Crystal solution for full Automotive temp range

"Ulf Samuelsson" skrev i meddelandet news:f0lqhv$h4o$ snipped-for-privacy@aioe.org...

google://crystal oscillator automotive/ gave the answer

If you are having a hard time finding anything suitable within your Google search hits I'd contact a vendor. Since you are in Europe maybe these guys:

Personally I am not a great fan of using crystals in environments that shake a lot though.

You use a resonator, crystal oscillator or what?

I only know a source for USA..

-70 to +200 degrees Celsius range..

Interesting. Their life test is 2,000 hours. That is not long.

The frequency range is specified at DC to 300MHz. A crystal for DC will be quite large.

The temperature limit changes. It is -100C to +250C in

and -100C to +300C in

Regards,

Mike Monett

First I think really hard about whether the app really needs a precise clock. Surprisingly often you can get away with ratiometric tricks and a LC or even RC oscillator, or a trimmed internal clock like the MSP430F2xxx family offers. Maybe Atmel has that feature as well. Rugged resonators is the 2nd path but, of course, usually you still don't get crystal accuracy.

Oscillators in a can is something I try to avoid. Too often I have seen that turn into a purchasing nightmare.

When you have a CAN bus, I suspect you need good accuracy.

Not really. The tolerance as wide as 1% is good enough.

Vladimir Vassilevsky

DSP and Mixed Signal Design Consultant

Well, you hadn't mentioned CAN yet ;-)

I am not familiar with CAN but on other buses it is possible to discard the first few transmissions and use these for synchronization.

The first time we did that was actually on an Atmel 89LV51. Classical situation: It was in a separated interface that gets banged around a lot so I didn't want a crystal in there. Also, it had to be clock-silent for noise reasons when not in action (PCON) and after wake-up the oscillator wasn't particularly expedient in settling. More like a Harley on a cold morning. That meant we had no other choice but to synchronize to the data stream.

Anyhow, sometimes I wish European car mfgs would occasionally take their new designs out into the wilderness here. Doesn't have to be as extreme as our "alternative road to Lake Tahoe" in the link below but I've had people turn pale and grab the dashboard. Including the occasional scream. On roads we'd consider "bad but kind of usable".

try

-jg

Seems to me that an internal CPU oscillator will not stay tuned withing 1% over the requested temperature range. Typical values are 7.6 to 8.4MHz over -40 to +85 deg.

Meindert

When I am looking at a spec sheet from there like this one

I am somehow missing a G-rating. There are automotive applications such as On-Star where automotive electronics have to function for some time after a major crash. The deployment of the airbags isn't necessarily the end of things, electronically speaking.

If I'd be in Ulf's situation and it absolutely had to be crystal precision I'd also look at defense suppliers.

in my business, 2000 hours is a long time.. !! i'm happy to get 1000 hours from a tool.. at 200 C

LOL.. a crystal for DC.. :)

Vectron has some VXCO:s for -55 to +125 degC.

e.g.

This one has a G-rating, since it's a mil spec component.

Yep, Mil-Std-202 is what to look for.

The LIN bus does an auto-baud transaction, perhaps it is time for the CAN bus devices to offer the same (or similar) feature ?

On chip OSC are steadily improving, but some baud-check is a good way to help

-jg

Size might be an issue.

Best regards, Spehro Pefhany

Look at the bright side: In America everything is bigger. Maybe we can build it 8-D