My stuff flew on all the Apollo moon rockets. And on the C5A, the AH130, and tons of ground test stuff for B52's and U2's and JSF and lots of commercial jet engines. Sailed on lots of ships, too. We never had a failure from gold embrittlement.
But why would a tiny bit of intermetallic at the periphery of a solder joint cause a copper trace to fracture? It wouldn't penetrate the solder or the copper trace any serious distance. We're talking micro-inches of gold here, probably with a nickel barrier layer.
Solder creep was once and still is an issue, and so is embrittlement.
A failure is a failure, and if there is a chance that such a failure can result in a broken connection, then it is not an acceptable system of interconnection.
How about a few pictures of them, both bare, and built?
--
aioe.org is home to cowards and terrorists
Add this line to your news proxy nfilter.dat file
* drop Path:*aioe.org!not-for-mail to drop all aioe.org traffic.
http://improve-usenet.org/index.html
If it were true, you couldn't use digital logic signals crossing clock domains. Otherwise you would have some chance of hitting a metastability window.
--
These are my opinions, not necessarily my employer\'s. I hate spam.
Since the topic is a mechanical electrical connection, I think you need to find whatever group you think it is that your horseshit belongs in, because it ain't here, and it ain't in this thread. I suggest you look into the reasons why radiation hardened chip packaging is offered by some chip makers.
Consider yourself a failure mode. You know... bullshit.
"Your appeal has been denied, like you knew it would be..." -- Cardinal Richelieu
That's silly. Any system, mechanical or electrical, has a finite failure rate. In a given situation, depending on how many lives are at risk and how important the mission goal is, each part of a system is given reliability goals. A guidance computer may have a design MTBF of
200,000 hours, and may be triply redundant, but there's always some accepted level of risk. If no risk at all were allowed, no planes would ever leave the ground and no surgeries would ever be scheduled.
I think the Shuttle is estimated to have about a 1% risk of catastrophic failure per mission, but they accept that and fly.
Sure, a "failure is unacceptable" but in real life individual systems are assigned MTBF goals, and some composite risk is always accepted.
We are often asked to compute the MTBF of our products. We use the Bellcore method and FIT numbers, unless a manufacturer can provide specific values for his parts. Our field failure rate is generally several times lower than the Bellcore calculation suggests. Reviewing field failures, there's often a bad-guy part that's dominating the failures, like a tantalum cap or a voltage regulator. When we spot one, we do something, replace the part or use a bigger heat sink or whatever.
We helped replace a heads-up display on the AC130 gunship. The old system, designed by GEC in England, had an MTBF of 22 hours. Now THAT was unacceptable!
--
aioe.org is home to cowards and terrorists
Add this line to your news proxy nfilter.dat file
* drop Path:*aioe.org!not-for-mail to drop all aioe.org traffic.
http://improve-usenet.org/index.html
You're only worried about the part that gets soldered. Most boards, these days use a soldermask to prevent solder from getting sucked off the pad down the trace. If this is for a home made board without a soldermask, then disregard everything I've said. What I'm spewing is for commercially made boards.
After the gold is dissolved, the solder is in contact with nickle plated over the copper. Commercially plated boards have the copper pads as the base material, nickle plating over the copper, then gold plating over the nickle.
I sorta messed up and specified a 4-layer stack that's nonstandard for the cheapie proto houses, and blue solder mask besides, and 3-day turn, so we didn't get gold and 6 boards cost $1000. My bad.
Upper-left is voltage regulators, so we don't need too big a pile of bench supplies to run this. Then there are four main test circuits, then a bunch of miscellaneous doodads that might be handy. There's also a TDR test trace that snakes through all the layers, J28 to J29, to check impedances, and upper-right are some foil half-moons that let us see the actual layer stackup.
There are also a couple of experiments in using a 50-mil pitch dipswitch as a switchable RF attenuator. We'll see how high a frequency that works at.
Production is populating several versions so I can test the various circuits.
If you google you will, mysteriously, get links. A lot of work has been done recently to characterize solder joint MTBF for lead-free and BGA applications, especially as ragards temperature cycling.
Yup, it's the gunship version of the C-130. The HUD is used to help the pilot spot targets. We did a VME module that generates vector characters to drive the old CRT pedistal. Newer planes use LCDs, but this was a retrofit, to replace some terrible GEC electronics.
The statistic is actually kept as MCCBF. It is a function of the number of connections, in millions, between failures. MTBF doesn't apply on a per connection basis as there are many types of connection, all of which have different life expectancies..
Nice, John... although let me introduce you to MMCX connectors some time. :-)
Did the board house tell you what materials they use? (Heck, tell us who they are if you don't mind?) The biggest concern I've had with, e.g., Advanced Circuits (4pcb.com) is that their cheap ("prototype") service uses a fixed stack-up, and they specify a dielectric constant although usually not the exact material they're using (e.g., Rogers 4413 or whatever... I imagine they use whatever they have more of or is cheaper on any given day).
What frequency would you *like* your 50mil DIP switch attenuator arrangement to be good to? I'm guesstimating that your arrangement there will be usable to around 500MHz but probably not satisfactory by 1GHz.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.