Tin whiskers and other packaging problems

[Pat Pending]

Hello all. First time post on this board. I've found my way here through reading an excellent article written by Mr Mark Hennessey on the subject of transistor packaging problems. In the late 1980's I was contracted to an industrial control gear company to design a P.I.D. unit to control induction motor loading in industrial process machinery. The project went well, and the finished prototypes worked to a higher level of functionality than was outlined in the original brief. So far so good. I was paid my pittance and left, but then the accountants got at the project, (I bet this is beginning to sound familiar), and the production of the P.C.B.'s was outsourced to the lowest tender. In the original design BC5xx series transistors had been specified, with tropicalization of the whole board to seal against moisture. When the end users started having problems with the assemblies I was summoned to my customers emergency board meeting to explain myself, as to why I had tried to sabotage their business by designing such rubbish! I asked to be allowed to inspect the faulty boards, and when I saw them I laughed in their faces, the (British) assembly company had used lockfit Mullard transistors and had modified the P.C.B. design to suit! Worse still the tropicalization process had been completely missed on the offending boards. Using my prior knowledge of these transistors, (and the fact that they had been obsolete for at least 8years by then) I made the suggestion that they be replaced with BC107 & 177, if they wanted to use obsolete parts. They must have been irony impaired, because thats exactly what they did . In the years since I've come across these units many times, and the biggest offenders for transistor failure are the ones that were retrofitted with the TO18 packages, without tropicalization, what happens it seems is in atmospheres of (relative) high humidity the tin plating creeps the small distance from case (collector) to either emitter or base leadout. The most reliable were the ones that were assembled with the lockfits but had been tropicalized. What can we learn from my ramblings? Well, TO7 packages are not the only one's with tin issues, lockfit transistors can remain stable provided you keep the atmosphere out, and don't let an accountant anywhere near your brainchild. As for me, I now work part time as a maintenance tech. and spend some of my spare time fixing old radios for people. I still smile when I see the inscription "AF117". Best regards. Andy. Ps. Lockfit in this context = Lokfit 

[ppppenguin]

Welcome to GVR. Thank you for telling a salutary story of how accountants can mess up a perfectly good design. We engineers are perfectly capable of doing it oursleves on occasion

I wonder how BC5xx transistors survive in a relatively harsh industrial environment.

[Refugee]

I once worked for a company that was owned by an ex banker. He told me to reuse a transformer that was above the voltage required and below on the amps.
I told him it would not work but he told me to make up an SLA charger with it. The customer made an insurance claim for a new carpet due to the molten plastic dripping on the floor.
The company went bankrupt about a year later.

[Pat Pending]

Hello Jeffrey. The problems were never really caused by the transistors, even the BC5xx types, which to be fair are just commercial commodity type's, would have been Ok if the accounts department had not saved the £1.20 on the tropicalization. The unit that the board was a central part of cost £105 to manufacture and sold at £1200. Spoiling the ship for a ha'penny of tar I think is the expression. Regards. Andy.

[Mark Hennessy]

Hello Pat, and welcome from me, too

Thank you for the compliment! From your experience, it does sound like moisture ingress is the basic problem with the Lockfit package - I know practically nothing about the detail of encapsulation, but wonder if the problem is between the metal and the plastic? Relatively speaking, there is a large overlap compared to other packages, and it's easy to imagine how imperfections in the process could cause that boundary to allow moisture in. Your reports might be anecdotal in the strictest sense, but still very useful as far as I'm concerned. What sort of sample sizes are we talking about here?

Regarding the spelling of Lockfit, I have seen alternatives. I went with "Lockfit" as that's what Mullard seemed to use most often in their documentation, although "Lock-fit" is an alternative. I don't recall them dropping the "k", but I'd welcome any clarification. Another factor, of course, is to consider what people are most likely to type into a search engine

Thanks again, and enjoy the forum,

Mark

[Pat Pending]

Thanks Mark. The sample size was not great. However if you consider that these types of equipment were put into cold/hot temperature gradients and high humidity, basically damp factory's, not peoples living rooms, if this problem could crop up, it did. Incidentally the TO18's weren't tossed out, as the problem with them was on the underside of the package, not internally, they would just be removed, cleaned, dipped in black enamel paint, and resoldered back in place, a Soviet semiconductor manufacturing idea, but it works!  Earlier in my TV and radio repair trade days, a colleague would use this trick on transistors he replaced maintaining "it keeps out the damp", so the idea was not mine to claim. Regards. Andy.

[Pat Pending]

I once worked for a company that wasowned by an ex banker. He told me to reuse a transformer that was above the voltage required and below on the amps.
I told him it would not work but he told me to make up an SLA charger with it. The customer made an insurance claim for a new carpet due to the molten plastic dripping on the floor.
The company went bankrupt about a year later.

Sorry I missed your post. I made a mistake once with 
a customers old radio, resoldering a tap plug on 
the mains transformer in the wrong place, a pitch filled one no less, 
I had to pay the chap who revarnished the ladys
antique sideboard. The only banker involved on that occasion was me

[Refugee]

In my case it was "told you so" and that was it.

[Pat Pending]

Rather than just leave this topic with my anecdote. I thought I could elaborate a little, with some help from an old gent with whom I have become acquainted over the last couple of months. The chap in question, George, after college and national service started work for Mullard at Mitcham in the late 1950's. He continued working within the Philips group throughout the next twenty years. Today, my good Lady and I had to take my old Mum to Norwich hospital for an appointment, and as we were going to be passing quite close by his new home, we called in to black our beaks, so to speak, and wish him and his Lady wife a happy new year. While we were there, I thought of bringing this topic up with him to see if he could shed any light on it, and illuminate it he did! On the subject of the TO-7 he said that at the design stage it was not even known that the phenomenon of tin whiskers even existed, and it was thought that the problem was due to not enough protection putty being used at the assembly stage, allowing contamination to get to the semiconductor. Adding more putty to bed the germanium in and raising the height of the glass seal "pips" that the leadout wires passed through apparently helped, but not for the reason assumed. The modification did not cure the fault as the dispensed volume of protective putty could not be controlled precisely enough, allowing the "whiskers" easy passage to the interconnecting wires via voids in the putty. the cure was only found when the real problem was discovered and lead was added to the plating, as reported by Mark Hennessy in his article on the subject. http://www.markhennessy.co.uk/articles/v....htm#treat . By the time the cure was found, the production life of the AF-114 through AF-118 was nearly at an end, a great shame because for their time they were remarkable transistors. My informant stated his opinion that the aluminium oxide that was added to the protective putty was the cause of the phenomenon, but as the foil had been found he said the organisation wouldn't fund further research. Now the LOCKFIT package. George was personally involved in this one, he claims and made himself very unpopular with the "upstairs boys" over it too! He said if you break open a faulty early version of the lockfit package, you will (under a magnifier if your eyesight is as good as mine), most probably see one of the problems with them , moisture ingress, the phosphor-bronze (?) interconnections will be tarnished. He went on to say that there was pretty much nothing right with the way this package was designed, firstly the silicon was bonded to the collector connection, then that and the base and emitter connections were bonded to the flat back of the package, the base and emitter connections were made and the rest of the package was molded on that with an epoxy/inert filler mixture. The problems were, the flat leadouts were not prepared very well for bonding and therefore made for poor sealing allowing moisture ingress and slow but steady contamination of the transistor. Next, because the transistor was surrounded by the epoxy, it had to be quite soft, i.e. mostly filler, and apparently had the properties of a bath sponge! The other problem with this George said, was because the silicon was imbedded, the fine interconnections would fracture at manufacture (a poet but does n't knowet!) but stay in contact at first, but the contact becoming more unstable as the device was used, eventually becoming very noisy, or failing completely. The team of which George was a member, were tasked with the job of improving the manufacturing of this package, and he said what they came up with was sintering the leadouts to bond better with the package, making a void for the silicon to sit in, to stop the fractured connections, and finally to use a "hard epoxy". Their proposals were all initially rejected on the basis of cost to re-tool, the company saying that they had been designed for consumer electronics and would last decades in a domestic setting, a fair and accurate assesment for the most part, (unless you keep your portable radio in the potting shed?!) a slanging match ensued, with two of the team leaving their jobs to go to work with BEL in the U.S. (from frying pan to skillet?), although the modifications were later adopted. George was also later sacked for conflict of interest, as through the sixties and most of the seventies, until Philips management found out, he had also been running a T.V & RADIO repair business with the help of his wife and brother, the energy of some people. George is unfortunately now almost blind, but still loves to chat about the old times and has agreed to let me do some audio recordings of his memories . I was unpopular with my folks over our long conversation, but it is so rare to actually meet someone who was actually there at such an interesting point in history. When we arrived home I fished an old BC148 out of the bin, that came from a noise infested Fidelity (a quite distinct intermittent low rumble), and squeezed it with a pair of cutters, it fell apart showing me the back of the connections inside the transistor, the position of the silicon on the other side of the collector is marked with the blue dot in the attached picture, and the tarnish is plainly visible. Best Regards. Andy.

[pwdrive]

That was worth posting, interesting read.

Lawrence.