Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

determining smps transformer type and primary turns.

Status
Not open for further replies.

Salvador12

Full Member level 4
Joined
Jul 16, 2014
Messages
201
Helped
3
Reputation
6
Reaction score
7
Trophy points
1,298
Activity points
3,862
Hi folks , please help me understand what kind of transformer i got and how many primary turns it needs, I was driving the transformer but after a while it gets hot and it should let more power through.
I will post what I have, the schematic of the smps im using the transformer in and the photos of the transformer itself , mayeb you can help me out telling me what type transformer this is and how many turns it would require on the pirmary.

all i know is this transformer came from an older philips tv set which had a smps power supply, im using the oriignal litz wire found on the transformer but i assume i had too many turns on the primary from athe previous smps i used it in , about 38 turns primary, maybe that caused core saturation and heating.
any help appreciated.

20151118_150659.jpg20151118_150707.jpg20151118_150720.jpg20151118_150740.jpg20151118_150800.jpg50khz-4n25-sg3525-smps-ir2110-smps-900w-70v-power-supply.png
 

This looks like a voltage mode half bridge, (current mode would have required the output inductors to be coupled.)
You correctly have a 1uF DC blocking capacitor in series with the primary coil (& its blessed leakage inductance). The LC resonance of this {Lpri(leak):C(blocking)}must be at a frequency which is below the switching frequency of the SMPS. Preferably well below it.
No PFC stage means its likely illegal today for 900W.

To design a half bridge, …..your vin is half the DC bus voltage (155v for you). Your vout is 140v (total).
Pick a D of 0.7 say, at max load.
If this were a simple buck, then you would need vin to be 140V/0.7 = 200v…..you havent got 200v, you got 155v, so make the half bridge traxfmr have a step up of 200/155 = NS/NP. Make it a bit more than this to account for duty cycle loss due to txfmr leakage.

"3c85" marking means its probably a ferroxcube ferrite core.
 
Last edited by a moderator:

3C85 is a discontinued Philips/Ferroxcube ferrite quality, almost equal to 3C90. The core shape doesn't seem to correspond to a catalog product. But you can calculate the core parameters by putting in effective area and length.
 

Lets think about this scientifically. The Phillips TV was probably about 250 watts maximum. You now want to use the same primary winding to make a 900 watt power supply. It will not work.

If you have a dial caliper and you can measure the diameter of a single strand of the litz wire and then count how many strands there are we can calculate how much current the wire is good for.

Also if you look at this jpg and give us the dimension of your transformer and also the winding width of your bobbin we may come up with a way for you to make this work.

- - - Updated - - -

Does anyone think the 040 marking on the transformer could reference a gap?
 

Attachments

  • E core labled abcdefg One-Powerful-Decade-fig-2.jpg
    E core labled abcdefg One-Powerful-Decade-fig-2.jpg
    21.2 KB · Views: 125

thanks guys, you are very supportive , i didint got even close this kind of straight forward advice in other places were i seeked help.

well the tv power could have been about 250w but reconfiguring the litz wire making it thicker and so on i believe the core can push more., im not acutally aiming for 900w , 600 would be enough without the core getting too hot , as when i was driving it I already got close to 450/500 upon voltage measurment that is dropped across the load resistance.
ok here are the measurments, A-4.5 cm
B-2.3cm
C-3cm
D-1.6cm
E-3.2cm
F-1.4cm
M-0.9cm
L-almoust 7mm or 0.7cm

Well i'm not that good at maths but if you can help me with the calculation also you could show me the formulas and so.

thanks.
 

I would download the "ferroxcube ferrite databook 2013", and look through to see which one matches those dimensions.
I might be named slightly dfferent than above "ferroxcube ferrite databook 2013", but you should find it.
 

This is definitely a non-standard core.

C x F = Ac = Area of the center leg.
3CM x 1.4CM = Ac = 4.2CM^2

240 VAC input x 1.414 = 340 VDC

You have a half bridge circuit so the transformer will only see 1/2 the bus voltage. 340 VDC/ 2 = 170 VDC

Primary turns = Vpri x 10 8th/4 x f x B x Ac
170 vdc x 10 8th = 17,000,000,000
4 = constant
f = frequency in hertz, at the transformer.
B = Flux, Gauss, do not exceed 1600 Gauss, general rule.
Ac = Area of the center leg = 4.2CM^2


Primary turns = 17,000,000,000/ 4 x 50,000 HZ x 1600 Gauss x 4.2CM^2
= 12.65 turns, use 12 turns.

Wind 6 turns primary.
Your secondary.
wind 2nd 6 turns of primary.

The two primary windings must be connected in series and in proper phase.


If you want to work with the equation yourself the only parameters you should change are the frequency if you want something other than 50khz and the Gauss which is the hardest parameter. Best guess 1200 for low Gauss and 1600 for high Gauss. Lower Gauss generally means cooler transformer, but then you need more turns.
 
Last edited:

, hmm thanks folks, well what do you mean lower Gauss, if I understand correctly Gauss is the unit of magnetic field desnity and teslas are magnetic field strength even though maybe i'm messing someting up here.
but how could more turns give me less magnetic field desnity if more turns in a solenoid usually give more ampere turns which give a stronger magnetic field and for a given area it's density also then must increase?

also heres the interesting thing , there is another schematic of basically the same circuit just using different mosfets it shows using the same core but the turns ratios instead of 12 shown in the first one is 25 turns for primary?
I'm using the IRFP 460 instead of the ones mentioned in the schematic does this have anything to do with the turns count given that the core is the same and everything else is too.?
 

.1 Tesla = 1000 Gauss = Same thing different scale.

For a DC solenoid the amp-turns determine the strength of the magnet. Since DC has no inductance the current is only limited by the DC resistance of the wire.

For the transformer to work it must be AC for transformer action to work (a changing magnetic field). A wire wrapped around a core with AC current is an inductor. Inductors impede AC voltage. The more turns you have the more AC impedance you have and the lower the Gauss. With less turns the current goes up because of less impedance and your amp-turns actually go up. With one turn you would probably blow your circuit breaker.

Regarding your schematics, you are looking at internet crap. Every high school kid that crossed two wires together and made a spark publishes a schematic of a circuit that "works" meaning that it did not immediately burn up when he turned it on. The Ac for a ETD59 is around 3.545 if you want to calculate turns for that transformer. But from your story i would guess that one person forgot to divide the bus voltage by two for a half bridge and used 340 VDC in the transformer calculation.

At 1200 Gauss the turns for your transformer = 16.86 turns, so you would split this up to two 8 turn primaries.

The FET or igbt switch does not make any difference. Technically they have a small voltage drop if you want to subtract 1 volt from the calculation it will not make any difference. By the way, i used 240 VAC for the input, use what ever voltage your average power is.
 

how would you calculate secondary turns please show steps
 

well as for nebo, as long as the pirmary turns are the right count and the transformer doesnt saturate the pirmary secondary goes 1:1,
for example if you have +-100 volts going into the primary and your primary is 10 turns , if you will put the same 10 turns on secondary you will get +-100 out of the secondary. any other voltage scan be calculated approximately.
 

righ FlapJack, an ac inductor can only push as much current through as the inductance allows, so waht do you think about 15 turns should be good? for this particular transformer ? some folks have said it might be iron powder core but im not sure is that true or not.

anyway I hope my IRFP460 will be able to handle the curren. I will wrap those turns and see what happens , I wioll also use two parallel litz wires for each turn since the original litz wire is not that thick.as you can also see in the phoos.
 

Salvador12, you need an even amount of primary turns so that each 1/2 winding has the same amount of turns to balance things out.

Using two parallel litz wires is a good idea and this will help determine the number of primary turns. For best results you should fill the winding space across the transformer with wire, except for the 2 to 3mm margin's at the ends. Measure the diameter of the litz wire and see how close you can get to the desired primary turns with two parallel wires. You should be around 6 to 8 turns per 1/2 winding.

The transformer core is marked 3C85 i am sure this is Ferroxcube ferrite material. I am a little worried about the 040 marking. This could mean there is a gap ground on the center leg. Can you look closely at all three legs and see that they all touch and there is no gap when the two half's are put together, a gap the thickness of a piece of paper makes a difference.

Do you have a picture of the bobbin empty, it looks a little strange to me with the molded in margin's. Was this a high voltage transformer?

- - - Updated - - -

nebo123, The easiest way to calculate the secondary is a simple ratio.

For an unregulated power supply if you want 30vdc out and you have 170vdc in then 170vdc/30vdc = 5.666, with a total of 16 primary turns then 16/5.666 = 2.82 turns = 3 whole turns. With losses and diode drops you may want to add in one more turn.

If it was a regulated power supply and you wanted to account for a low voltage condition then use a lower primary voltage. Example 200vac low x 1.414 = 283vdc / 2 for half bridge = 141vdc on transformer. 141vdc/30vdc=4.7 16 primary turns/4.7 = 3.4 turns use 4 turns and maybe add in one more turn for losses and diode drops.
 

well I made 12 turns let's see how that works out. well the problem is the transformer core is glued together and since ferrite is brittle taking it apart will destroyd the core and the bobbin plastic is not transparent so there was no way for me to see whats inside , well i doubt it was a HV transformer since the isolation layers are not that thick between the primary and secondary for it to be a crt tube transformer.atleast to my mind, but who knows.

- - - Updated - - -

i actually hate semiconductors for their litle ability to whitstand anything beyond the smallest problem, heres the deal previously my mosfets were dead cold but under heavy loads the transformer got real hot and the voltage dropped, this time the transformer is cool 9atleast for that little time when i got to test it) and also the output was surprisingly stable, no matter the load the output stayed at 160v DC.not a volt less.
but heres the problem now, mosfe heatsink got really hot, mosfet driver IC blew up nicely and everything among the semicondcutors is shorted and dead.both IC's the small voltage regulator for the IC's and both mosfets are dead.

I guess the most likely explanation is that 12 turns is a bit too litle for this transformer it neds more turns.
 

Math does not lie, 12 total primary turns is OK, the core could have gotten to hot, but you would not draw to much power because of it.

How did you count your turns. You should have 6 turns of 2 parallel wires for each primary winding.

6 turns of 2 parallel wires.
Secondary.
6 turns of 2 parallel wires.

For a total of 12 primary turns after each half winding is wired in parallel.
 

well I made the primary all one winding , because otherwise I would't be able to wind the windings due to having to use more layers of insulation and there is not that much space.I did make 12 turns of two parallel litz wires.
as my stress test showed the wire diameter is ok because changing loads resulted in no voltage decrease.

But the mosfets did get very hot , not sure what failed first, whether the mosfet driver IC' or the mosfets themselves , well i guess the mosfets themselves because both IC's were blown short even their voltage regulator was blow short sop I guess the mosfets failed and their gates joined electrically with the drain.
so the IC's got a teste of what's it slike to have 325v DC running through them :D
 

One of us can not count. In your first post you have some pictures, 12 wires across the bobbin almost fills it up, 14 would probably fit, but 24 wires would definitely not fit.

If you made one winding of 12 turns of two parallel litz wires you would have to fit 24 wires across the bobbin there is no room for that many wires.
 

Looking sharp at the post #1 photos reveals at least two winding layers, the first layer covered by a blue tape.
 

no guys in the first post the photos were taken to reveal the core as I was asking about the core , in the photos is the old one wire 36 turn primary, those are old photos , i only recently made the new two parallel litz wire 12 turn primary, i made it winding it from one side filling up all space till the other then inserted a layer of insulation and made the rest of the necessary turns over from the side at which i stopped back to the first side.

so its like that , but still I havent heard what you think blew the mosfets and also made them really hot as they were hot before they ended up shorted.
 

Looking at the schematic the 4.7 ohm gate resistor is usually 15 to 18 ohms. But i do not think this would cause the fet to burn out.

The 7812 regulator should be 7815 to give you a little safety room for switching the fet off and on.

The feedback system is a cobbled up mess. It may be causing oscillation and heating the fets up.

An oscilloscope would be the weapon of choice to debug this problem. Look at gate wave forms and if you have a high enough voltage probe the switching wave forms.

Also the board layout is critical post pictures of both sides.
 

Status
Not open for further replies.

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top