9
\$\begingroup\$

I heard that someone mentioned the method of screening: to have 10 MCUs working at -55 degree, and find out the ones which can work properly, throwing the broken ones away.

Is the method applicable? I am worried that the MCU may work properly at -55 degree at my screening test and fail at the real working environment.

If not, what could be the possible solutions? We are using stm32f4, due to its very good DSP performance. The MCUs working at -55 degree we found do not have DSPs and can only work at low frequencies around 20MHz.

\$\endgroup\$
  • 7
    \$\begingroup\$ Any time you operate a part outside of its absolute maximums, you are playing with fire. Whoever said that is just asking to get burned in a big way. \$\endgroup\$ – Matt Young Aug 7 '15 at 14:11
  • 6
    \$\begingroup\$ The possible solution is to find an MCU with appropriate temp rating. Or add some heaters around. Or just leave as is an do many-many tests. \$\endgroup\$ – Eugene Sh. Aug 7 '15 at 14:12
  • 1
    \$\begingroup\$ @MattYoung more like playing with dry ice, as the case may be :D \$\endgroup\$ – Passerby Aug 7 '15 at 15:31
  • \$\begingroup\$ At such low temperature you will most likely encounter issues with your PCB material, Standard FR4 cracks and its epoxy breaks at such low temperatures. \$\endgroup\$ – Lior Bilia Aug 7 '15 at 15:40
  • \$\begingroup\$ Welcome to the often expensive and brutal work of batch destructive testing and statistics. This is why quality assured parts are so much more expensive than their seemingly identical, non-QA counterparts. Companies will spin off whole new firms to handle the cost and paperwork of this one thing, then market QA parts to offset the cost of what they originally needed. \$\endgroup\$ – Sean Boddy Aug 8 '15 at 6:09
14
\$\begingroup\$

The crude way you'd ensure you were not on the edge of operation would be to test it outside the range. For example, you might test the parts at -65°C at voltage at a higher clock speed and higher/lower voltage than normal.

The manufacturer probably does not test at temperature extremes themselves, but they know how much margin is required under test conditions and they test to that. They also know how to ensure that they are testing everything. You don't know any of that. For example, something like an oscillator might work fine down to -40°C, and once started work to very low temperatures, but some may fail to start at -45°C. One particular instruction may start to fail first because of some timing conditions.

If the manufacturer can supply units qualified to that temperature that would be best. Or lobby for a relaxed requirement. Or put heaters in there to guarantee a minimum temperature after an acceptable warm-up period (perhaps inhibit operation until acceptable temperatures are reached).

Chances are if the parts need to meet a military lower temperature range, you really need to be sure that it works reliably.

\$\endgroup\$
  • 2
    \$\begingroup\$ Chances are if the parts need to meet a military lower temperature range, you really need to be sure that it works reliably. Exactly. You want such extreme tests with a reason. The most common reason is military-grade requirements. Testing the way OP is proposing is absolutely not up to such standards! \$\endgroup\$ – Mast Aug 8 '15 at 21:04
11
\$\begingroup\$

As you said, it is impossible to tell whether you degraded the unit during the tests outside of its temperature range. You have two options:

  1. Contact the manufacturer asking them what temperatures extremes the parts should be able to withstand. Because you may have noticed that temperature ranges are very similar. The manufacturers may choose a temperature range that has been commonly specified, which is feasible and yet appealing to the potential customers, then test to that temperature range even though they might think the parts could be more robust. Testing is expensive. If they miraculously answer, I guarantee they will never answer with certainty and they will say it is your own responsibility if your parts die. However you might be more reassured about testing your unit and using them at that extended temperature range, depending on your reliability goal.
  2. Use heaters, temperature sensors and a control system (which can be the same microcontroller + a driver) to control the temperature within set limits. You are lucky enough to want to operate at low temperatures, which is easier and cheaper for the same ambient temperature difference. Honestly, this is not much trouble for a much more reliable system. Just make rough calculations for the power requirement of the heater, then solder a power resistor and a thermistor near your chip, add a driver for the heater (which could be a simple transistor), and control those from a proportional-integral controller running at low frequency in the background. That's 1 equation to solve, ~5-6 components, and ~10 lines of code. You don't even have to use PI control, comparators with hysteresis could do the job. Just make sure the entire PCB is heated uniformly (say, with long thin traces for the heater), because I suspect it will be even more at risk than your chip, becoming brittle.
\$\endgroup\$
  • 1
    \$\begingroup\$ The heater argument is a good one. If you make sure that the PCB is in good thermal isolation it won't even take much power. \$\endgroup\$ – Nils Pipenbrinck Aug 7 '15 at 18:45
10
\$\begingroup\$

What you are referring to is sometimes called 'uprating'. It's the opposite of 'derating', which you'd do to some or all of your components depending on your application and reliability needs.

Here is an old article on the subject of uprating. Their recommendation at the end is a good one -- contact the manufacturer to understand what might be affected by operating at low temperature. They will never guarantee operation outside their limits (unless you are a big/strategic customer for them) but they may be able to provide some guidance on what they would be most concerned about, which could help you formulate a good life test/screen.

The real answer depends on tons of factors. Is it going to see thermal cycles (going between hot and cold) or just operate down at -55C? Thermal cycling induces mechanical failures in bond wires and IC packaging. Is it a 'one off' vs 'mission critical' application, ie., what are the consequences of seeing a failure. If it's a 'one off' (single unit being built for short term use), you might be ok with testing a few units. If it's a mission critical situation, or the part will permanently be operated at a low temperature, you probably will want to spend more effort on the qualification.

Screening like this has been done for military applications for years. The important thing to understand is where the real "cliff" in the parts performance is. We can all agree the parts will probably not perform at -200C. And we can probably all agree the parts will probably perform just fine at -41C (just outside the STM32F operating range). The manufacturer has put in some guard band on their components operating range.

The relevant questions are-- can you figure out where the guard band is (and does it include your desired lower temp range), and will it ever change across multiple lots.

Figuring that out will require testing many parts to get good statistics on the reliability of the parts at low temperature, and what the distribution of their failure looks like, so you can predict if the failure mode is likely to appear in your implementation. And then, once your product is in production, you will have to monitor the parts performance with some kind of acceptance sampling.

An alternate approach to all of this is to install a heater, and use the STM32F's die temp sensor as feedback in the heater control loop. Doesn't help for a cold start but if it's a continuously running unit it might be ok.

\$\endgroup\$
2
\$\begingroup\$

I am assuming the MCU is CMOS though you do not say so. All MCUs suffer from self-heating problems which limits the maximum operating temperature. For example an iPhone with the charger plugged in could feel about 50C to the touch but be 125C or more internally when operating. So the test limit for your MCU, normally controlled during qualification with a thermostream, will guarantee that the design limit is OK. Once you go below that limit, transistor delays will reduce which introduces the possibility of race hazards. Additionally intrinsic carrier concentration will reduce which will have an effect on mobility. If your MCU has A/D or D/A converters, their characteristics, for example maximum error, could increase, or not work at all.

Derating the frequency will not help at all (this may help with high temperature). The chief downside with using the device outside of its range is that even if the probability of error is low, it will still be significant with millions of instructions per second being executed. If you are not too particular about power consumption, you could disable the power saving routines in your code (such as halt, sleep, etc.) and this will result in a small self-heating effect, which could be enhanced by using thick insulation. However if your device has to work at very low as well as high temperature, this would be a problem.

Pre-qualifying your device will not be much help unless you have access to slow and fast lots from your manufacturer; these will be extremes of doping and other parameters such as metal thickness to assess the reliability.

If you have a fat budget you could license your own processor from ARM or one of its competitors and harden it yourself to your own temperature specification. This is known as a customer own-tooling (COT) approach. If necessary you can license memory controller IP and peripherals as well. An alternative would be to approach a manufacturer that specialises in customisation and ask them to prequalify your required product over an extended temperature range.

A manufacturer that does customisation will have access to the all the computer aided design (CAD) databases needed to verify a chip. It is then a simple matter to revalidate the design at a lower temperature. However they may be reliant on a second vendor to characterise the silicon at a temperature outside of the usual range. This requires an extensive range of SPICE simulations and associated library characterisation experiments, which may be outside the scope of what they are willing to do for all but the largest customer. As part of this process the previously mentioned thermostream may be used to check that the split lots still pass their test vectors at the low temperature you specify. This may also result in a yield loss as mentioned by other answers.

\$\endgroup\$
  • \$\begingroup\$ What does the customization manufacture do? \$\endgroup\$ – richieqianle Aug 8 '15 at 15:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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