1kw or more new SMPS project, help with parts

[undrtkr]

hi all

i've been colecting some parts to make a SMPS for my UCD amplifiers, and on the past days i got the crucial part for start the project, some E cores from big old smps boards that a friend gave me

i'm thinking on half bridge with +-80v and 10~15 amps output to supply the UCDs, and reading the forum i liked so much Ludo's topic and his smps, maybe it will be my starting point

i already have almost all the parts for making the smps, just need to buy output capacitors, wire for transformer, another small parts and make the PCB, i'll list the parts that i have and possible doubt about each one:

IR2110 and KA3525A think no problem here, is this KA3525A the same SG3525?

GBPC2506 rectifier for AC input, think no problem here

2x 1200uf/200v and 4x 820uf/200v caps for input, can i use all these capacitors together?

IRFP460A, you can see on the pictures there are a small ferrites on source legs of IRFPs, how about that?

BYV42E-200 diodes for output, the datasheet says 30A/200V, 15A per leg ultrafast rugged diodes, even the pinout don't let me use the 2 diodes of package on negative rail, can i still use it like a 15A diode at least?

the core is E55/28/21 from N67 material, 3,54cm2 of Ae, by the size i can imagine this core can provide a lot of watts, but i don't know about this material, anyone have faced N67 before?


my final questions are:

how many watts can i get with this core and IRFP460A?

can i build Ludo's smps using these parts with no problem? (i know that i have to recalculate the turns for my core)

anyone knows an alternative for MUR460? i have some MURHF860 8A/600v it's like 2x MUR460 on same package, so can i use MURHF860 instead MUR460?

that's all for now, i'll be attaching some pictures and datasheets

all kind of help and comentary are welcome
thank you all

 

[wally7856]

“is this KA3525A the same SG3525?”

Looks like it to me.

IR2110 is good driver also.

“GBPC2506 rectifier for AC input, think no problem here”

At 25A and 600V looks good to me to.

“2x 1200uf/200v and 4x 820uf/200v caps for input, can i use all these capacitors together?”

Yes.

“IRFP460A, you can see on the pictures there are a small ferrites on source legs of IRFPs, how about that?”

How about that what?

“BYV42E-200 diodes, even the pinout don't let me use the 2 diodes of package on negative rail, can i still use it like a 15A diode at least?”

http://www.nxp.com/documents/data_sheet/BYV42E_SERIES.pdf

Each diode is good for 30A at 200vdc.

“the core is E55/28/21 from N67 material, 3,54cm2 of Ae, by the size i can imagine this core can provide a lot of watts, but i don't know about this material, anyone have faced N67 before?”

The material for that core is N87 or N27. To know what one wrap 20 turns of wire across the middle leg of the E-cores and hold them together tightly and take an inductance measurement. Let me know what number you get and I or you can calculate the Al value to find out what material it is. Try not to let your hands go near the winding’s as you measure it, that really messes up the reading.

“how many watts can i get with this core and IRFP460A?”

Output power of the core is proportional to the switching frequency. And you did not list that.

The IRFP460A can give 13A at max junction temperature. Fig 70% efficient SMPS that leaves you with 10A. With a half bridge you have half the voltage so take your low line voltage as worst case then subtract some ripple voltage and multiply x 10A. Say 155vdc - 20volts ripple = 135vdc x 10A = 1350W. Assuming i am not missing any other inefficiency’s.

“can i build Ludo's smps using these parts with no problem? (i know that i have to recalculate the turns for my core)”

If you would have posted his schematic or given a link i could have looked.

“anyone knows an alternative for MUR460? i have some MURHF860 8A/600v it's like 2x MUR460 on same package, so can i use MURHF860 instead MUR460?

I do not think the MURHF860 have enough current capability to work on the output, each diode is only rated for 4 amps average current. If you run them in parallel you will get 8 amps at 80vdc giving you 640watts and you need 500. But this is a non optimized design and those diodes are barely enough to meet 1kw spec (2 x 500W).

Microsim has a lot of experience, maybe he will comment on this.

 

[undrtkr]

thank you wally7856 by the soon reply

and sorry maybe my english not good enough, but i'll try to be more clear

i refer to this schematic

refering to schematic you can see 2x MUR460 diodes between Drain and Source of IRFPs, can i replace these diodes by the MURHF860? the BYV42E i want to use at output

about the IRFPs refer to this picture, these IRFP460A that i got have a small ferrite on source legs, how this can influency the circuit, that's good or not?

about the core, it isn't the same core of the datasheet that i've uploaded, the size is exactly the same, but the material is N67 and it's writed on the side of E, you can see it on this picture

i think 140khz on 3525 (meaning 70khz at trafo) a good frequency to start

about the test to find the Al of the core, can i roll the 20 turns of wire on bobin, and then mount the E pieces inside the bobin to measure without touching the cores? and any specific AWG wire for this test?

i'll try to do the test of cores tonight and i will post the results soon as possible

i want to post all the process in the making of this SMPS here, any comentary sugestion are welcome

thank's for the help

 

[undrtkr]

made the core test for Al

got 2087mH for 20 turns of wire, need help with calculations

i'm getting Al = 5197500 is this right?

thank's

 

[wally7856]

“refering to schematic you can see 2x MUR460 diodes between Drain and Source of IRFPs, can i replace these diodes by the MURHF860? the BYV42E i want to use at output”

Yes you could, but personally i do not think you should put those diodes there or their RC snubber’s. The IRFP460A already have internal body diodes.

That schematic is hard to follow. This one is looks very similar but is much easer to read, go down to post 60. I think it would be easier for all of us if you used this one.

http://www.diysmps.com/forums/showthread.php?250-IR2153-smps-with-short-circuit-protection./page6

“about the IRFPs refer to this picture, these IRFP460A that i got have a small ferrite on source legs, how this can influency the circuit, that's good or not?”

The ferrite beads were put on there to stop ringing. If you are near the switching frequency they were used for they will help you. If you are lower then they will not do much. If you are way to high they could possible clip your switching waveform but that is a small possibility.

“about the core, it isn't the same core of the datasheet that i've uploaded, the size is exactly the same, but the material is N67 and it's writed on the side of E, you can see it on this picture”

From your picture i can not verify 2 dimensions. The height of the core should be 21mm and the length 27.8mm can you check those for me. Many times the only difference in transformers are one of those 2 numbers. I have found other epcos transformers with N67 material but slightly different dimensions.

“i think 140khz on 3525 (meaning 70khz at trafo) a good frequency to start.”

Well do not paint yourself in to a corner. Before you set the frequency you have to know what your transformer is, them make some trial calculations and see what power is available at different frequency’s, and make some trial calculations for wire size and number of turns.

“about the test to find the Al of the core, can i roll the 20 turns of wire on bobin, and then mount the E pieces inside the bobin to measure without touching the cores? and any specific AWG wire for this test?”

Tough search but i found the curves for N67 material. As you said your transformer is stamped with N67 so we will belive it is N67. But as i asked above please verify the 2 dimensions for me so we can be certain of the Ae value. Since we have identified the core as N67 for sure there is no need to measure the Al value.

Also measure the inside width of the bobbin so we know how much winding area we have.

N67 material specs here.

http://www.thierry-lequeu.fr/data/SIFERRIT.pdf

Starting on page 64 of 78 are the curves for N67 material.

“i want to post all the process in the making of this SMPS here, any comentary sugestion are welcome”

Good, if everyone watches for mistakes and helps out you should get a good SMPS and all can learn.

 

[wally7856]

made the core test for Al

got 2087mH for 20 turns of wire, need help with calculations

i'm getting Al = 5197500 is this right?

thank's

You missed a decimal point. your meter says 2.079mH not 2079mH. Then the answer is Al = 5,197

The Al value is specific to core size. We can not verify this number unless we find an old catalog with this transformer and N67. This material is discontinued.

But notice the calculation is not to far from N87 = 6,400Al and N27 = 5,800Al, from the transformer spec sheet you posted so your numbers sound reasonable.

 

[wally7856]

WOW, i am hot on Google tonight.

http://www.farnell.com/datasheets/49690.pdf
E 55/28/21 with E67 material.

It says N67 has an Al value of 6400, so either your measurement is a little off, or someone is trying to fool us. It looks like you have 3 pieces of E-core do they all say N67.

 

[undrtkr]

thank's a lot for helping me out wally7856, very nice this info about N67 material, i don't found anything on my researches

so i can confirm you the size of core, length = 27,8mm and height = 21mm, also width = 55mm
have attached some pictures, to explain better with a paquimeter

the internal width of bobbin is 33,8mm

i'm looking the schematic you linked, and it's look nice for me, with protection and regulation will be a nice smps for my UCDs, then i'll use this schematic for now

i'll try to make some calculations with this information, and i'll be posting here to you people confirm if it all ok, i'm using MicrosiM PDF guide and excel file for making the calculations

any advices?

thank you

 

[wally7856]

“so i can confirm you the size of core, length = 27,8mm and height = 21mm, also width = 55mm
have attached some pictures, to explain better with a paquimeter”

Good work. Now we know for certain what we have.

“the internal width of bobbin is 33,8mm”




“i'm looking the schematic you linked, and it's look nice for me, with protection and regulation will be a nice smps for my UCDs, then i'll use this schematic for now”

I think that was a smart move.

“i'll try to make some calculations with this information, and i'll be posting here to you people confirm if it all ok, i'm using MicrosiM PDF guide and excel file for making the calculations

any advices?”

First the power of this transformer is.
6,832W at 100khz
4,782W at 70khz
3,416W at 50khz

Second you must decide what safety margin to run on your bobbin to prevent arc over from primary to secondary. This will subtract winding area. When you have your final winding width then play with frequency and wire diameter to find winding’s that will fill one full width. Do primary then secondary’s then primary.

 

[undrtkr]

last night i was making some calculations using the information that wally7856 have found about my cores

i'm using MicrosiM excel file, and ExcellentIT calculator to point me, they gave me different results but think it's ok

for the excel file i used:

Vin = 135 Vdc (worst case)
Bmax = 1200 (based on the graphs from page 96 of PDF that wally7856 posted and MicrosiM pdf guide)
Frequency = 75khz (my starting point)
Ae = 3.54cm2 (found on datasheet of E55/28/21)
Vout = 80+80 (is this value Vac at output of windings? should i have to add 1 or 2 turns for regulation margin and diode drop?)

the results:
Npri = 10.593 turns (x2 for 2 sections of primary)
Nsec = 6.519 turns (x2 for 80+80)

so i have more of these transformers assembled, then yesterday i was disassembled one and take note about the original windings, i'll attach some pictures to you know and a txt with more details

refering to the picture and txt, the primary is exactly 10 turns, this will give me a good point to me, but this transformer come from an old smps that has PFC, so the input voltage at transformer don't change and should be around 150Vdc (correct me if i'm wrong)


talking about he circuit

the circuit that i'm looking have an error on output, but searching the forum i found a corrected version:

http://www.diysmps.com/forums/showt...ircuit-protection.&p=5296&viewfull=1#post5296

can this circuit provide me more than 1kw? my core and fets can do 1350w in the worst case following wally's calculation, but i mean about the entire circuit is that ok for 1kw or more?

think it need better dimensioning on input and output capapacitors and output inductor for more than 1kw, how about that?

with large capacitance on board is necessary a soft start circuit with relay or the SG3525A soft start can do the job?


i have decided to drawn my own pcb, so i got some layout advices like putting the driver near fets and avoiding large tracks, i'm looking for a lot of layouts that inspire me, soon i'll be posting here

i'm looking for ideas, advices

thank you

 

[wally7856]

undrtkr, very good work documenting this transformer. I am studying it now. A few questions for you.

You had 2 winding’s that were 1 turn of cooper foil 0.1mm thick 24mm wide. What size wire was connected to these winding’s.

Can you check to see what the winding’s connected to on your PCB.

RFI Faraday screens are connected to the high voltage input side circuitry of the transformer, usually through a capacitor. Or directly to the DC output circuit common line, and is a low current connection.

Safety screens are connected to the ground plane or chassis and is a high current connection.

I can not figure out what this is. “01 turn common wire” Can you tell me anything more about it. What size wire and were does it go to.

And last do you know what switching frequency this transformer was running at. You can look at the old controller chip and see what RC network was used for the oscillator and look it up on the spec sheet.

 

[undrtkr]

"undrtkr, very good work documenting this transformer. I am studying it now. A few questions for you."

thank you, no problem about that, ask me what you want

i still have one board complete yet including PFC circuit, thermal control with lm35, it's a industrial design i think, this smps was used to power VHF radio of many watts for local services, it's designed in USA... these boards have failure in consequence of thunderbolts and then replaced

i learn so much with you, please share with me your theory

"You had 2 winding’s that were 1 turn of cooper foil 0.1mm thick 24mm wide. What size wire was connected to these winding’s."

this wire have 0.55mm diameter, 2 foils connected at same point on bobbin and to pcb ground

"Can you check to see what the winding’s connected to on your PCB."

i'll try to make a drawn on a picture of the board

"I can not figure out what this is. “01 turn common wire” Can you tell me anything more about it. What size wire and were does it go to."

this wire diameter is 0.95mm, look at the picture i've attached, looking better i se the wire have something in the middle, disassembled and got a 10Amp fuse, this winding is connected to a point called charger in mainboard, and have a snubber circuit with a 4.7ohm 1/4w resistor and 102 small yellow axial capacitor, seens interesting?

"And last do you know what switching frequency this transformer was running at. You can look at the old controller chip and see what RC network was used for the oscillator and look it up on the spec sheet."

i will do this now, hehe

thank you wally, very much pleased to share this info with you

 

[undrtkr]

i've discovered the frequency of the oscillator, the controller chip is the UC2845N i've have attached the datasheet, at the Rt/Ct pin have a 2.4k precision resistor and a 472 axial yellow capacitor

on datasheet page 4, figure 1, looking at the graph we will got 150khz for 2.4k and 4.7nF
correct if i'm wrong

if you want pics of the small control boards, PFC, and more i can share with you, all pcbs are dual layer

can you calculate the original power of this SMPS wally7856 ???

 

[wally7856]

I am trying to make the math come out for this transformer.

Was this a uninterruptable power supply with a battery backup. If so what battery voltage.

What was the fuse rating of the unit.

What was the input voltage.

 

[undrtkr]

"I am trying to make the math come out for this transformer."

very nice

"Was this a uninterruptable power supply with a battery backup. If so what battery voltage."

can't be sure about that, but if i remember the batteries are 12v 150aH each, a lot of
the board don't have any 12vdc input that i can see, operates only from Vac,
but exists another battery module, maybe a inverter don't know

"What was the fuse rating of the unit."

10A 250V

"What was the input voltage."

can't confirm you for sure, but this smps have PFC based on UC2854N, maybe working from 90~240vac
here was working on 220vac

the outputs are +26 +15 +5 -5 outputs, each output have a small control board, with a BYV42E diode and a IRF640
think these outputs are made from the bigger one coming of the trafo

 

[wally7856]

OK, i am getting it figured out, but the one turn winding with a fuse i still do not understand. Are you saying that one end of this wire connected to the PCB. But the other end went around the transformer 1 time and did not connect to anything?

 

[undrtkr]

"OK, i am getting it figured out, but the one turn winding with a fuse i still do not understand. Are you saying that one end of this wire connected to the PCB. But the other end went around the transformer 1 time and did not connect to anything?"

looks strange for me, this side of winding go to a point called VAUX in pcb that are the same 26Vdc output, still looking on 26vdc output a 8A fuse and then a diode feed the +15 +5 modules

the caps of input are 820uf 400v rated
the board only works from 190~240vac, have confirmed that on a small back part of power connector

edit

think it's not a half bridge, because i can't see the floating capacitor

the primary windings are directly connected to fets

 

[wally7856]

Are the 820uF 400VDC input caps in series or parallel. And are there any other big electrolytic caps near them.

Are there only 2 power switching transistors.

 

[wally7856]

Do you think that wire with a 10 amp fuse could have been soldered to the middle of one of the winding’s or copper foil as a center tap.

 

[undrtkr]

"Are the 820uF 400VDC input caps in series or parallel. And are there any other big electrolytic caps near them."

the capacitors are in paralell, before the caps have a big coil in a E42 core gapped
there are 2 IRFP450 in parallel driving this coil, is it the PFC right?

"Are there only 2 power switching transistors"

yes 2x IRFP460 at the transformer primarys

i've attached a picture that explain a little better, now i can understand better how it works with the batteries

think it's only a big 26Vdc supply, that charges the batteries with help of 1turn winding to provide a little more voltage to charge, and a 10amp fuse to protect
the other outputs works from this 26vdc supply, when AC off the small modules works from batteries