Why are there so few 105C rated embedded WIFI solutions??......

[Smokey]

There are only a couple of options for adding wifi to an embedded system with elevated temperature requirements.  Almost all the chips stop at 85C.  It seems like there should be a bigger market for at least 105C rated parts....

I've only found one part that's actually in stock anywhere, but it's pretty expensive:
SiLabs WF200DR
https://www.mouser.com/datasheet/2/368/wf200_datasheet-1633290.pdf

There is an NXP part that looks hard to get in small quantities:
NXP 88W8987
https://www.nxp.com/docs/en/data-sheet/88W8987_SDS.pdf

And a part from TI that isn't available yet...
https://www.ti.com/lit/ds/symlink/cc3300.pdf

.... and yes... I know... ESP32...

What do the automotive guys do for this? I know it's not ESP32. 
It looks like there are some new parts coming out specifically for automotive WIFI, but I have a hard time designing in a part when TI won't give me a date they will actually be available in distribution.

What other 105C rated WIFI solutions am I missing?

[tszaboo]

Industrial temperature ranges and Wifi makes very little sense to me. One expects a highly demanding, high reliability environment, where a connection loss could be damaging. The other, where connection is optional, with unreliable wifi.
How do you even power your device? Do you already have a battery wich is rated to 105C? If so, then the cost of the wifi chip is the least of youf concern. And if you power the device through a cable, why do you need wifi in the first place?

Automotive is different, because the infotainment systems usually have much worse rating than the rest of the car.

[voltsandjolts]

Almost all the chips stop at 85C

Many chips will operate above manufacturer temperature ratings. Try one and you might be surprised. Of course for production use it would be a bit risky.

[Smokey]

Industrial temperature ranges and Wifi makes very little sense to me. One expects a highly demanding, high reliability environment, where a connection loss could be damaging. The other, where connection is optional, with unreliable wifi.
How do you even power your device? Do you already have a battery wich is rated to 105C? If so, then the cost of the wifi chip is the least of youf concern. And if you power the device through a cable, why do you need wifi in the first place?

Automotive is different, because the infotainment systems usually have much worse rating than the rest of the car.

It is for automotive.  It's essentially a wifi data link for monitoring and data logging download.  A connection loss would not be damaging, just annoying.  But it would be great if someone happened to be driving through phoenix in the summer that it didn't spontaneously combust.

[TheUnnamedNewbie]

Infotainment systems in automotive are generally much lower rated (I think Grade 3 usually, which is 85C). This is because of the combination of being in-cabin and non-safety critical.

[Smokey]

Lets say you are parked in phoenix in the summer.  Outside temp is 110F, cabin is 160F (71C).  Lets also say the device is under the dash and has a chance to normalize to that temp.

Wifi needs a fair bit of power and that chip will have it's own temp rise above the ambient (which is 71C). 

So even if you get in the car and blast the AC, the cabin temp will drop quickly, but the device will still be elevated when the Wifi starts self heating. 

I'm just trying to make a case why 85C could be marginal in real world conditions.

[TheUnnamedNewbie]

Silicon starts behaving weirdly in the microwave region at high temperatures, doesn't it? I'd expect to see detuning of certain components (oscillators, mixers) in the signal chain over 100C ambient.

Sure, but you design for it and take it into account (temperature sensors controlling tunable capacitor banks/VGAs/temperature compensated biascurrents). At 2.4 GHz it is quite 'trivial' to do this - keep in mind there are companies that make 77 GHz and 140 GHz sensors that are automotive rated, doing in-cabin 2.4/5 should be easy.

[coppice]

If the flash is on die there are issues. Many flash processes have retention issues above 85C. So, you have a restricted number of process choices if you want 105C operation. Of those you need a process that is also rated for RF applications. I'm not sure how many choices that leaves you with. There are multiple 105C WiFi choices, two of them from major industrial and automotive suppliers (NXP and TI), so maybe the market is adequately served. Achieving rock solid long term 105C operation doesn't come without cost.

[Geoff-AU]

It is for automotive.  It's essentially a wifi data link for monitoring and data logging download.  A connection loss would not be damaging, just annoying.  But it would be great if someone happened to be driving through phoenix in the summer that it didn't spontaneously combust.

Use a bimetallic thermostat to turn it off if it gets too hot.

[Smokey]

It is for automotive.  It's essentially a wifi data link for monitoring and data logging download.  A connection loss would not be damaging, just annoying.  But it would be great if someone happened to be driving through phoenix in the summer that it didn't spontaneously combust.

Use a bimetallic thermostat to turn it off if it gets too hot.

Ha.  I can go grab the one from the thermostat at my parents house.  It's been running for about 70 years, so it should be proper burned in at this point. 

[Smokey]

... turn it off if it gets too hot.

This brings up an interesting question.  Is this actually an acceptable solution?  If there is a non-critical component that doesn't meet the temp rating of the rest of the system, can you just not use that lower rated part when it's too hot in the enclosure, but otherwise run as normal? 

[radiolistener]

Almost all the chips stop at 85C.  It seems like there should be a bigger market for at least 105C rated parts....

Higher temperature usually used for military grade or automative grade components, they are more expensive

[T3sl4co1l]

... turn it off if it gets too hot.

This brings up an interesting question.  Is this actually an acceptable solution?  If there is a non-critical component that doesn't meet the temp rating of the rest of the system, can you just not use that lower rated part when it's too hot in the enclosure, but otherwise run as normal? 

If the part specifies storage temperature in the range, sure.  Be careful to isolate everything, not just power; often inputs/outputs can parasite power the device.  Check that it won't, or provide a means to disconnect everything.

Tim

[coppice]

Almost all the chips stop at 85C.  It seems like there should be a bigger market for at least 105C rated parts....

Higher temperature usually used for military grade or automative grade components, they are more expensive

Remember when 74xxx meant 70C max, and 54xxx meant 125C max? There was also a 64xxx series with 105C max. The 64xxx series didn't get much market traction and disappeared. As time went by a lot of basic parts moved to 85C max. because it was easy to achieve, and it gave an edge to those parts. Possibly the low end extension from a 0C to 70C range going to a -40C to +85C range gave a lot more value to the average customer, but it happened because it had become easy to do. Now a lot of industrial customers are saying if 125C is hard (expensive), we'd be interested in 105C parts, if they don't have a huge price premium. I suspect the 105C market may grow a lot in the next few years.

[Gribo]

Cheap RF designs use crystals - These drift with temperature. AFAIK WIFI can tolerate 80ppm between transceivers, beyond that, data rate and range drop drastically.