LM3478: Instability issues while simulating SEPIC topology in LTspice

Martin M.

Part Number: LM3478
Other Parts Discussed in Thread: LM5155

Hi,

I’m currently facing problems with the simulation of the LM3478 in a SEPIC topology. I think the problem is not related to the “light load” problem of the LM3478. I’m also considering using the LM5155, however, since there is no unencrypted simulation model of the LM5155 available, I tried doing the simulation with the LM3478 model.

My goal is to implement a SEPIC with the following parameters:
Vin = 5..20V
Vout = 12V
Ioutmax = 5A
fsw approx.. 600kHz

However, since I didn’t get the simulation running as desired, I decided to take a step back and simulate the circuit dimensioned in AN-1484 (Vin=3..5.7V,Vout=3.3V, Ioutmax=2.5A, fsw=330kHz). I used components as calculated and dimensioned in AN-1484.

When simulation with an input voltage of Vin=5V, everything works as expected. Inductor currents show behaviour as expected. Also, the output voltage looks good. There are no discontinuities in the switching behaviour (DR pin of LM3478).

When simulating this with an input voltage of Vin=3V, I get ugly output voltage and especially “instable” inductor currents. (see figure). There are also discontinuities in the switching behaviour (DR pin of LM3478).

I could share the simulation file if required.

Am I simulating something wrong?
Can you find any error in my simulation?
What is going wrong in the circuit?
Any ideas on how to get the circuit stable and working?

I would be happy for any kind of tip.

Thanks and regards,

Martin

over 5 years ago

0 Youhao Xi over 5 years ago

TI__Guru* 78060 points

Hi Martin,

Thank you for your interests in our product. What is the sim tool that you used? TINA or Pspice? Can you send your model circuit to us so we can do an investigation?

Thanks,

Youhao Xi, Applications Engineering

0 Martin M. over 5 years ago in reply to Youhao Xi

Prodigy 170 points

Hi Youhao

I'm using LTspice. To get the Pspice model work in LTspice I had to make little modifications (e.g. remove unnecessary {}-brackets) to the model. Please find the circuit and the model attached.

Thanks and regards,

Martin

SEPIC_AN-1484.zip

0 Youhao Xi over 5 years ago in reply to Martin M.

TI__Guru* 78060 points

Hi Martin,

We are not able to use LTspice owing to copyright issues. Can you use TINA or Pspice?

Thanks,

Youhao

0 Martin M. over 5 years ago in reply to Youhao Xi

Prodigy 170 points

Hey Youhao,

I just downloaded TINA and tried to adjust the average model available on the product page to the parameters given in AN-1484. It's not working properly yet. Seems that the output voltage is not correct when doing a transient analysis. However, I'm new to TINA, so probably 'm doing something wrong.
Also do you have any transient model available for the LM3478? There's only an average model for AC analysis on the product page.

Regards,

Martin

snvmb83_SEPIC.tsc

0 Youhao Xi over 5 years ago in reply to Martin M.

TI__Guru* 78060 points

Hi Martin,

I am afraid the average model is set for boost operation. It may not work for SEPIC. We do have a transient model but it is in Pspice. Let me ask our colleague to see if we can extract the SEPIC TINA model from Webench. Will update you in a few days.

Thanks,

Youhao

0 Youhao Xi over 5 years ago in reply to Youhao Xi

TI__Guru* 78060 points

Hi Martin,

I am still waiting for our modeling expert to advise. I will let you know as soon as I get the answer.

Thanks,

Youhao

0 Brian (CPM-USA) Cheng over 5 years ago in reply to Youhao Xi

TI__Prodigy 270 points

Hi Martin,

Thanks for the patience.

I would like to check if LM5155 TINA model could help you in this case or not.

Based on your original email, you were trying to use LM5155 in the first place, but were forced to use LM3478 due to lack of TINA model for LM5155?

LM5155 TINA model will be released on TI.com within few days, so if it can help to solve your issue, then that would be great.

If you have to use LM3478 TINA-Trans model for any reasons, please let us know. Thanks.

Regards,

Brian

0 Martin M. over 5 years ago in reply to Brian (CPM-USA) Cheng

Prodigy 170 points

Hi Youhao and Brian,

Thanks for helping.

Yes, I want to use the LM5155 but tried to simulate SEPIC topology with LM3478 due to the lack of LM5155 TINA model (resp. unencrypted PSPICE model).

I'm looking forward to the LM5155 TINA model. Hopefully it will work for SEPIC topology.

Regards,

Martin

0 Brian (CPM-USA) Cheng over 5 years ago in reply to Martin M.

TI__Prodigy 270 points

Hi Martin,

Sorry it took long to get the LM5155 TINA model.

Please find the model from the link below. Thanks.

Regards,

Brian

0 Martin M. over 5 years ago in reply to Brian (CPM-USA) Cheng

Prodigy 170 points

Hi Brian,

Thanks for the model. I'll try to simulate the SEPIC with it this week.

Regards,

Martin

0 Youhao Xi over 5 years ago in reply to Martin M.

TI__Guru* 78060 points

Hi Martin,

Please update us asap.

Please be noted that we will be able to offer follow-on support next week. Wish you a happy Thanksgiving!

Regards,

Youhao

0 Martin M. over 5 years ago in reply to Youhao Xi

Prodigy 170 points

Hi Youhao

The model seems to be working. Implemented the SEPIC and the first tests where successful. Thank you!

While waiting on the model, I made my one little evaluation board with the converter. I'll probably come back to you with more questions on the converter and SEPIC topology when putting the board into operation. However, I'll open a new post if required, since it will not be related to LM3478 instability issues anymore.

Thanks for your help and also a happy Thanksgiving.

Best regards,

Martin

Power management

Power management forum

LM3478: Instability issues while simulating SEPIC topology in LTspice