Design for extended temperature range

You can improve the Topswitch thermals by adding an additional heat sink. You may have to use the esip package TOP266EG. You can use a larger Topswitch device (TOP267E) that will improve the device thermals. Use higher current rated rectifiers with heatsinks. You can reduce the temperature rise in the transformer by using the next size larger core and larger diameter wire. Also the enclosure design is important. If it is a closed box (no air flow), you will need to conduct the heat from the heatsink to the enclosure surface. You may be able to use a thermal conducting potting compound. Another way is to thermally connect the heatsink to the case through thermally conductive insulators.

Z. Cochrane

Hi PI-Cochrane, thanks's for your informative reply.
Since it is going to be potted and in a closed box I am still quite worry that the heat trapped inside the box. We are designing to a certain safety standards that could prohibit us from effectively heatsink the case through thermally conductive insulator that is thin enough. The requirement needs >3mm gaps between the board and the casing.
Can you please point me to a good PI's reference design that has high operating temperature (higher than 60°C)?

Thanks,
bbmilo

Hi Cochrane,
Another related question is: how the accuracy of the design be improved to ±2% instead of ±5%?

Thanks,
bbmilo

The output accuracy is determined by the error amplifier and reference. If you use a 1% reference and accurate error amplifier with 0.1% resistors, then you should meet 2% regulation.

Z. Cochrane

I do not have a design example for a high temperature application. Thermal management is too dependent on the specific design requirements. Check your safety requirements to make sure that the 3 mm applies to the insulator thickness. Usually insulator system will be approved for specific applications and breakdown voltages and be much thinner than 3 mm. In many cases you need 3 mm creepage and clearance from primary and secondary systems. This means that the insulator needs to be at least 3 mm larger (in each direction) than the device attached to the case. Also you may be able to extend the insulator past the PCB to meet the required safety spacing. Verify this with your safety requirements.

Z. Cochrane

Hi Cochrane,

Do you mean the LMV431A to be 1% and the programming resistors to be 0.1%?
As with the error amplifier, do you mean the opto-isolator, e.g U2 LTV817D in RDR-242? What is the criteria to look for in an device to make it accurate?
Thank you.

If you are using a LMV431A with a 1% voltage tolerance, that device has a built in error amplifier (the device has significant gain). Combined with the programming resistors tolerance of 0.1%, the power supply will have sufficient gain such that the opto coupler variations will not degrade the regulation. If you use a LTV817D (CTR 300% - 600%) you will be fine. Even if you use an opto isolator with a CTR variation of 50 to 600, the LMV431A’s gain will maintain the required regulation of 2%. However, using an opto isolator with such a wide CTR ratio will make your loop compensation more difficult. It is better that you stay with a limited CTR variation of 2:1.

Z. Cochrane

That makes PI very worth considering for our new project.
Thanks a lot for your help.

bbmilo