Guys, we are here to discuss and learn about this, not bash at each other.
Microsim444: if you have some helpful input, that would be great. I do agree with needing PFC for 3kw. Please try and be more polite and Luka, please don't get caught up in it. If someone want's to be rude, don't encourage them.
I am learning about this stuff along with you guys and I would hope we can all help each other. This is not a Knowledge competition and I certainly don't care for "gloating" The pictures I saw of Luka's supply look very nice and professional. I am sure he has tested it to do what he set out to accomplish and it worked for that purpose or he would be re-designing it.
OKAY, I'll get off my soap box now. What were we talking about.... oh yeah, massive ferrite cores and getting ridiculous amounts of power through them.
PQ 5050 H7C1 material has been reliable for 1500 watts in several supplies I built. 2x 2 turns foil for primary , 18 turns X2 secondary litz wire to get +- 65 volts from 12 volt boat system.
Microsim444: if you have some helpful input, that would be great. I do agree with needing PFC for 3kw. Please try and be more polite and Luka, please don't get caught up in it. If someone want's to be rude, don't encourage them.
I am learning about this stuff along with you guys and I would hope we can all help each other. This is not a Knowledge competition and I certainly don't care for "gloating" The pictures I saw of Luka's supply look very nice and professional. I am sure he has tested it to do what he set out to accomplish and it worked for that purpose or he would be re-designing it.
OKAY, I'll get off my soap box now. What were we talking about.... oh yeah, massive ferrite cores and getting ridiculous amounts of power through them.
PQ 5050 H7C1 material has been reliable for 1500 watts in several supplies I built. 2x 2 turns foil for primary , 18 turns X2 secondary litz wire to get +- 65 volts from 12 volt boat system.
Attachments
Yes you are right...anyway
For crazy power, you can always go on the path car audio uses, to use several smaller cores (but still with pretty big cross section of them) and series the secondarys. that way you would get what you would not have with the biggest standard core, apart from just using ferrite square rods, and still use the same or less space. I must say I haven't seen this yet, I guess there could be some problems... on other hand, go up with freq., @150k+ you can already get ridiculous amount of power from small cores, let alone big ones
but as far as I know, best to build several supplys that are not that powerful, by themself, they have say 1/4 the voltage of needed one, and connect 4 like this in series.
Does picture work for anybody?it doesn't zoom for me
For crazy power, you can always go on the path car audio uses, to use several smaller cores (but still with pretty big cross section of them) and series the secondarys. that way you would get what you would not have with the biggest standard core, apart from just using ferrite square rods, and still use the same or less space. I must say I haven't seen this yet, I guess there could be some problems... on other hand, go up with freq., @150k+ you can already get ridiculous amount of power from small cores, let alone big ones
but as far as I know, best to build several supplys that are not that powerful, by themself, they have say 1/4 the voltage of needed one, and connect 4 like this in series.
Does picture work for anybody?it doesn't zoom for me
Last edited:
Good
RICK,
I will show you soon, u can go to diysmps, there is a snapshot of my 1KW design, still under development.
Also, You dont know any thing, because you cant prove the 3KW draw from that supply, or at least tell the CORE used. dont talk if you cant show.
RICK,
I will show you soon, u can go to diysmps, there is a snapshot of my 1KW design, still under development.
Also, You dont know any thing, because you cant prove the 3KW draw from that supply, or at least tell the CORE used. dont talk if you cant show.
Luka, the problem I ran into going to higher frequency was that there was not enough window area for the size of conductors I needed in the smaller cores. I agree, you could stack cores to double or more the Ae2 but then you run into a problem of not having a ready made bobbin. This is not an issue with a toroid of course. I will try and re-post the picture. I see that it will not enlarge.
Boat amp.
I will try this again. sorry for the previous small image. This was built on the heatsink of an old Zerox word processor. The fan is 5 1/4 inch diameter to give an idea of scale. The amp was built into a closet on a 34 foot Cigarette boat. We turned the bathroom into a subwoofer. I built this in 1992. Still running from my understanding.
I will try this again. sorry for the previous small image. This was built on the heatsink of an old Zerox word processor. The fan is 5 1/4 inch diameter to give an idea of scale. The amp was built into a closet on a 34 foot Cigarette boat. We turned the bathroom into a subwoofer. I built this in 1992. Still running from my understanding.
Attachments
Last edited:
2x 2200uf, etd59, no pfc since i live in romania, the grid here can hold 20 amps with no problem. I don't have photos since i built it some years ago. I am making another luka design in the neer future (just ordered components), but only for 1 kw. After this maby if i have time i will demonstrate you that you can use that design in H bridge for 6 kw. Untill then... STOP BEEING A HATER AND PROOVE YOURSELF. Anyone can speek words
2x 2200uf, etd59, no pfc since i live in romania, the grid here can hold 20 amps with no problem. I don't have photos since i built it some years ago. I am making another luka design in the neer future (just ordered components), but only for 1 kw. After this maby if i have time i will demonstrate you that you can use that design in H bridge for 6 kw. Untill then... STOP BEEING A HATER AND PROOVE YOURSELF. Anyone can speek words
OK
Friend.
I cant show uncompleted work, but also its impossible to draw 3KW from
ETD59 with 2200UF, EE65 Cannot draw 3KW continuous, due temperature raise. Unless you have a good cooling technique.
I only can draw 1100W constant from my prototype. with 4x2200UF, otherwise waves goes into bad shape at that load, for PULSES its OK
I have to believe you, since you cant show any thing, with outstanding results 😛
FORGOT ONE THING, you will show me how to draw 6KW in HALF BRIDGE mode from LUKA Supply?
I think you will break all engineering rules in the world with that invention!
Friend.
I cant show uncompleted work, but also its impossible to draw 3KW from
ETD59 with 2200UF, EE65 Cannot draw 3KW continuous, due temperature raise. Unless you have a good cooling technique.
I only can draw 1100W constant from my prototype. with 4x2200UF, otherwise waves goes into bad shape at that load, for PULSES its OK
I have to believe you, since you cant show any thing, with outstanding results 😛
FORGOT ONE THING, you will show me how to draw 6KW in HALF BRIDGE mode from LUKA Supply?
I think you will break all engineering rules in the world with that invention!
Last edited:
3kW through an EE65 should be perfectly possible at 100kHz, with a good core material (N97 or similar), about 20% copper in the window, convection cooling and a Rac/Rdc of 1.5. Under the same conditions an ETD59 should be able to deliver even more.
What's the Rac/Rdc of your transformer? Did you neglect the proximity effect perhaps?
What's the Rac/Rdc of your transformer? Did you neglect the proximity effect perhaps?
Check out that beast:
Meanwell Power Supply Products
This is a 24V 125A (3kW) power supply. It runs in phase shift with current doubler.
The transformer is, believe me or not, a small E42 core...
The fan cooling is huge so that it makes possible to use a smaller core.
If you can accept an higher temperature rise of the transformer or if you use
strong forced air cooling you will be able to pass much more power from the transformer.
ETD59 or even better EE65 can pass much more than 3kW continuous if forced air cooled.
Meanwell Power Supply Products
This is a 24V 125A (3kW) power supply. It runs in phase shift with current doubler.
The transformer is, believe me or not, a small E42 core...
The fan cooling is huge so that it makes possible to use a smaller core.
If you can accept an higher temperature rise of the transformer or if you use
strong forced air cooling you will be able to pass much more power from the transformer.
ETD59 or even better EE65 can pass much more than 3kW continuous if forced air cooled.
Where did you get 500kHz from? The datasheet for the Meanwell says 100kHz. Also, 3kW from an ETD49 at 60kHz isn't as impossible as you'd think with forced air cooling, good windings and a low-loss core material.
Did you take proximity effect into account when you designed the windings of your transformer or not? If not, then you'll have high losses for sure.
mag: Do you know if the Meanwell runs at 100kHz transformer frequency or 50kHz? The datasheet says fosc = 100kHz and that can sometimes mean 50kHz transformer frequency, but for a E42 at 3kW, that sounds low, even with really good air cooling. 😱
Did you take proximity effect into account when you designed the windings of your transformer or not? If not, then you'll have high losses for sure.
mag: Do you know if the Meanwell runs at 100kHz transformer frequency or 50kHz? The datasheet says fosc = 100kHz and that can sometimes mean 50kHz transformer frequency, but for a E42 at 3kW, that sounds low, even with really good air cooling. 😱
Hi Megajocke,
the oscillator is running at 200kHz so the transformer is at 100kHz.
BTW it is not at E42, the current doubler inductances are wound on E42.
The transfomer is an ETD49.
I don't think that at those frequencies the core material is important. If you check a standard 3C90 core or a low loss 3C92 core they exhibit similar losses at 100kHz. It is just a matter of cooling.
The windings seems to be made of litz wire to reduce Rac/Rdc ratio.
@microsim:
take care about proximity effect. It is a killer effect at those power levels.
In my first transfomer prototype of a 24V/100A SMPS I was using for the secondary 5 turns of copper foil of 0.5mm thickness. The copper foil was og the same length of the coil former so that it was a 5 layer winding.
The transformer was at around 130°C with only 20A loading... After taking the proximity effect into account I calculated a current density in the secondary of around 150A/mm^2 instead of my 4A/mm^2 due only to skin effect. After having rewound the secondary with 6x litz wires of 240x0.1mm everything stays cool...
I suggest you to don't use more than 2 layer copper foil.
the oscillator is running at 200kHz so the transformer is at 100kHz.
BTW it is not at E42, the current doubler inductances are wound on E42.
The transfomer is an ETD49.
I don't think that at those frequencies the core material is important. If you check a standard 3C90 core or a low loss 3C92 core they exhibit similar losses at 100kHz. It is just a matter of cooling.
The windings seems to be made of litz wire to reduce Rac/Rdc ratio.
@microsim:
take care about proximity effect. It is a killer effect at those power levels.
In my first transfomer prototype of a 24V/100A SMPS I was using for the secondary 5 turns of copper foil of 0.5mm thickness. The copper foil was og the same length of the coil former so that it was a 5 layer winding.
The transformer was at around 130°C with only 20A loading... After taking the proximity effect into account I calculated a current density in the secondary of around 150A/mm^2 instead of my 4A/mm^2 due only to skin effect. After having rewound the secondary with 6x litz wires of 240x0.1mm everything stays cool...
I suggest you to don't use more than 2 layer copper foil.
Mag.
Thanks for the informations, My 4KW transformer was made at specialist Transformers manufacturer in the USA, and its Calculated to handle that power correctly, also its foil windings, carfully manufacturered to deliver the power I need, EE65 with 30C90 material.
I think per there calculations, it wont hit 60C at 2KW RMS Load. cant remember exactlly.
Thanks.
Fox
I wonder how high one dares push the core losses in an ETD core with forced air cooling... The point where the center legs meet doesn't have that good cooling.
I was thinking of core materials similar to this:
http://www.ferroxcube.com/prod/assets/3c96.pdf
100°C, 100kHz 100mT losses are half of 3C90, 40 instead of 80 kW/m³ (1,0W vs 1,9W for ETD49)
100°C, 100kHz 200mT losses are 35% lower than 3C90, 300 instead of 450 kW/m³ (7,2W vs 10,8W for ETD49)
for reference:
http://www.ferroxcube.com/prod/assets/3c90.pdf
I was thinking of core materials similar to this:
http://www.ferroxcube.com/prod/assets/3c96.pdf
100°C, 100kHz 100mT losses are half of 3C90, 40 instead of 80 kW/m³ (1,0W vs 1,9W for ETD49)
100°C, 100kHz 200mT losses are 35% lower than 3C90, 300 instead of 450 kW/m³ (7,2W vs 10,8W for ETD49)
for reference:
http://www.ferroxcube.com/prod/assets/3c90.pdf
4KW supply.
Just got this today, bought it off Ebay for $30.00! It is new, in the original box.
Output rated at: 27 volts, 145 amps! Phase shifted IGBT power stage, three W20nb50 IGBTs per leg. I will post more info as I have time to do more reverse engineering. There are two toroid cores stacked inside the windings. They are about 2 inches diameter and 7/8 inches tall each.
This has another part that is a power factor correction supply with an output of 400 volts. It's toroid is even smaller. I have removed the heat sinks to take the pictures. The whole thing was in a nice square box with forced air cooling.
Just got this today, bought it off Ebay for $30.00! It is new, in the original box.
Output rated at: 27 volts, 145 amps! Phase shifted IGBT power stage, three W20nb50 IGBTs per leg. I will post more info as I have time to do more reverse engineering. There are two toroid cores stacked inside the windings. They are about 2 inches diameter and 7/8 inches tall each.
This has another part that is a power factor correction supply with an output of 400 volts. It's toroid is even smaller. I have removed the heat sinks to take the pictures. The whole thing was in a nice square box with forced air cooling.
Attachments
How do they manage to get good enough primary-secondary insulation? Has it something to do with that white wire perhaps? It is unusal seeing toroidal transformers in offline SMPSs.
Megajocke, the primary is made of insulated wires, (the white ones) There are about twenty of them in parallel. The secondary is enameled wire. I was surprised to find the toroid also. It makes it a lot easier to rewind than an epoxied etd or ee core. Not sure what I am going to rewind it for yet. It would make a killer supply for a big amp. If I was still doing car audio, it would make a nice 300 amp 12 volt supply for testing.
i borrow this thread for my question, i really hope u guys and girls have a solution to my problem....i hope its okey.
i want to build a transformer but i cant find a suitable core for my calculated values....
here is some data:
n: 20/1
P: 2.5kW
f: 100kHz
I1rms ~ 7A
I2rms: ~ 115A
the thing is, i want to use a UU, UI core, and with U93 from ferroxcube i get a flux at 10mT .... im basically not using the core at all 😉
if i go down in size, the windings wont fit because i need atleast n1 to be 40 turns due to the large turnsratio of 20 to 1 ....
am i thinking wrong??, i think u experienced builders have a solution to this....
i want to build a transformer but i cant find a suitable core for my calculated values....
here is some data:
n: 20/1
P: 2.5kW
f: 100kHz
I1rms ~ 7A
I2rms: ~ 115A
the thing is, i want to use a UU, UI core, and with U93 from ferroxcube i get a flux at 10mT .... im basically not using the core at all 😉
if i go down in size, the windings wont fit because i need atleast n1 to be 40 turns due to the large turnsratio of 20 to 1 ....
am i thinking wrong??, i think u experienced builders have a solution to this....
- Status
- Not open for further replies.