And if they would just admit that war is inherently dangerous and that risk is acceptable if they choose to continue in it, they could buy their parts from Digi-key like the rest of us. And save a lot of money. Oh wait - that would cut into SOMEBODY's profits.....🙄
I used to have this guy renting space from me in a warehouse. He had a big oven/freezer thing he used to put things in and heat and cool them repeatedly to check there durability. I believe it was part of some contract to supply the military. Only problem was the machine would break half way thru a ten day test and he would have to start again. It was really pretty funny. The operator was on a pitiful wage, but he was with this thing 24/7 so the OT really added up. He was writing a book at the time! He was there for a good three months
No. For large system it does not work that way. As an example let's say your aircraft has 10,000 critical parts. Then if you buy frm Digikey there is a on in 100,000 chance a part might be bad. This means your aircraft has a 10,000/100,000 chance of failure. Flying such an aircraft is orders of magnitude worse than fighting in any modern war. Plus the cost of loosing one in any ten aircraft is unacceptable.
The math is simple: With critical parts, probabilities add together.
With VERY large systems like space missions the number of critial parts is SO LARG that unless you really do go "nuts" with testing and traceability the mission has a poor statistical chance of working at all and you will just blow up rockets 99% of the time.
There are two methds to address the problem at "probabilities add" (1) use designs with fewer critical parts by building backup systems and using fault tolerant designs. or (2) use ultra-reliable parts. Modern system use both and try to strike a balance.
Actually your idea of simply buying from Digikey and accepting a few crashes is used some times but as system become more complex you find that 100% of them crash. Yes 100% will fail if you build a big enough system with questionable parts because to get to Mars EVERY ONE of those 10 millions parts has to work. The chance of that happening with DigiKey parts is about zilch.
Oh yeah? What about Mundorf?😀
No, this is a very interesting discussion. I have bags of fancy parts that came out of aerospace surplus in Dallas. The guy who sold them to me at the hamfest said he is "not allowed to say" what kind of project they came out of.
Each gold plated leads, ceramic-encased, precision film resistor has a hand inked serial number and a bunch of tiny different colored paint dots that I presume are markers that tests have been performed. I have hundreds of these. Really difficult to imagine this level of tracking but this thread shines the light on that process.
What happens is they need the get good statistics on a device. The only way to get stistically valid data is to build hundreds of them and test hundreds of them. Even if you only need one device the laws of statistic require you to test hundreds. They many times build far more than they need just to support a test program. Likely what you have is some of the tested devices that were not needed.
Years ago there some people at Hughes Aircraft build an atmospheric probe to be sent to Jupitor. It used an 8-bit micro processor. (I think a standard Motorola part at the time.) But how do you know the little uP is radiation hardened? Well ,they buy an entire batch of the parts, thousand of them and place them in a radiation test chamber of weeks or months then measure the failure rates to build a dosage vs. failure curve for the part. You have to kill hundreds of them. They do the same thing with acoustics and heat and cold and make curves for all of those things. Then peole wonder why they "wasted" so much money on a $2 part. MOST of the parts survive the tests but they only need a few.
The problem is that you don't know what the test did to the part. It could be on its last legs, you just don't know.
- Status
- Not open for further replies.