Problem flyback AC/DC.

3 posts / 0 new
Hello, I´m using TNY279G for a flyback AC/DC, with two outputs: 5V to 1A and 24V to 0,7A. The problem is that the exit of 24V is not stabilized. If R106 is not mount and there is no load in the outputs, the 5V output ripple is very low and the 24V output is good. But when I put a load, i have the following problems: - If I only put load in 24V, this output goes down to 12V and the 5V output work perfectly. - If I put load in both outputs, the ripple in 5v output is very low and it is stable, and 24V output goes down to 18V with very high ripple. When I mount R106 the results are very similar. Which is the problem or mistake?? Thanks.
첨부 파일Size
24V_5V_A.pdf25.44 KB
transformer datasheet.pdf46 KB
FAE, Power integration product line prasunkuls@hotmail.com 1)Remove R101. 2)Further, why it is wrriten R106=NC, is it not connected? if not connected, connect it. Assure the value of It as well. Note: as you trying to control both the outputs, regualtion would be poor.

The behavior you describe is not completely unusual but let me try to explain what I think is going on. At a high level the way to think about the feedback loop is this:

The regulated output is the 5 V, this is where feedback is derived ie the TL431 (D27) in your schematic acts to keep the 5 V output at 5 V. This sets the turns per volt on the secondary of the transformer (Vo + Vf of rectifier diode). Thus the other outputs are determined by the number of turns minus the voltage drop of their rectifier.

However what we've not take care of is the leakage inductance term. In this schematic
http://www.powerint.com/sites/default/files/images/pi3394.jpg the two secondary leakage terms have been added.

These leakage inductance terms cause voltage errors between the two (or more output windings). Again back to the 5 V output. The voltage across the winding is set by VO + Vf + Vleak(5V) (the voltage across the leakage inductance on the 5 V winding).

Back to your question
The problem with the value of Vleak is it varies output by output depending on the output current. So if the 5 V is fully loaded (Vleak(5V) is large) and the 24 V is lightly loaded (Vleak(24V) is low) then the 24 V output voltage will increase as the turns per volt is higher. 5 V lightly loaded and 12 V fully loaded the situation reverses, Vleak(5V) is low and vleak(24V) is high so the 24 V output voltage will fall.

It's possible to trade off regulation between the two output by doing sum regulation from both outputs (the purpose of R106). This doesn't eliminate the effect of leakage inductance but allows the 5 V output to vary to improve the 12 V regulation.

There's one more situation. If the 24 V output is unloaded then this output will peak charge due to the small leakage inductance induced spike at turn off of the primary each switching cycle. Every cycle the output voltage will incrementally increase. This usually requires a small pre-load on the output from which regulation is not derived (24 V in your case).

Hope that helps - I started writing this expecting the answer to be short so I hope I've not confused everyone.

There is a much more detailed explanation as part of an answer to Puzzler 3:

http://www.powerint.com/en/community/documents-quizzes/puzzler-3#answer2

and although targeted at TOPSwitch this AN has some background on designing multi output flybacks: http://www.powerint.com/sites/default/files/product-docs/an22.pdf

Conclusion
It's not possible to get perfect regulation from a multiple output flyback when one output is fully loaded and another is at zero load. In most applications this condition does occur but if you do have a condition where the 5 V can be fully loaded and the 24 V is at zero load or vice cersa there are active circuits to help cross regulation (http://www.powerint.com/node/231). Finally reducing the value of the leakage inductance terms helps. For example the windings (especially the 5 V output as it has fewer turns so occupies less of the bobbin width) should be arranged to fill the width of the bobbin by increasing the wire size and/or using multiple parallel wire (multi-filar). This is discussed in AN22 also.

Cheers

PI-Chekov