I'm involved in a product design (on the mechanical side, to be clear that I'm not an EE) where we need a very compact ac-dc switching power supply that will take 90-300vac, 50-60hz and output 5 watts at 12 volts. From a mechanical engineering perspective, it would be extremely advantageous if this power supply could deliver this power while operating for extended periods of time (many hours) at up to 105 °C ambient air temperature.

Just to make this more challenging our module would need to fit into a space about 15mm high, 25mm wide, and around 50mm long—so a really tight fit. The supply will need to be certified, so details like isolation (3000VAC) and noise that would effect wifi and Bluetooth radios needs to be controlled.

What I would like to know is what kind of considerations would go into whether designing such a power supply is feasible? For example, my limited understanding is that the capacitors are usually the components that limited temperature rating and reliability, but it also seems like there has been significant advancement in high-temperature ceramic and tantalum capacitors that would make such a power supply feasible today (although perhaps relative expensive).

Thank you in advance for any perspective you can provide.

  • \$\begingroup\$ It shouldn't be a problem but the devil is in the detail - what isolation from the AC do you require and what AC level is the input? What range of AC voltages can the design expect to see. Also is there any forced air cooling? \$\endgroup\$
    – Andy aka
    Jun 6, 2015 at 17:41
  • \$\begingroup\$ Ideally voltage should work in most places around the world, something like 90 to 300 VAC, 50-60Hz. Isolation should be 3000 VAC for certification. Double insulated would be nice. And, yes, we can force cool this component to keep the surface at ambient. \$\endgroup\$ Jun 6, 2015 at 17:51
  • \$\begingroup\$ @user2259745 If you can force cool the surface to ambient, why would it go up to 105degrees? Just trying to get the complete picture. \$\endgroup\$
    – Asmyldof
    Jun 6, 2015 at 18:20
  • \$\begingroup\$ The ambient air temperature will be 105C, hence the components will get hotter than that. \$\endgroup\$ Jun 6, 2015 at 18:24
  • \$\begingroup\$ Ah, yes, that can be a problem. You will need to over dimension either a lot of the circuit or the airflow to keep them cool enough. A 20 degree rise inside your power supply is very easy. Let alone inside the driver chip, even if its surface is "only 115", it might be 150 on the inside, that's often when they start doing weird stuff that they aren't supposed to. However, given the right tools and materials it could be doable. \$\endgroup\$
    – Asmyldof
    Jun 6, 2015 at 18:27

3 Answers 3


Probably not a severe requirement- 105°C is not that hot, you can even get electrolytics easily that are rated for 125°C. Assuming (from your tag only) that the input will be AC you may require a fairly large electrolytic filter cap. At a sufficiently high switching frequency you may be able to use MLCC capacitors elsewhere in the circuit.

The size might be an issue- usually high efficiency (which you want at high temperatures) can require a bit more size, and 15mm height (inclusive of housing?) is a pretty tight restriction.

Expect more limited lifetime if it has to operate for a substantial duty cycle at very high temperatures and electrolytics are necessary. Temperature rating for e-caps is for a life measured in thousands of hours, not tens of years, so we usually like to operate at an average temperature much less than the rated temperature.


I've just had a quick scan for ideas about circuits and this came up: -

enter image description here

It uses the fairly reputable TNY267 from Power Integrations and they have design tools on their website - you can plug-in numbers such as input voltage range and output power and voltage and it designs a circuit for you. There is still considerable work to do to get it in the envelope and I would recommend that you use a consultant etc but you should be able to get a decent idea about space envelope by using the self-generated Bill of Material from PI.

I also think you may be looking at getting a supplier to wind the isolation transformer for you too (that's the thing with 140 turns and 8 turns in the diagram).

  • \$\begingroup\$ .The little flyback is going to need a lot of work to do the job .At 300Vac what would the peak switch voltage be ?Efficiency is important at these high ambient temps .How many parts would you need to add to increase the efficiency to say 90% so self heating would not be an issue .The half wave rectifier is not legal in some countries .There will need to be EMC filtering to assure that it passes EMC . \$\endgroup\$
    – Autistic
    Apr 28, 2017 at 21:36
  • \$\begingroup\$ As I said, a consultant might be required to get this done properly so please don't nit pick @Autistic. \$\endgroup\$
    – Andy aka
    Apr 29, 2017 at 0:04

You did not define an input voltage, and since i don't believe you actually are going to use 220VAC (because the you can just use wall mount adapter and because it would never fit your size, at least safe and reliable kind), lets see what would i do.

So first, there are DCDC ICs, with external components. Once you have automotive grad (125deg) you only need efficient design of like 95%, then it's own power will be very low.

There also are modules. The usully require less components around. TI has one that i evaluated and found very nice, although i am not sure about temperature.

Problems: what will you do with the wires? Without filter, which may be quite big, the wire may radiate, you will not pass regulations. Is the outside hotter that the inside of the box? I saw once a design where electronis was heated instead being cooled by the case. You may need some thermal insulation.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.