The buzz that is coming from trafo comes only at start, but not as much loud as at turn-off and even at turn-off that is only while bulk caps have enough voltage to power the supply.Since there is 13k uF+ on secondary side and controler is power from there it can be powered much longer than voltage on bulk caps can fall to minimal level of few volts.
I did use it with 0.5 ohm load and that is what I've got:
12.5v @ 25A with ripple voltage being almost 1v.I don't have complited output filter, but 1v at 25A is not that much if you consider that I hade in mind only 17A.I will see what I can find for coat.
I did use it with 0.5 ohm load and that is what I've got:
12.5v @ 25A with ripple voltage being almost 1v.I don't have complited output filter, but 1v at 25A is not that much if you consider that I hade in mind only 17A.I will see what I can find for coat.
In a properly filtered and regulated design you should get way less than 1Vp-p ripple at full load, being mostly switching ripple and spikes as opposed to 100Hz ripple. Something seems to be wrong there despite the fact that you are getting some hundreds of watts of output without blowing anything.
You have to analyse ripple waveform components with oscilloscope first. 100Hz ripple is reduced by improving the regulation loop and increasing primary side capacitance. Sine-like ripple at twice the switching frequency is reduced by increasing output filter L and C values. Triangle-like ripple at 2*Fs is reduced by using output capacitors with lower ESR and increasing filter L. Switching spikes are reduced by improving the overall layout, using lower ESL capacitors and with the help of an additional output "PI" filter.
Ripple
I think you might want to calculate your output filter rather than guess the values , it seems to me you have a lot of L and C for your load requirements. If your load requires 13 Volts @ 20 Amps
and your fsw is 100kHz then:
Lo(min) = (27 V - 13V)(4.25us) / 1.4(2.0) = 22uH
and for .05 Vp-p ripple
Co(min) = (20A)(4.25us) / .05 Vp-p = 1700uF
Flc = 823Hz
Now you need to compensate your supply with a filter that has high gain at 125Hz to reduce input ripple and then rolls off at about half the switching frequency to reduce spikes and switching noise then when you scope the output you will find .05 Vp-p ripple more or less. If you would like to squeeze this you can add a pi type filter as was suggested. With a properly designed feedback your transformer noise should disappear.
chas1
I think you might want to calculate your output filter rather than guess the values , it seems to me you have a lot of L and C for your load requirements. If your load requires 13 Volts @ 20 Amps
and your fsw is 100kHz then:
Lo(min) = (27 V - 13V)(4.25us) / 1.4(2.0) = 22uH
and for .05 Vp-p ripple
Co(min) = (20A)(4.25us) / .05 Vp-p = 1700uF
Flc = 823Hz
Now you need to compensate your supply with a filter that has high gain at 125Hz to reduce input ripple and then rolls off at about half the switching frequency to reduce spikes and switching noise then when you scope the output you will find .05 Vp-p ripple more or less. If you would like to squeeze this you can add a pi type filter as was suggested. With a properly designed feedback your transformer noise should disappear.
chas1
filter caps
Luka
I am sure you have thought about this but if you change values in the output filter then you change the whole feedback network and the supply will not function as the orginal design and therefore it will not work properly since I have no idea about what article you are referring to then maybe I shouldn't comment.
chas1
Luka
I am sure you have thought about this but if you change values in the output filter then you change the whole feedback network and the supply will not function as the orginal design and therefore it will not work properly since I have no idea about what article you are referring to then maybe I shouldn't comment.
chas1
Re: noise 2nd time
FWIW: I didn't mean to suggest to use varish to fix a design issue.
Its still not a bad idea to varnish transformers windings. Loose turns can vibrate together and slowly remove the enamel insulation. Fast Dynamic loads can also cause them to vibrate. An electrical varnish would be perferable over Poly since they have higher breakdown voltages and better heat transfer, but its much more expensive.. Modern transformers are usually vacuum inpregnated to prevent vibration and to increase the breakdown voltages betwen winding layers.
chas1 said:
FWIW: I didn't mean to suggest to use varish to fix a design issue.
Its still not a bad idea to varnish transformers windings. Loose turns can vibrate together and slowly remove the enamel insulation. Fast Dynamic loads can also cause them to vibrate. An electrical varnish would be perferable over Poly since they have higher breakdown voltages and better heat transfer, but its much more expensive.. Modern transformers are usually vacuum inpregnated to prevent vibration and to increase the breakdown voltages betwen winding layers.
noise 2nd time
Techguy
You are right in every respect and my remarks were not intended to diminish your advice. I would go so far as to say that over 90% of commercial designs are impregnated for the reasons that you state. I was trying to reinforce Eva's response since this is a DIY situation where proper closing of the feedback loop is often over looked which is one of the causes of this type problem IMHO and by all means when a project is complete one should take steps to insure protection of components against any environment. Thanks for the reminder and the courteous response.
chas1
Techguy
You are right in every respect and my remarks were not intended to diminish your advice. I would go so far as to say that over 90% of commercial designs are impregnated for the reasons that you state. I was trying to reinforce Eva's response since this is a DIY situation where proper closing of the feedback loop is often over looked which is one of the causes of this type problem IMHO and by all means when a project is complete one should take steps to insure protection of components against any environment. Thanks for the reminder and the courteous response.
chas1
Luka,
Wind them bi-filar, which means, wind them in the direction, at the same time to ensure the tightest coupling between the windings. Winding gthem in opposite directions will severly curtail the performance of the converter. It won't cause catastrophic fasilure, but it just won't perform very well, not responding to load and/or line changes very well.
Steve
Wind them bi-filar, which means, wind them in the direction, at the same time to ensure the tightest coupling between the windings. Winding gthem in opposite directions will severly curtail the performance of the converter. It won't cause catastrophic fasilure, but it just won't perform very well, not responding to load and/or line changes very well.
Steve
luka said:
Yes. The additive nature of the windings means that the flux will help keep both rails symmetric about the zero volt line. So, if there is a surge in power demand on the (-) rail, because it is coupled to the (+) rail, any crests or dips on it will be reflected on the (+) rail equally.
Without crunching any hard numbers, can't say exactly, but off the cuff, I would say your system response to line and load D'es will be slowed some. Regulation will be the same, but the speed at which your SMPS responds to these D'es will be slowed, anywhere from some to considerable, depending on how much over the desired inductance value you have.