UPGRADE YOUR BROWSER

We have detected your current browser version is not the latest one. Xilinx.com uses the latest web technologies to bring you the best online experience possible. Please upgrade to a Xilinx.com supported browser:Chrome, Firefox, Internet Explorer 11, Safari. Thank you!

cancel
Showing results for 
Search instead for 
Did you mean: 
Newbie menagerie
Newbie
364 Views
Registered: ‎04-09-2019

Consequences of exceeding maximum junction temperature

Hi

I hope someone can help me.  We are using a Spartan 6 LX Commercial grade device which is running in a variety of physical environments, some of which cause the maximum die temperature of 85deg to be exceeded by perhaps 5-10 degrees.  I am wondering what the likely consequences of this might be in terms of factors such as lifetime, MTBF, or perhaps in the effects on device speed . timing performance.  We would prefer not to have to fit an industrial grade part for a relatively small temperature violation, if possible.  I was also wondering whether there is actually any difference between commercial and industrial grade parts, or whether they are just tested differently, or whether speed grades are shifted, or whether there are other differences.

Thank you for any help!

Ian

0 Kudos
3 Replies
Xilinx Employee
Xilinx Employee
286 Views
Registered: ‎08-25-2010

回复: Consequences of exceeding maximum junction temperature

Hi @menagerie 

The most importan consideration should be timing performace. Different speed grade is chosen by devie screen. 

Thanks
Simon
-------------------------------------------------------------------------
Don't forget to reply, kudo, and accept as solution.
-------------------------------------------------------------------------
0 Kudos
Highlighted
Xilinx Employee
Xilinx Employee
259 Views
Registered: ‎08-01-2019

Re: Consequences of exceeding maximum junction temperature

Hi @menagerie,

Generally above the recommended temperature, the functionality is not guaranteed but the device should not be damaged. Operating above the absolute temperature is when damage is possible. 

Here are some other forums with more information from others in the industry that may help.

https://forums.xilinx.com/t5/Versal-and-UltraScale/Max-Junction-Temperature-Exceeded-for-MPSoC/td-p/1003869

https://forums.xilinx.com/t5/Virtex-Family-FPGAs-Archived/Exceeding-Operating-Temperature-Virtex-6/td-p/474264

Here are some tips on trying to decrease the junction temperature which you may want to look into for your design if you haven't already done so.

https://forums.xilinx.com/t5/Other-FPGA-Architecture/Zynq-XC7Z007S-overheating/td-p/822940

This app note also goes into some detail on mechanical and thermal design guidelines which could help lower your junction temperature.

https://www.xilinx.com/support/documentation/application_notes/xapp1301-mechanical-thermal-design-guidelines.pdf

Hopefully this helps!

Sincerely,

Courtney

----------------------------------------------------------------------------------------------------------------------
Please don’t forget to reply, give kudos, and accept as a solution if you found this helpful!
----------------------------------------------------------------------------------------------------------------------

Tags (1)
Scholar u4223374
Scholar
163 Views
Registered: ‎04-26-2015

Re: Consequences of exceeding maximum junction temperature

@menagerie To put it simply, if you're operating outside the rated specs then you have no guarantee that the chip will behave correctly. To use the chip in these conditions you would need different timing data (assuming that the whole chip is actually functional at that temperature). I don't know if Xilinx even has that data available; my guess is that they don't, because it would have been a waste of money to measure it. There are some chips where they provide extra timing data (I think one of the older Spartan chips is dual-rated as -4I and -5C, so to run in the industrial temperature grade you just have to accept slower timing) but the Spartan 6 isn't one of them.

The different temperature grades are fundamentally the same silicon, so it won't wear out any faster than an industrial grade chip running at this temperature.

Realistically, the -I part is something like 10% more expensive than the -C one. A small price to pay for the peace of mind. Otherwise you have a risk that at some point you'll order a thousand Spartan 6 FPGAs and find out that, unlike the previous batch, these ones do not work at 95°C. First step here is spending weeks of engineering time tracking down what is probably a horribly intermittent fault (that only occurs when the chip is very hot - so you get to do your testing in an oven!). Then you're faced with desoldering all the chips that have been fitted, scrapping those (no market for used FPGAs unless they're extremely obsolete parts), purchasing new industrial-grade chips (with associated lead time), and fitting those. And explaining to management why there's suddenly a $10K hole in the budget, and explaining to customers that they'll be without a board for a month while you get the right chips and perform the swap.

Tags (1)