Haa!! Some stupid scientists are still researching this stuff ...
Fools ...
Right, this is the keystone to the next generation of audio products. As Jimi said blah blah woof woof.
where have I heard those two words together
A week ago, I made an attempt to explain the significance of termination errors in numerical simulations, in which cases they are of importance, and when not, to a rather irrational individual.
Afterwards, the lights were still off.
Every die I have ever seen, touched, or used, has an aluminum bonding pad for every node which is intended for the outside world. IIRC, there was one SOS device which used gold. And I believe current gens use copper. If a pad is not created for a node, I do not know how to attach to it.1) Do you think that they could attach a bond wire even though there may not be any feature on the die to do so? I always thought that the die had to have a gold pad or similar to attach the bond wire to...
In this neck of the woods (NY), there were a few very small hybrid houses (10 employees or so) that I would think could do the work. You'd best discuss packaging with them prior to purchase, as well as die type. Bonding of the die can be epoxy, solder, gold/tin or gold/germanium.. Reflow either nitrogen furnace, scrubbing eutectic gold/silicon or ge, and sealing via epoxy, solder again, gold tin, or one shot welding. (one shot is just sooooo easy and quick) The hybrid house would need to have the equipment in house.
If you choose this path and wish more information, PM me and I'll give you what I know. (well, I'll answer any question to the best of my ability anyway...)
The major issue is always tce..
The glass is a low temp fired glass, and it is physically sensitive. So much so, that any bending of the lead near the meniscus tends to crack it at the bonding interface, and that location will suck in flux at the tail end of a soldering operation as the package atmosphere cools down and contracts. This was a typical field failure mechanism for military hybrid returns.
Kovar was the normal lead material. It bonds well to the glass, and has a compatible expansion coeff. I recall copper only leads 20 years ago or so, used for IR drop in high current packages and skin/proximity reduction with switchmode.
I suspect it's just tce. I recall some vendor doing a copper to alumina bond (direct bond copper, or DBC), but it was a very tricky process. Copper oxide melts 2 degrees C lower than pure copper and bonds to aluminum oxide, so the manu had a belt furnace and process control capable of roughly .2 degree C accuracy at 1060 degrees C.
Nah, I'm tapped..
Charles, as I offered...pm me if you would like more details, I'd be happy to help...
Same applies to all here..
jn
Last edited:
The expanded wafer format is great for automatic pick and place with pattern recognition. Waffle packs are typical for small quantities, so which is better will depend on wafer size, die size, quantity purchased, and things such as min orders and cost to pick.
I actually had some vendors which refused to sell probed wafers because they didn't want us to know what their yield was, and others who refused because our specifications were so tight.. Devices which we couldn't use were still product they could sell.
The bad side to waffle packs??? One of my operators dropped a full waffle pack of screened 1N4148's on the floor.. IIRC, they were either 15 or 25 mil squares 12 mils thick.
jn
Given what I know about assembly, quality and reliability, I would not be recommending any of the proposals floating around here. Assembling semis is not something to be taken lightly.
For the record, my current employer assembles 70 billion devices a year and we have a problem when ppb hits 3 digits.
For the record, my current employer assembles 70 billion devices a year and we have a problem when ppb hits 3 digits.
Yes but would he do 2000 units for less than the price of a midsized compact?
BTW Funny how you judge outfits so easily as unreliable...
Jan
I agree it should not be taken lightly. But for the rel levels in this industry, 100 ppb is kinda in the mud. I suspect CH doesn't have an issue with PPM levels, nor even 100 PPM.
Besides, I'm confident he burns them in. I suspect your current employer does not perform 100% burn in.
jn
Actually Jan, I believe he intended to point out how difficult it can be to do this kind of stuff and get reliability, and not so much judging outfits.
He is correct. How many non chip types even know what a spittle shield is??
So many things can go wrong, that's why I mentioned small hybrid houses.
jn
I suspect with what has been suggested here the reject rate at final test would be in the % level and there would huge issues with infant mortality, even after burn in.
And, what about the parametrics after an assembly job like this? Some of the problems I ended up with on products were incredibly subtle:- they did not perform to spec and the final root cause and analysis showed that small changes and mundane every day actions lead to problems. In the semi industry, most of the problems in running production arise in assembly - both for initial rejects and field failures.
Semi companies spend a huge amount of effort to build reliable, repeatable products. I would not want anyone here to think this is easy, or that they could buy a second hand machine, train a tech up and get anything like a quality product.
And, what about the parametrics after an assembly job like this? Some of the problems I ended up with on products were incredibly subtle:- they did not perform to spec and the final root cause and analysis showed that small changes and mundane every day actions lead to problems. In the semi industry, most of the problems in running production arise in assembly - both for initial rejects and field failures.
Semi companies spend a huge amount of effort to build reliable, repeatable products. I would not want anyone here to think this is easy, or that they could buy a second hand machine, train a tech up and get anything like a quality product.
The15 year old article JC quoted is interesting to me because it explored ballistic resistor at room time if they are small in size . I agree if it liquid helium cooled it not a mass production product.
Might be, good for you to point out the problems.
I believe failures after burn in are no longer called "infant" mortality by definition. At the single digit production scale, burn in is pretty easy.
You have succeeded in that regard. Again, that is why I pointed to small hybrid houses.
jn
- Status
- Not open for further replies.