lm3478 sepic

Hi Daniel,

You mentioned that you are trying to use uncoupled inductors and the recommended Webench values worked for you, is this right ? 

On the LM3478, there are details about sizing input an output inductors seperately. The assumption of L1=L2=L is true only if you are using coupled inductors.  

Webench follows the same procedure as the Datasheet.

Regards,

Ismail

Hi, Ismail

Sorry about the late late answer but I have been working in other proyects but now the sepic is slowing me down. Is urgent to make it work.

Here is the thing: I need a sepic with 9-40V in, 16V out and 6A out at a frec of 470Khz. I´m using the lm3478. I tried to use webench but webench cant simulate this circuit. I search all the formulas of all the data sheets to try to desing my circuit.

The first thing I did was to test one sepic that webwench can simulate, so I assemble a circuit to achive just 2.5 Amps (still 9-40v, 16v out and 470Khz) and it works. The components get very hot (the fet and the inductors) I think thats normal. Is it??

Now I have to assemble the sepic that I need. For 9-40v, 16v out and 6amps out and 470Khz I calculate:

L1= 4.7uH and L2= 22uH. Am I right??

Please help me.

Thanks

Hi daniel,

I will try to run some calculations for you with my colleague and recommend some values.

Regards,

Ismial

Hello Daniel,

when choosing an inductor you must take in account current ripple, current rating, saturation and in particular for stability issue webench calculates the un coupled inductors such that:

Lout / Lin > Vout / Vin

I hope that answer solves your problem.

Regards,

Giuseppe

Hello all,

We are currently working on a minor update on webench for SEPIC. It should be ready soon. We recently had a similar conversation on E2E with another customer regarding choosing inductors for SEPIC. It culminated in to the customer designing using coupled inductors. Please review that discussion:

http://e2e.ti.com/support/power_management/non-isolated_dcdc/f/196/p/171942/636868.aspx#636868

Regards,
Akshay

Hi,

Thanks for your help.

Just one thing, I´m doing some test with my sepic circuit but the mosfet turns super hoy and dies. I look for several mosfets with los rds on and low Q but still cant find the right one. All of them turns super hot and melt the solder. Now I´m Trying with heat sinks but I dont want to use them in my final pcb desing. I want to use the pcb as a heat sink.

My sepic conditions are: 9 to 40v in, 16v out and 6A out. I´m using the lm3478.

please help.

Thanks.

Hi Daniel,

Would you share your schematic? Your design requriements, as such, are very stringent. 6A of load current would require a much larger input current at low input voltages in  SEPIC. Therefore for this design you would need to rate your MOSFET for a high current and a high voltage (Vin+Vout) also.

At what switching frequency are you opearting your desing at? Running at high switching frequency helps with sizing the BOM small, but has an adverse effect on efficiency. The switching component of the MOSFET losses increases dramatically at high frequencies.

Regards,
Akshay

Hi Akshay,

Thanks for your quick response.

This is my sch.

Some values that I calculate are:

Duty cycle: 0.65

Iripple: 4.26A

IL1rms: 11.2A

IL2rms: 6A

IL1peak: 13.3A

IL2peak: 8.13A

L1: 4.4uH , L2: 18.47uH

IQpeak: 21.46A

IQrms: 13.87A

The mosfet in the sch gets super hot so I´m going to test the BSC123N08NS3 GCT-ND.

Whats wrong with my desing??

Thanks for your attention.

Hello Daniel,

The load requirements for your design are very large. As you've calculated, the FET rms current is about 14A. The FET that you were using (SI7322) has an RdsOn at 10V Vgs of about 50 mOhms. This will result in a large conduction loss in the FET and that explains why its getting so hot. The other FET you've mentioned (BSC123N08NS3) has a much lower RdsOn. But it does have a larger gate charge. I would suggest that if you are using that other FET, you should lower your switching frequency. I hope this helps.

Regards,
Akshay