Hello friends,
After several tests, I've done a SMPS push-pull with 2 x IRF540. Input 16V and output 21V (just for testing). Measured efficiency was somewhere around 85...90% which I guess is quite ok.
For trafo I took a ATX one (E+I), half-bridge classic design, I saw from originally windings has ~1.666 wd/volt at 30KHz and I scaled to my needs. Everything is working nice except at ~80W delivered to my load, trafo is getting very hot, measured ~88...91 centigrade.
I know trafo core withstand @200W since that was the power of ATX PSU or whatever, at least 100W without over 40-50 deg. I measured efficiency quite well with two 1% 30W 0.01 Ohm precision resistors and DMM for voltages.
Can someone explain and suggest what should I do ?
Thanks very much in advance,
After several tests, I've done a SMPS push-pull with 2 x IRF540. Input 16V and output 21V (just for testing). Measured efficiency was somewhere around 85...90% which I guess is quite ok.
For trafo I took a ATX one (E+I), half-bridge classic design, I saw from originally windings has ~1.666 wd/volt at 30KHz and I scaled to my needs. Everything is working nice except at ~80W delivered to my load, trafo is getting very hot, measured ~88...91 centigrade.
I know trafo core withstand @200W since that was the power of ATX PSU or whatever, at least 100W without over 40-50 deg. I measured efficiency quite well with two 1% 30W 0.01 Ohm precision resistors and DMM for voltages.
Can someone explain and suggest what should I do ?
Thanks very much in advance,
Excess eddy currents from incorrect core mix?
Are the cores held together without a gap (even in PP, a slight imbalance in the DC or waveform can cause headaches)?
Did you make "litz" wire by twisting several fine wires together or use copper foil instead of plain wire as if it were a 60Hz tranformer?
Those were the ones I ran into when playing with SMPS.
Cheers!
Are the cores held together without a gap (even in PP, a slight imbalance in the DC or waveform can cause headaches)?
Did you make "litz" wire by twisting several fine wires together or use copper foil instead of plain wire as if it were a 60Hz tranformer?
Those were the ones I ran into when playing with SMPS.
Cheers!
Perhaps... I used this arrangement (first layer -> last layer):
<primary>
<primary>
<secondary>
<secondary>
and as far as I read, I was wrong. I should use:
<primary>
<secondary>
<secondary>
<primary>
but I glued everithing with epoxy
Initially I don't use any gap since originally PSU trafo haven't any such. I added some gap (paper scotch) for testing but no significant improvement. More likely worst, as gap increase -> efficiency decerease dramatically.
No. I used single 0.9mm wire for all windings. I judged if originally 5V / 20A winding had 3 * 1mm scheme and 12V / 8A had 1 * 1mm, for 5-6A which I need will be enough 1 * 0.9mm (same as originally 12V winding)
I couldn't find any practically (simplified) design of trafo for push-pull so any link (thread) will be very appreciate. Maybe my driver isn't properly designed, I use some PNP (BC327) as buffer between MOSFET gate and TL494 and from scope, looks ok but I read a lot of posts about driving.
I stabilised TL494's Vdd to 9V so in theory, MOSFET's gate couldn't be over 9V so perhaps need more to full ON ?! I can attach some schematic if anyone need.
Best regards,
The incorrect interleave alone will make it heat up and the plain wire will have a heckuva lot of losses.
New transformer time before doing anything with the circuit.
Cheers!
New transformer time before doing anything with the circuit.
Cheers!
Read up on skin effect on wires at high frequency. At higher frequencies ac currents travel at the surface of the wire and not in the middle. For example at 100KHz switching frequency only a 0.4mm wire would fully utilized. The power supply which you are using for a reference for your design may well have been running at a slow frequency than the one you have designed. Which switching frequncy are you using?
It is also worth bearing in mind higher switching frequencies allow you to use smaller transformers while lower switching frequencies require a bigger transformer.
PCB Via Current | PCB Trace Width | Differential Pair Calculator | PCB Impedance
some usefull calculators, also has skin depth under conductors, worth downloading as its free.
Copper foil wound transformers are good but may require some screening as quite a lot of capacitive coupling. If you look at a lot of up market SMPS's they use planar transformers (basicly coils are designed on a PCB), a great way of doing it as you can control such things easier and are cheeper to produce (in volume).
some usefull calculators, also has skin depth under conductors, worth downloading as its free.
Copper foil wound transformers are good but may require some screening as quite a lot of capacitive coupling. If you look at a lot of up market SMPS's they use planar transformers (basicly coils are designed on a PCB), a great way of doing it as you can control such things easier and are cheeper to produce (in volume).
Are you using cooler? PC SMPS are designed for forced cooling! Also check PWM balance between arms of the push pull, this topology is very unstable and can saturate the core if unbalance is allowed.
- Status
- Not open for further replies.