Hello again friends and thanks for looking at my thread. My question is this....when i complete a new build and burn test the machine, How long is long enough. Previously i have tested units for about 2 hours but occasionally a faulty unit slips through. I use a professional 3rd party burn in program but I'm not sure how long to burn for......Your thoughts please.
Tek-Tips is the largest IT community on the Internet today!
Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!
-
Congratulations strongm on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!
New system burn in...How long??? 1
- Thread starter ckes101
- Start date
-
1
- #2
I would expect that the software would have some sort of recommendation, but I'd hazard a guess. Back in the old days (386 days) most shops advertised a 24-hour burn-in period to make sure that there are no dead parts. That may be feasible in your case, or it may not.
Keep in mind that there will always be machines with bad parts that make it through a burn-in period (if you build in sufficiently large quantities for a long enough period of time). Keeping that in mind, the purpose of the burn-in period isn't to guarantee that all of the parts are 100% good and in working order, but to minimize the number of units that come back in for service due to faulty parts after they have been sold.
If you wanted to guarantee 100% no failures, you could burn in each system for a full week or longer. Of course, then your customers would have to wait an extra week for their PC and your electric bill would go up considerably. You would either probably have to raise prices a bit to offset the extra time/effeort/energy, and that with the extra 1 week wait could actually lower your sales. There's a concept of diminishing returns that basically states that the closer you get to 100% of anything the higher the cost of incremental increases becomes. So catching 50% of faulty parts is easy. Catching 90% of faulty parts is harder, but catching 99.999 of them is extremely difficult and costly to do, and probably isn't worth it. At some point the cost of further burn-in testing will equal the cost of a warranty replacement during the first couple of weeks of customer use, at which point further testing is pointless.
So the real question ends up being, how do you catch most of the low-hanging fruit (from a faulty part standpoint)? If a part will fail if it is running at 100% for 7 consecutive days, that probably isn't an issue since it is extremely unlikely to be used in such a fashion. So you basically want to cover the most common usage scenarios. I don't know how everyone uses their PC, but most people that I know leave their PC on 24/7 and just let it go to sleep when it isn't being used. In a work environment it is pretty standard that the PC would be in use for 8-10 hours per day (or even 24/7 if the business is around the clock). So a burn-in period that simulates that sort of a usage cycle would probably be appropriate.
I would guess that if a PC ran for 12-24 hours at or near full utilization then it you could be pretty certain to have weeded out most of the bad parts.
Keep in mind that there will always be machines with bad parts that make it through a burn-in period (if you build in sufficiently large quantities for a long enough period of time). Keeping that in mind, the purpose of the burn-in period isn't to guarantee that all of the parts are 100% good and in working order, but to minimize the number of units that come back in for service due to faulty parts after they have been sold.
If you wanted to guarantee 100% no failures, you could burn in each system for a full week or longer. Of course, then your customers would have to wait an extra week for their PC and your electric bill would go up considerably. You would either probably have to raise prices a bit to offset the extra time/effeort/energy, and that with the extra 1 week wait could actually lower your sales. There's a concept of diminishing returns that basically states that the closer you get to 100% of anything the higher the cost of incremental increases becomes. So catching 50% of faulty parts is easy. Catching 90% of faulty parts is harder, but catching 99.999 of them is extremely difficult and costly to do, and probably isn't worth it. At some point the cost of further burn-in testing will equal the cost of a warranty replacement during the first couple of weeks of customer use, at which point further testing is pointless.
So the real question ends up being, how do you catch most of the low-hanging fruit (from a faulty part standpoint)? If a part will fail if it is running at 100% for 7 consecutive days, that probably isn't an issue since it is extremely unlikely to be used in such a fashion. So you basically want to cover the most common usage scenarios. I don't know how everyone uses their PC, but most people that I know leave their PC on 24/7 and just let it go to sleep when it isn't being used. In a work environment it is pretty standard that the PC would be in use for 8-10 hours per day (or even 24/7 if the business is around the clock). So a burn-in period that simulates that sort of a usage cycle would probably be appropriate.
I would guess that if a PC ran for 12-24 hours at or near full utilization then it you could be pretty certain to have weeded out most of the bad parts.
I think a person should leave their new systems on a lot when new. Like turn it on every morning and let it stay on till end of the day. This is what I always do and I have been really lucky over the years to not have any failures. I know your meaning system builders but it's also up to the buyer of the new system to give it a good running. I read somewhere that most computer items fail in the first 30 days of being put into service.
ckes101
I understand the thought process but "Burn in" software programs can only test certain parts of a new system.
What about the DVD writer? the floppy drive? that annoying northbridge, graphics card or case fan? the wireless keyboard you bundled that keeps loosing it's channel?
Burn in "software" goes only part way to thoroughly testing a new setup.
Martin
We like members to GIVE and not just TAKE.
Participate and help others.
I understand the thought process but "Burn in" software programs can only test certain parts of a new system.
What about the DVD writer? the floppy drive? that annoying northbridge, graphics card or case fan? the wireless keyboard you bundled that keeps loosing it's channel?
Burn in "software" goes only part way to thoroughly testing a new setup.
Martin
We like members to GIVE and not just TAKE.
Participate and help others.
Excellent point Martin. In those cases you may need to just verify once or twice that something works as intended. You could probably set a new PC to run 3DMark or something similar for an hour or so to test your video card, you could burn a couple of test DVDs, etc. But the more thoroughly you test it the most expensive it becomes.
- Status
Similar threads
- Locked
- Question
- Locked
- Question