[time-nuts] aging/failure of un-powered xtal oscillators?
lists at rtty.us
Fri May 17 09:08:34 EDT 2013
The target design lifetime of your typical clock oscillator is in the 10 to 20 year range. These parts are well past their intended replacement date.
Things like package leak rates, and epoxy issues do indeed kill oscillators. Both are a time dependent thing and are not particularly dependent on operation / non operation. There are of course other issues that do depend on the unit being in operation. More obscure things like dendrite growth on traces and plating issues are also an issue. Weather they are operation dependent depends on which experiment you look at.
If you replace them, I suspect you are in for some sort of basic re-qual. It's highly doubtful that the original 1980's BOM for those oscillators is still something that can be built. The divider chip designs used back then are probably long gone….
I would almost guarantee that it's a divide by 64 or a divide by 128 inside the package. In that era, odd frequency blanks were cheaper than non-binary dividers.
On May 17, 2013, at 8:07 AM, Grant Hodgson <grant at ghengineering.co.uk> wrote:
> A client company has sourced a quantity of 'New in Box' iSBC series memory modules manufactured by Intel in the 1980s for a MULTIBUS based computer system. These are still in their original, sealed packaging and have been stored (for 25 years) in controlled conditions. These cards are required as part of a refresh programme for a mission-critical application (electricity generation), which are currently using original Intel cards from the same era.
> The memory cards use a 64.1kHz oscillator module as the refresh clock for the DRAM. (I suspect that the oscillator module uses either a 2.5MHz or more likely a 5MHz crystal whose output is divided down within the module by 39 or 78 to 64.1kHz).
> As this oscillator is used only for DRAM refresh timing, the accuracy and drift are not particularly important. However, reliability is important - i.e. the oscillator must function, even if it is a few ppm off frequency.
> AS a matter of precaustion, all the tantalum and electrolytic capacitors on the NIB cards are to be replaced as a matter of course, and any socketed ICs will probably be removed, cleaned and replaced (to be decided). There are several dozen cards to be replaced.
> There is an option to replace the crystal oscillator module, but it is uncertain as to whether it makes sense to to the extra work. The replacement oscisllator module will be from a different supplier to the one Intel used, and would need some form of qualification.
> Is there a known failure mechanism whereby a cystal oscillator module that has been sat on a shelf (as part of a complete card) in a benign environment develops one or more characteristics that could impact it's reliability?
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
More information about the time-nuts