I assured there is no cross conduction, the duty cycle is under 60%, the voltage rectified from ac line (approx 317V) sags very little since the filter caps are far bigger than they must (240uF, two 120uF aluminum caps)
Notice that this only happened using schottky diodes (unsnubbed), never happened using fast or ultrafast diodes, there is way too much ringing in the secondary side, and the more I snub the primary, the more ringing is on the secondary side, can't imagine why... today I'll build a transformer to run at lower freq, say 50 to 75KHz and see what happens, maybe for the components I'm using this freq range is more suitable
Thanks!!!
Notice that this only happened using schottky diodes (unsnubbed), never happened using fast or ultrafast diodes, there is way too much ringing in the secondary side, and the more I snub the primary, the more ringing is on the secondary side, can't imagine why... today I'll build a transformer to run at lower freq, say 50 to 75KHz and see what happens, maybe for the components I'm using this freq range is more suitable
Thanks!!!
sorry i misunderstood about it shorting with uf rectifiers. certainly schottkys must be snubbed since they with "oscillate' due to their junction capacitance.
the swings of the ringing can exceed their reverse breakdown voltage. soooo...... go back to the UF rectifiers or snub the schottys.
secondary or primary ringing will show up on the opposite side of the transformer from where the ringing is occurring. you may have to experiment with snubbing on both sides.
from what i have learned most hard switched inductive elements must have something, if it is properly designed it does not take much.
the swings of the ringing can exceed their reverse breakdown voltage. soooo...... go back to the UF rectifiers or snub the schottys.
secondary or primary ringing will show up on the opposite side of the transformer from where the ringing is occurring. you may have to experiment with snubbing on both sides.
from what i have learned most hard switched inductive elements must have something, if it is properly designed it does not take much.
The ringing you see is a artifact of the leakage inductance. The more leakage the more pronounced the ringing. If schottky's are used the need for a snubber is minimized on the secondary side since there isn't much Trr (reverse recovery) though there is capacitance wich will ring with the leakage inductance at it's resonant frequency.
Here's a little bit about transformer construction...
To reduce the leakage further, try this. Cut a small slit in the sidewalls of the bobbin on both sides of the bottom of the core. Then wind all the primary turns on a single layer against the bottom of the core. Pull the return wire through the slit on the end of the bobbin as we will return it at the very end. This also helps keep the leakage down. Add a layer of tape then add another layer of primary on top of the first layer. Next Bifilar wind the secondary so that the secondary wires fill the entire length of the bobbing bobbin. Leave the windings of this hanging out the slit in the bobbin. Add another layer of primary. then another secondary and primary. Each primary/secondary layering will halve the leakage. Now add a layer of tape and return all the primary wires. Now add 3 layers of tape and the secondary wires.
I hope that helps.
Here's a little bit about transformer construction...
To reduce the leakage further, try this. Cut a small slit in the sidewalls of the bobbin on both sides of the bottom of the core. Then wind all the primary turns on a single layer against the bottom of the core. Pull the return wire through the slit on the end of the bobbin as we will return it at the very end. This also helps keep the leakage down. Add a layer of tape then add another layer of primary on top of the first layer. Next Bifilar wind the secondary so that the secondary wires fill the entire length of the bobbing bobbin. Leave the windings of this hanging out the slit in the bobbin. Add another layer of primary. then another secondary and primary. Each primary/secondary layering will halve the leakage. Now add a layer of tape and return all the primary wires. Now add 3 layers of tape and the secondary wires.
I hope that helps.
I think that's the hell of a winding technique that will help a lot, since it's nothing else wrong on the supply, in fact, it works fairly good at designed power, but only for a while, when output diodes begin to get really hot and I have to turn it off.
I'll try this technique, but... to know I'm doing right if you could upload a pic of it, real better
Thanx a lot!!
Regards
I'll try this technique, but... to know I'm doing right if you could upload a pic of it, real better
Thanx a lot!!
Regards
dtprof
a picture will clear this up.
are the primary windings going out the same side of the bobbin? or opposite, radially or axially?
when you say bifilar wind the secondary. bifilar wind it with what? if the primaries are already down and separated from subsequent primaries.
i am not trying to be obstinate, i just am not clear about your description
a picture will clear this up.
are the primary windings going out the same side of the bobbin? or opposite, radially or axially?
when you say bifilar wind the secondary. bifilar wind it with what? if the primaries are already down and separated from subsequent primaries.
i am not trying to be obstinate, i just am not clear about your description
james, I think I understand it, but a pic wouldn't be bad
I understand that the thing's to make a "plane hole" in the opposite side of the begginning of the primary, when you wind the whole layer and then you leave hanging enough wire to weld it to the begginning of the other layer of primary, and the wind a layer of secondary and do the same and so on, think it's the hell of a technique since all the windings will be pretty plane and smooth and will not return over itself, to the current will look like it's always going in an infinite spiral, never returning
I understand that the thing's to make a "plane hole" in the opposite side of the begginning of the primary, when you wind the whole layer and then you leave hanging enough wire to weld it to the begginning of the other layer of primary, and the wind a layer of secondary and do the same and so on, think it's the hell of a technique since all the windings will be pretty plane and smooth and will not return over itself, to the current will look like it's always going in an infinite spiral, never returning
Now for the layering... I'll supply a transformer symbol in a minut but essentially the construction for the secondary is for a center tapped transformer N1 is one winding to the rectifier and N2 is the second winding to the other rectifier. N1x and N2x are however many additional parallel windings needed to cover the full width of the available area. This means all teh windings will see the same viltage differentials, and the paralleism means that they are very well compled. This technique should yield a leakage/primary inductance ratio of about 0.5% or about 5uH for every 1mH of primary side inductance.
At the end you will pull the wires through the slot in the other side of the transformer.
After you have completed the layering you pull the wires back across the transformer and terminate the wires on the pins.
All views are from the bottom side of the transformer (the side the lines are on).
At the end you will pull the wires through the slot in the other side of the transformer.
After you have completed the layering you pull the wires back across the transformer and terminate the wires on the pins.
All views are from the bottom side of the transformer (the side the lines are on).
Attachments
One more thing I noticed...
In the first message you mention this is a 38V output. For a center tapped HB design the output rectifier should be rated for 2xVo+Vfd+VLk. Or 2x38+0.55+(43.5-48)=81.55V. This means you need a 100V Schotky on the output rectifiers. This may be why you can't last very long before things blow up.
For 250W@38V you would to rate the diodes for (250/38)*0.475(half the duty cycle)*safety margin or >>3A. Normally I would want at least 5A here because of start up issues The caps on the HB will be fully discharged as will all the secondary side caps so the inrush current will be maxed out during start up unless a long softstart period is used. Do not forget that the rectifiers have to support what ever sustained current you can have right before your OCP circuit trips. This should be at least 25% and I would usually err clser to 50%.
If you use full bridge rectification (4 diodes) on the secondary side you would only need a 50V part.
Of course maybe I just misread your original post in which case blow off everything I said.
Tony
In the first message you mention this is a 38V output. For a center tapped HB design the output rectifier should be rated for 2xVo+Vfd+VLk. Or 2x38+0.55+(43.5-48)=81.55V. This means you need a 100V Schotky on the output rectifiers. This may be why you can't last very long before things blow up.
For 250W@38V you would to rate the diodes for (250/38)*0.475(half the duty cycle)*safety margin or >>3A. Normally I would want at least 5A here because of start up issues The caps on the HB will be fully discharged as will all the secondary side caps so the inrush current will be maxed out during start up unless a long softstart period is used. Do not forget that the rectifiers have to support what ever sustained current you can have right before your OCP circuit trips. This should be at least 25% and I would usually err clser to 50%.
If you use full bridge rectification (4 diodes) on the secondary side you would only need a 50V part.
Of course maybe I just misread your original post in which case blow off everything I said.
Tony
Thanks Anthony.
Duly noted. I haven't read the 60065 document.Anyone have a spare copy lying around?
I am a power supply guy and most (but not all) my work is consumer type applications with some telecom work thrown in on the side.
Always want to make sure I follow the rules because I HATE it when it either smokes or I get shocked...
Tony
Duly noted. I haven't read the 60065 document.Anyone have a spare copy lying around?
I am a power supply guy and most (but not all) my work is consumer type applications with some telecom work thrown in on the side.
Always want to make sure I follow the rules because I HATE it when it either smokes or I get shocked...
Tony
dtproff said:
Regardigs this -unread by me- post, the output is 16V, the secondary peak to peak wave is somewhat about 70V (when input from AC line is the lowest this should be around 60 and when highest 85V) normally 35V + 35V, didn't think that I must block peak to peak voltage... For now anyway i'm using 200V ultrafst diodes
Regards
If you are measurung 70Vpk-pk then I would probably use 100V schottky's. You gain two improvements... first the lower Vf and second the PSU will not emit as much EMI since schokky's don't really have much reverse recovery (though they do have extra capacitance). Be sure that you do your measurement at the lowest line voltage you are likely to see since the current flowing through the leakage inductance at the time of the FETS turning off will determine the heighth of the leakage inductance spike.
Anyway... best of luck.
Tony
Anyway... best of luck.
Tony
Tony: I wound the transformer using the technique you depicted the best I could, the results where significant over the amplitude of the unsnubbed spikes, working at D=60%, at 82W output power, the spikes did not last until the other cycle and where approximately half in amplitude than they were, the freq of the ringing is still beetween 5 and 6 Mhz, guess I can still improve this by winding better because I did not do a really great job, I was eager to see what happened...
What 100V schottkey do you recommend? is there one that you prefer?
Regards!
Fernando
What 100V schottkey do you recommend? is there one that you prefer?
Regards!
Fernando
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.