[time-nuts] Lifetime testing

[Bob kb8tq]

Hi

> On Sep 1, 2020, at 4:09 PM, Hal Murray <hmurray at megapathdsl.net> wrote:
> 
> 
> kb8tq at n1k.org said:
>> I believe that they simply don’t have a body of long term data “in house” to
>> study. 
> 
> Interesting topic.
> 
> How many units do you have to test to get useful data?

Likely a couple hundred spaced out over a few years of production.

>  How often do you have 
> to measure the output frequency?

They flat out die ( = no output ). If you catch that +/- a month it’s probably “good enough”.

> 
> Are electrolytics the typical weak link?

They have been in the ones I’ve  fixed. 

>  Do their vendors have good data?

Apparently not.

> What's next on the list?

Roughly 20 other parts in the device ….

> 
> I assume the first step is to run them 24/7 at max rated temperature.

You likely would run them at multiple temperatures. 

>  What's 
> the next step and how much more do you learn?

Power / temperature cycle them.

> 
> How much life testing to vendors of other gear do?  PCs?  Disks?  Memory?  
> Instruments? (counters...)

Since this is a “many years of testing” sort of thing ( = you run them for maybe 20 years)
I’d say zip ...

> 
> 
> Is there a good web page or video on long-term-testing of electronic gear?  

Nope. Nobody has the money to do it on large groups of parts. (well, almost nobody …). 
What does happen is that the customer may be able to track field failures. The
issue there is that the environment (and other drivers) is rarely well documented. 
That information is normally conveyed in quarterly review meetings. 

Bob

> I'm looking for the hour or three hour level rather than a PhD  program.
> 
> -- 
> These are my opinions.  I hate spam.
> 
> 
> 
> 
> _______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com
> To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
> and follow the instructions there.