LM5015 Evaluation Card efficiency

Firtst, thank you for considering the LM5015 in you design.  Here I have a few questions:

(1) When you replaced the transformer T1, did you verify that the winding polarities were used correctly?

(2) After modification, is the circuit stable in operation?  You can monitor the LO pin voltage waveform to see if the pulse width is consistent.

(3) How did you get the 84% effieciency?  Was it the number that you read out from the LM5015 app note efficiency curve, or you did get reading of 84% with yours 3.3V? 

Also, a few suggestions:

(1) Change R10, R4 and R8 back to the original values.

(2) Change R5 to 14.7k.

(3) Change R9 to 64.9k.  Your R9=750k and R10=86.9k does not have a correct UVLO setting for your 9V application.

Thanks,

Youhao Xi, Applications Engineering,  Texas Instruments SVS-PPD

Hi Youhao,

   First, thank you for your reply.  I should point out right away that my output is 2W, not 1W (my mistake in the initial posting).

   To answer your questions:

(1) I have the correct polarity on the transformer

(2) The circuit is not stable under all conditions (unstable from about 17V and up), but at 9V and 2W output, it is stable.

(3) I got the 84% efficiency with the card un-modified (i.e. it was producing 5V).  I measured the input and output currents and voltages and calculated 84% efficiency, I did not just read it off the efficiency curve.

   You made some suggestions but I'm reluctant to implement them (because of (c) below, but I will also list why I did what I did):

   (a) For the values of R9 and R10, I used the LM5015_Isolated_Quickstart.xls from TI's web site to find these values.  If you insert 8.5V in box E36 (to set the UVLO rising threshold) and 80.6k in box E37 (value of R10) and "Yes" in box E35 (to indicate you want to use the EN as UVLO), the spreadsheet will show 752k in box E38 (value for R9, I used 750k).  I prefer using values much higher than in the few killo Ohms normally used to improve the efficiency (I usually deal with very low power levels).

   (b) For R4 and R5, they produce the required output with the value I had chosen and they consume less current (I know that R4 affects the compensation, but at this point I'm only trying to evaluate if I can make this more efficient - the circuit is stable at the 9V input voltage and 2W output load I have).

   (c) For R8, I think I would be starving the opto if I changed it to the original value of 590 Ohms (because the output is now 3.3V, not 5V).  Starving the opto would cause U3's output to clip at its reference level and the output would increase to compensate.

   I realize now that I was not clear enough in my initial posting.  I'm already guessing that the efficiency dropped because the losses in the switches have increase by a factor of 16 (i.e. the voltage went from 36V min on the original card to 9V min on my modified card, a factor of 4x smaller so the current will be 4x bigger for the same output power level; since the switching losses are Rdson*I^2, a current 4x bigger will cause a loss 16x bigger)).  What I'm after is if you can think of any way I can improve the efficiency (better diodes, better transformer)?  If so, any part(s) in mind?  Also, because of the fact that I changed the input voltage from 36~75V to 9~36V, do I need to do change the transformer?  The Coilcraft FCT1-33K2SL transformer I'm using is rated for 4W (I only need about 2W), but I'm uncertain about if the transformer would do the job or if I need something else (I'm using it outside of it's designed range, but a transformer is hard to damage and I'm at half the power it was designed for).  If I need something else, would it be just to have less copper loss in the winding (i.e. same turn ration and inductance, just bigger wires) or would I need a different magnetizing inductance?  Should I change the frequency of operation or is it okay as is?

   I will eventually need to tackle the stability also and at that point I may have to change some of the resistors you pointed out.  How did you come up with these values?  This is my first Forward converter (I'm quite familar with flybacks however).  Are the formulas for loop gain the same or very different?

   Thank you for your help,

Daniel

I haven't seen any reply from you yet.  Do you need anything from me?

I was on a business trip and just returned to office today.  Sorry for the delay.

Your concern (c) is not a problem.  We have other designs with 3.3V output and 590 Ohm did not cause any problem.  So please make the changes as I recommended in my previous response.

FCT1-33K2SL should be the right transformer for your 9V input and 2W load. The efficiency drop is mainly caused by the rectifier diodes.  The biggest loss is in the output rectifier diodes.  For 3.3V output, its 0.4V drop reduces the efficiency by about 12%.  For 5% output, the forward drop causes an efficiency reduction of 8%.  This is why you see the 3.3V version has lower efficiency than the 5V version.

To achieve  higher efficiency, you may need to consider to use sync FETs to replace these diodes.

Thanks,

Youhao Xi, Applications Engineering, Texas Instruments SVA-PPD