Offline Flyback SMPS diode at 120 degrees C

Hello,

We have designed a 60W PFC’d Flyback LED driver. Output is for 60V LED string.
VAC input is 240VAC.

The output diode is STTH3010D.

When the LED driver is in its plastic enclosure, which is itself inside a metal enclosure, we record 120 degrees on the tab of the diode. This is with the equipment in 22 degrees centigrade ambient. We cannot afford to buy a bigger heatsink than the one that we have for it. And in any case, we couldn’t afford to re-do the board so we could fit a bigger heatsink in. –This is because the extra cost would make our product uncompetitive with already existing offthe shelf products made by huge electronics corporations.
The diode datasheet says it is for use up to 170degrees Centigrade at the junction. However, going above 100 degrees Centigrade obviously bares with it the danger of thermal expansion/contraction damage. However, our equipment is only switched on and off once per day.
Due to this, do you think that the product will not suffer badly from expansion/contraction damage, due to the low amount of power cycling events? (the product will be on full power for 6 hours per day)
The external ambient that the product is installed in has temperatures ranging from -10degC to 24degC.

STTH3010D datasheet:
https://www.st.com/resource/en/datasheet/stth3010.pdf

Can you still consider alternate parts such as silicon carbide?

https://www.digikey.com/product-detail/en/cree-wolfspeed/C4D02120A/C4D02120A-ND/2679410

the reverse recovery will get worse with increasing temp then => short ...

- - - Updated - - -

you need to test ...

Thanks, sorry i forgot to say we are in DCM so not getting too much reverse recovery.

One possible solution would be to use 2 diodes in parallel (if the cost is not too high). Several SMPS use parallel diodes when current handling capacity needs to be increased. I guess you can mount 2 diodes in parallel without changing the PCB layout (if its SMD component)

Hi,

paralleling diodes is critical.

The problem is the thermal drift of V_f. dV is about -2mV/K
This means the hotter diode will increase the current .. and further increase temperature .. then again increase current ... and so on.
They will drift apart and one diode will carry almost all of the current if you don´t carfully design it.

Read this for detailed informations: https://www.st.com/resource/en/application_note/dm00098381.pdf

Klaus

Silicon carbide diodes parallel well.

If you work out the forward dissipation of the diode ( ~ Irms x Von ) and these losses are somewhat below what you are observing - then there must be some element of switching loss coming into play - in that case a better diode needed, perhaps 1200-V SiC ...

The ARRL Handbook advises load balancing between parallel diodes. Put a resistor inline with each diode.

Suppose two diodes are rated 1A, and they have 0.1V difference in V_fwd. Then a 2A load burns up one diode. The other diode gets 50mA.

So add 0.3 ohm resistor in each leg. This reduces the imbalance, and you can use tiny 1/8 watt resistors.

Just put a couple of holes in the enclosure that shall help air flow. If the device is to be mounted vertically (or otherwise) just ensure that a decent airflow path is available.

If the 120C measured (in the equilibrium state; after 1/2 of warmup say) steady state, there is no serious problem I can think of. The danger of thermal expansion or contraction is seriously misplaced.

I also suggest a couple of holes in the PCB that will allow air flow: but it depends on the assembly. Few small holes are better than one large one.