Electronics > Power & Renewable Energy

Are there any integrated thermal protection PMICs / FETs?

(1/3) > >>

I'm looking for a compact IC that switches off my supply rail above a certain temperature, bascially like a bimetallic switch. But I can't really find such a thing?

The problem is overheating in the device at a fault condition. The standby current is in the order of 10mA, with short bursts to about 5A for something like 2s. The supply voltage varies between 5.5V and 8.5V and switches off at a current above 5.5A. So a hard short circuit is not my concern. But with a fault condition and a continuous current of something like 4A the device would completely overheat and possibly start to burn, so I definitly need to implement some kind of temperature protection.

In theory I could use a temperature sensor, a comparator and a FET. But building this with descrete components will become too large, I only have a few mm^2 of PCB space.
I assumed there would be ready to use temperature protection ICs to do exactly that job, but at least with the keywords I used to search I could not find anything.
Thermal protection PMIC results in fan control ICs. Thermal protection FET actually results in FETs with integrated thermal shutdown, but only with the typical fixed junction protection threshold of something like 150°C and nothing I could set to a reasonable case-temperature of 75°C or so.

Does anyone know how I can find the desired IC? Any keywords that might be helpful? Maybe even a specific chip suggestion?

There are several kinds.
Some keywords to search:
"smart fet"
"smart switch"
"load switch" (these are available with over-temperature / over-current protection)
"circuit breaker ic"

Some are drivers for a switch.
Some are stand-alone protection + power switch integrated devices.
Some are power switches like mosfets with internal protection.

e.g. there are many more variants.

Your distinction of case temperature vs. IC temperature @ say 150C is a little problematic.

In a fast fault scenario like a hard short circuit if the switch internal resistance is significant compared to the lead resistances, the IC junction is dissipating
a lot of the actual power compared to the wiring / lead frame and the junction temperature is quickly going to reach high temperatures up to the limit e.g. 125C+
perhaps before the external case / leads absorb enough heat to reach 80C or whatever.  Depends on the electrical and mechanical / thermal parameters.
But anyway if something is going to protect an IC from a destructive overcurrent surge that could destroy the IC in sub-millisecond time ranges it's probably going to be based on reacting faster to the internal current / temperature and not so much limiting the external temperature.

If your scenario is something like a failed fan / heatsink attachment or operating under maximum load for a long time then a gradual overload so ambient temperatures are already reaching ~80C and you're looking for more of a "thermal circuit breaker" based on equipment internal ambient / case temperature then that's a bit different. 

Even a passive polymeric PTC solid state overcurrrent "resettable fuse" will be able to provide "slow" protection to sustained over currents or faster protection to extreme over currents but whether a PTC will protect your pass semiconductor from a short circuit depends on the trip sensitivity / speed of the PTC and what the SOA of the pass device is.

Some of these smart / protected switch / protected fet devices may have temperature / current limits that you can adjust or monitor externally so there is some flexibility in solutions.

You can also choose devices with more or less current and temperature capability so that the time / temperature / current a device trips at with a given
ambient temperature and given current load profile meets your requirements.

Ah, these keywords actually were quite helpful already! Unfortunatly it seems like there is no such IC with an easily settable shutdown temperature.

Regarding your second post: Yes, I am only looking for a protection against long overloads. Currents up to 5A are normal use case, but only for something like 2s. But when a connected device draws 5A for a whole minute, my board would overheat (due to limited trace width).

For the case of a "slow over-temperature" situation:

There are plenty of tiny digital temperature sensor ICs which include a programmable or fixed "alarm" threshold set-point temperature value and an "ALARM" or similar digital output pin which becomes set when the temperature is at/over that threshold.


If your power switch or over all system has a "fault" / "stop" / "shutdown" / "interrupt" or whatever signal which causes it to go into fail-safe mode then you could just "or" or "wire or" the temperature alarm into such an input to stop the system.

Or doing similar with a thermistor + resistor going into a ADC or comparator  to detect over-temperature true / false.  That'd only take two resistive components that could be as small as 1x0.5mm assuming you have a monitor input somewhere.

Also I thought some of the smart switches and similar parts above had some kind of temperature sense output pin so an external MCU or circuit could monitor the IC's sensed temperature.  If you have something that could react to that then that may be sufficient.

e.g. just some of many examples with device junction temperature outputs.

But anyway it sounds like you have multiple distinct possible concerns and approaches:
1: Protect the load switch itself.  In this case if it has over-current and over-temperature protection integrated, it is probably trying to protect itself vs. its maximum ratings.  So if you set the current limit (if adjustable) correctly and it has thermal protection then the device itself should be safe under any so specified condition, but the actual junction temperature at which the device switches off might not be under your control and might not be similar to the overall PCB ambient temperature because of local device area "hot spot" temperature since that device might be dissipating much heat compared to other areas in the system.  I am not sure what the problem is with relying on the device level protection as long as the rest of your system can also function safely at whatever ambient temperatures could be reached until the pass device goes into this protection mode.

2: Limit the operating temperature of "something" in your system by a slowly responding limit to protect against long term over load / over temperature.  Maybe the sensed spot could be near a power switch pass device for instance its junction temperature.  Or maybe it will be some distance away from a pass device if you are interested in more the general equipment temperature instead of a case / junction temperature primarily.  Anyway if you conservatively limit this sensed temperature to something well under the limit of what the components may be at at the same time then that's fine.  You can just have a 2-6 terminal analog or digital temperature sensor (thermistor, digital temperature sensing IC with alert output, whatever) monitor the temperature where you want to sense it, detect the over-temperature threshold being exceeded by a MCU or ADC or comparator or use a digital output directly.  Then digitally (or by analog means?) shut the system down into your desired fail-safe over-temperature mode.

3: But if you are concerned with this over-temperature being slowly reached because of a sustained over current then maybe you can not primarily worry about limiting the temperature to a programmable accurate low value (e.g. 80C) but instead figure out the *worst case* supported scenario use case for your system.  e.g. assume the environment has very hot ambient temperatures, direct sunlight if applicable, minimal supported or no ventilation, poor conductive / convective cooling as supported, whatever.  Then figure out what maximum current / load the system MUST support and continue operating in this worst case scenario.  e.g. 4A load infinitely on X outputs.  Ok that load will rise your pass device and system temperatures by X over and above the ambient no-load temperatures.  You'll get TA+X+Delta junction temperatures and so on.  Make sure your equipment is safe to operate at such limits.  Then if you DO have a accurate "ILIM" current limit control on the protection IC,  you can control on the pass devices, set that limit to be epsilon over that "maximum worst case supported load current" level.  And the pass device will to into over-current shut down just higher than your worst case supported current at the extreme steady state temperature limit.  Yeah if the environmental room ambient temperatures and cooling and stuff were instead a lot better than worst cast you'd still cut off the current at a lower limit than would be possible in that non-worst case scenario but arguably that shouldn't matter since if you designed the system to be usefully "in service" at the worst case environment the more conservative current limit at lower actual temperatures doesn't inhibit the system's useful function at any temperature.

4: If you're worried about slow over temperature and slow over-current you could probably just monitor the current and/or the PCB temperature somewhere convenient slowly and shut down based on either / both metrics if you have some logic / MCU or whatever that can measure those things.  If the pass devices need instant over-current or TJMAX or whatever protections those can be independent fast acting protections.


[0] Message Index

[#] Next page

There was an error while thanking
Go to full version