LM3478 Sepic Converter Question.

Hi TI, I would appreciate some help, as it appears from the traces that the switching frequency is way too low, but checking Rfadj = 37.4k, as per WEBENCH design?

mb.

It appears from the traces that the switching frequency is way too low, but checking Rfadj = 37.4k, as per WEBENCH design? I could use your help.

mb.

Hi MB,

Can you please share your WEBENCH schematic or your design input conditions; VinMin, VinMax, Vout and Iout? We will investigate the issue.

webench_design_LM3478_FINAL.pdf

Happy to do so, also Lout is 100uH which is the size most available at the time.

Also here are two traces of the input voltage to the cct taken across the input capacitor, which is fed from my commercial bench top power supply set to 15V.

I think it would be helpful to include two more captures of the output voltage, at a time base of 1uS, and at 100uS.

Also how critical is the input capacitor to correct operation? ieESR, size and value.

mb.

Hi Jazmine, thanks for looking into my issue. I posed the question about the input capacitor on my last reply because as I was examining the components, I cannot identify which input filtering capacitor was used as a college who populated the board can't be reached at this stage. It is of course a surface mount tantalum with case markings (156V 90702). I have found two part numbers in his catalogue, I believe its TPSC335K035R0700 or TR3C335K035C0600. The output capacitor is easily identified by its markings, and is as per WEBENCH design.

Hi Martin,

I'm going to order and build this board in order to debug in our lab. I see no obvious problems with your set-up and you seem to be using most, if not all WEBENCH recommended components. I should have an answer for you shortly.

Regards,

Jazmine Livingston

Ok, thanks Jazmine, looking forward to what you find.

mb.

Hi Jazmine. ?

Hi Martin,

I'm still looking into your design. I've built a board and am getting ready to test on the bench. How many copper layers does your board have? And how are those layers connected to the ground plane? If you are using vias to connect the two layers, that could be the cause of the added ripple at the output. 

I will continue to look into this and can also send you a coupon code that can be used to order a blank LM3478 SEPIC (uncoupled) board along with all the components WEBENCH recommends so that you can test in your own environment. If you would like one, shoot me an email at Jazmine.Livingston@TI.com.

Regards,

Jazmine Livingston

Hi Martin,

I can only duplicate the issue you've described by using non-electrolytic caps on the output. In which case, I get about 250mV ripple with a lot of ringing. With an electrolytic cap, I'm well within the 170mV ripple WEBENCH spec'd. Since you are using an electrolytic cap and still seeing much higher than expected ripple, I suggest taking a look at your board layout, specifically at how many copper layers you have (our boards have 2) and how those layers connect to the ground plane. If you are using vias, that could introduce more resistance to the system.

If you would like me to send you a coupon code to order the board and exact components from your specific design, feel free to shoot me an email at Jazmine.Livingston@TI.com.