TO220 heatsink is way too small for 60W Offline Flyback?

Hello,
Our contractor has designed a 60W offline Flyback LED driver for us (VAC Input = 190-265VAC).
This is for 7 hours use per night in UK, and the PCB is inside a totally sealed plastic enclosure.
It is 89% efficient. This contractor has chosen the Wakefield solutions 274-1AB heatsink for both the T0220 power FET and the TO220 Secondary diode.
He insists that this is all that’s needed, and is not allowing any more room on the PCB for any bigger heatsink. We believe that this heatsink is way too small…do you agree?

We havent thermal tested it because the PCB isnt yet made.

Output is 60V, 1A.
The flyback operates in boundary conduction mode.
The switching frequency is around the 80kHz mark.
There are no fans in the enclosure.


TO220 Heatsink (P/N = 274-1AB by Wakefield solutions)
https://uk.farnell.com/wakefield-so...y=https:en-GB/Element14_United_Kingdom/search

Hi, I agree. It has to dissipate like 7W, the thermal resistance is 28C°/W. The temperature at the junction will be around 225C°, that is too much. But in Syberia it could work on colder days ...maybe.

The heatsink parameters are valid for natural convection, not for hermetic enclosure. I wonder however if the efficiency calculation is substantiated and if it tells the exact FET and rectifier losses.

Yeah, you need a whole 'nother enclosure design plan.

Given the risk of thermal runaway on a hot day, plus natural variation among components, I'd strongly query the current design.

Sounds like a case for a metal case or a metal back-panel used as heat sink.

You need to study some anatomy!

The total dissipation is slightly more than 6W which is very small but ALL depends on the source.

Without knowing the details, let us assume that the diode dissipates 2W and the FET does the rest.

There are two heatsinks right? One for the diode and the other for the MOSFET.

What is the nature of the plastic enclosure? If it is black and thin, perhaps you can get away with it. If you have a choice, select a corrugated surface.

An adult, with a 2 m2 surface and body temp 32C, radiates about 100W when the ambient is around 20C.

I guess you will be fine.

Sounds like a case for a metal case or a metal back-panel used as heat sink.

Click to expand...

Thanks, this is interesting.
The following are “long-life” outdoor LED drivers, and they have a plastic case, with no metal back panel in them...

https://www.docs.lighting.philips.c....3-1.0A_SNLDAE_230V_C133_sXt_929001408506.pdf

What would you say should be the maximum allowable junction temperature of the FET and Diode?
(eg in a thermal test done at 23 degC ambient)

- - - Updated - - -

I have heard of a rule whereby the lifetime of a semiconductor is doubled for every 10 degrees below Tj(max), but i can't find anything official on this...do you know where i can read about this?

I have heard of a rule whereby the lifetime of a semiconductor is doubled for every 10 degrees below Tj(max), but i can't find anything official on this...

Click to expand...

This is clearly wrong but the underlying principle is simple.

Rate processes have an exponential term of the form exp(-E/RT).

If the process is taking place with a moderate velocity at room temp then a 10C rise will approx double the rate. (you expand the exponential)

If the mean expectation of life at the beginning is 10 years and you reduce the junction temp by 10C and you expect the mean life expectancy to become 20 years? Possible but only if temp is the only cause of device failure.

Do I guess right that you got the 89 % efficiency by copying the Philips specification and didn't yet determine individual component power dissipation?

In this case, you could continue to use their thermal parameters:
- maximum ambient temp 55 °C
- maximum case temp 80 °C

case overtemperature of 25 K seems as plausible estimation. It corresponds to about 11 W/(m²K) heat transfer coefficient for the total case surface, a reasonable ballpark value.

Assuming similar natural convection inside the enclosure (may be worse, depending on PCB temperature distribution and internal geometry), you end up with a maximum air temperature inside the enclosure >= 105 °C.

In other words there's little margin for additional heat sink thermal resistance. If you examine e.g. notebook power supplies, you'll find relative large area cooling plates all around the enclosure.

Using contact heatsinking by thermal gap filler or molding is another approach.