >> >> >> The transformer is driven by 8xIRF540's on each primary >> >> The IRF540's are driven using a IRF2110 MOSFET gate driver >> >> A dsPIC supplies 98kHz square wave at 22, 37, 42% duty cycles >> >> for transformer testing purposes. The duty cycles and frequency >> >> have been verified with a logic sniffer. >> >> >> >> (1) Core Selection >> >> EPCOS E55/28/21, material=3DN87 >> >> http://www.epcos.com/inf/80/db/fer_07/e_55_28_21.pdf >> >> Frequency=3D100kHz >> >> Core can handle upto 1000W at 100kHz >> >> Output power required =3D ~200-300W >> >> B=3D100mT >> >> Ae=3D354mm^2 >> >> >> >> (2) Transformer calculations >> >> >> >> Duty Cycle max (Dmax)=3D0.5 >> >> Minimum DC Voltage (Vdc min)=3D9V >> >> >> >> >> >> (3) Estimating E-T value >> >> E-T=3D1*10^6 * (Dmax)*(Vdc min)/f >> >> >> >> =3D 1*10^6 * (0.5)*(9)/100,000 >> >> =3D 450 >> >> >> >> (4) Primary Turns Count. >> >> Determining minimum primary turns >> >> Np=3DE-T*10^2/(B*Ae) >> >> >> >> =3D 450*100/(100*354) >> >> =3D 1.2711 turns ~2 turns >> >> >> >> (5) Turns Ratio >> >> For a push pull converter >> >> >> >> Ns/Np =3D Vs/(Vdc min * Dmax) >> >> >> >> Ns =3D Vs * Np/(Vdc min * Dmax) >> >> >> >> =3D 250 * 1.2711/(9 * 0.5) >> >> =3D 70.62 ~71 >> >> >> >> >> >> Ns/Np =3D 71/1.2711 =3D 55.857 >> >> Ns =3D 55.87 * Np >> >> Np =3D 55.87 * 2 =3D 112 turns ~ 120 turns >> >> >> >> (6) Wire Size >> >> Assume 400 circular mils per 1 ampere >> >> Total circular mils=3D400*I >> >> for AWG30=3D79.2 circular mils >> >> for 250W, 250W/250V=3D1A >> >> >> >> For secondary at 1A, >> >> 400/79.2=3D5.05=3D6strands of AWG30 >> >> >> >> For the primary; with 250W output >> >> Core loss=3D3.8W ~4W >> >> with efffeciency of 87% >> >> >> >> Primary Power =3D 254*100/87 =3D 291.95=3D292W >> >> >> >> Since 2 primary coils, each primary powers 292 / 2 =3D 146 >> >> Current through each primary =3D 146/12 =3D 12.1A ~12A >> >> for 12A at primary, for 12A, 12 * 6 strands of AWG30 >> >> >> >> At hand, I am having some litz wire of 45SWG, 90 strands in all. >> >> So, for the Primary, I used 6 strands of the litz with 2 turns and >> >> for the secondary 120 turns with 2 strands of the said litz wire. >> >> >> >> The results what I get is as follows.. >> >> >> >> With Ferrite Core (E55/28/21) >> >> >> >> Test: (1) >> >> >> >> No Load, Inverter OFF, Battery Voltage: 14.5V >> >> Frequency=3D98kHz >> >> With 22% Duty Cycle, Resistive Lamp Load (230V): >> >> >> >> Load (Watts) Current Battery Voltage >> >> @230V DC (A) DC (V) >> >> -------------------------------------------- >> >> No Load 8.6A 13.7V >> >> 40W 8.4A 13.6V >> >> 60W 8.7A 13.6V >> >> 100W 9.6A 13.6V >> >> >> >> Test: (2) >> >> >> >> Frequency=3D98kHz >> >> With 37% Duty Cycle, Resistive Lamp Load: >> >> >> >> Load (Watts) Current Battery Voltage >> >> @230V DC (A) DC (V) >> >> -------------------------------------------- >> >> No Load XXX XXX >> >> 40W 14.6A 13.6V >> >> 60W 17.0A 13.6V >> >> 100W 18.5A 13.4V >> >> >> >> >> >> At hand, I have a Yellow Powdered Toroid Ring with the following >> >> dimensions. I don't have much details about this toroid, except >> >> that the shop from where I bought, the shop keeper told me that >> >> it was a powdered iron core from China, which could handle about >> >> 200kHz. >> >> >> >> I did a quick transformer with 2 turns of the same litz (6 strands) >> >> for the primary and 4 turns of litz (4 strands) for the secondary. >> >> The transformer thus has a ratio of 1:2, which should justify a 1:1 >> >> voltage transfer for a square wave 25% duty cycle. >> >> >> >> Test : (3) >> >> >> >> With powdered Iron Core (Yellow Toroid) >> >> OD=3D58mm ID=3D35mm >> >> >> >> No Load, Inverter OFF, Battery Voltage: 13V >> >> Frequency=3D98kHz >> >> With 42% Duty cycle, Resistive Lamp Load (12V): >> >> >> >> Load (Watts) Current Battery Voltage >> >> @12V DC (A) DC (V) >> >> -------------------------------------------- >> >> No Load 5.0A 12.5V >> >> 35W 7.5A 12.0V >> >> 70W 8.5A 11.9V >> >> >> >> >> >> The results are quite different from what I was expecting.. >> >> (a) Even at no-load the converter seems to be drawing too much curren= t. >> >> (b) The converter fails to deliver the proper output load >> >> (i) As in test 3, with 35W the lamp appears to be in it's prope= r >> >> brightness >> >> with 70W, the lamp lost a lot of the brightness. >> >> >> >> These 2 issues are visible in Tests 1 and 2 as well as can be seen. >> >> Any idea, what could be wrong with my transformer calculations/design= ? >> >> with the 40W (230V) lamp, though not at full brightness, it glows qui= te >> well. >> >> At 60W the brightness is even lesser, at 100W the filament is a bit >> more than >> >> yellow. It looks to me that the transformer is unable to deliver the >> required >> >> power, while it is drawing more than the required current. Is the >> transformer >> >> getting into saturation ? >> >> >> >> If so, what's wrong in my calculations/design ? >> >> >> >> What can I do to fix my calculations and hence my design ? >> >> >> >> Any help ? >> >> >> >> >> >> >> >> Thanks, >> >> >> >> Manu >> >> -- >> >> http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive >> >> View/change your membership options at >> >> http://mailman.mit.edu/mailman/listinfo/piclist >> > -- >> > http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive >> > View/change your membership options at >> > http://mailman.mit.edu/mailman/listinfo/piclist >> -- >> http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive >> View/change your membership options at >> http://mailman.mit.edu/mailman/listinfo/piclist >> > -- > http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive > View/change your membership options at > http://mailman.mit.edu/mailman/listinfo/piclist -- http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist .