I read the report section on the whiskers forming which bridged two contacts on a tin plated connector. I learned that the whiskers can be cleared by 5 milliamps, I hadn't known that. The flow diagram for faults was interesting..half the end results ended in an orange, unintended acceleration mode..
Pure tin should NEVER be used in a mission critical device. EVER. It is also unclear if conformal coating stops it, as some indications are that the whiskers can grow through the coat. I'm not sure about that. I think it can't hurt.. certainly won't help a bga.
Unfortunately, I believe that now 80% of the components we can buy are pure tin plated. So all that electronic stuff we buy now? Even more throwaway.
The nasa paper also pointed out that there seems to be no issue with tin/silver eutectics and whiskers...that parallels my experience. About 17 years ago, I was one of the largest users of tin/silver in the US (so I was told, a ton or two per year as rosin core, solid wire, and bars for solder pots.), so some mil guys were pinging me for technical details..
Cheers, John
Most chip type SMD resistors and caps etc are pure tin!
There is evidence and photos of the whiskers growing through conformal coatings.
http://www.npl.co.uk/upload/pdf/mitigation-of-tin-whiskers-using-conformal-coatings.pdf
The problem with the tin/silver/copper based solders is no long term reliability figures, we have to guarantee 15 years. Another problem with lead free soldering is when low Tg FR4 is used (mainly cheaper products!, and a glut of Tg 130 laminate after June 2006, that people used, as it was cheaper than the Tg 170 recomended for lead free) the Z axis expansion during reflow can cause barell cracks in vias, this can lead to intermittent failures (and no fault found returns) especialy when the units get hot, add to this the increasing complxity of PCB's even in the simplest of consumer good and we have a reduced shelf life.
NASA Goddard Tin Whisker Homepage
There is evidence and photos of the whiskers growing through conformal coatings.
http://www.npl.co.uk/upload/pdf/mitigation-of-tin-whiskers-using-conformal-coatings.pdf
The problem with the tin/silver/copper based solders is no long term reliability figures, we have to guarantee 15 years. Another problem with lead free soldering is when low Tg FR4 is used (mainly cheaper products!, and a glut of Tg 130 laminate after June 2006, that people used, as it was cheaper than the Tg 170 recomended for lead free) the Z axis expansion during reflow can cause barell cracks in vias, this can lead to intermittent failures (and no fault found returns) especialy when the units get hot, add to this the increasing complxity of PCB's even in the simplest of consumer good and we have a reduced shelf life.
NASA Goddard Tin Whisker Homepage
tin whiskers
My experience shows conformal coating to help.
Before audio I worked on RF. GE had a big batch of tuned cavities that grew Tin whiskers. Bad thing at 1GHz. We wire brushed them and put on conformal coat and they never came back again. They had to be retuned of course but it worked.
My experience shows conformal coating to help.
Before audio I worked on RF. GE had a big batch of tuned cavities that grew Tin whiskers. Bad thing at 1GHz. We wire brushed them and put on conformal coat and they never came back again. They had to be retuned of course but it worked.
Hi Wayne,
the IPC have some papers on this problem. Unfortenatley matte tin is becoming the de-facto finish for a lot of components and untill there is a 100% guarantee of zero whisker growth, certain industries are not going to go lead free, and due to supply limitations are going to have to work round current component finishes. This means either re tinning with a lead/tin solder or trying to ensure that during reflow the solder flows over the end terminations of devices (not easy). The easiest way to guarantee no tin whisker growth is to add approx 3% ww of lead (or silver?) to the tin.
There is lots of exciting reading on this matter on the IPC web site, trouble is its on the members area only.
http://www.ipc.org/MO_pdfs/Proceedings/apexExpo2010/2010-proceedings.htm
Conformal Coatings for Tin Whisker Risk Management
http://www.ipc.org/MO_pdfs/Proceedings/apexExpo2010/S11_01.pdf
Effect of Soldering Method, Temperature, and Humidity on Whisker Growth in the
Presence of Flux Residues
http://www.ipc.org/MO_pdfs/Proceedings/apexExpo2010/S11_02.pdf
Copper Tin Intermetallic Crystals and Their Role in the Formation of Microbridges
between the Leads of Hand Reworked Fine Pitch Components
And I wonder why so little new blood is taking up PCB design as a career
the IPC have some papers on this problem. Unfortenatley matte tin is becoming the de-facto finish for a lot of components and untill there is a 100% guarantee of zero whisker growth, certain industries are not going to go lead free, and due to supply limitations are going to have to work round current component finishes. This means either re tinning with a lead/tin solder or trying to ensure that during reflow the solder flows over the end terminations of devices (not easy). The easiest way to guarantee no tin whisker growth is to add approx 3% ww of lead (or silver?) to the tin.
There is lots of exciting reading on this matter on the IPC web site, trouble is its on the members area only.
http://www.ipc.org/MO_pdfs/Proceedings/apexExpo2010/2010-proceedings.htm
Conformal Coatings for Tin Whisker Risk Management
http://www.ipc.org/MO_pdfs/Proceedings/apexExpo2010/S11_01.pdf
Effect of Soldering Method, Temperature, and Humidity on Whisker Growth in the
Presence of Flux Residues
http://www.ipc.org/MO_pdfs/Proceedings/apexExpo2010/S11_02.pdf
Copper Tin Intermetallic Crystals and Their Role in the Formation of Microbridges
between the Leads of Hand Reworked Fine Pitch Components
And I wonder why so little new blood is taking up PCB design as a career
...... probably spent Uranium
I know some places with that in spades .....
Attachments
Not right now. BTW I posted a thread in Nelson's forum on doing noise with a sound card. There is actually more information in the complete spectrum than in a Quan-Tech plot.
Today's replacement for a Quan-Tech-
An old Quan-Tech. You can use a soundcard and a jig but its not trivial to make it work well enough to get out of the way of low noise transistors. Plus the jig needs a number of adjustments.
I'll be happy to provide the schematics of the last gen Quan-tech. Reproducing the test head isn't enormously difficult but not an afternoon task. That plus a soundcard (192KHz in) and some software will do the job. I would use the internal AGC (uses an internal oscillator) with the FFT to get a direct readout of the linearity of the dut as well as its noise spectrum. You could also use it to screen tubes with the right power supplies. The supplies would be manually set and I'm not sure if we could get the gm readout function, maybe. . .
I would even get the PCB's made if someone steps up to the task of the software and hardware design.
An old Quan-Tech. You can use a soundcard and a jig but its not trivial to make it work well enough to get out of the way of low noise transistors. Plus the jig needs a number of adjustments.
I'll be happy to provide the schematics of the last gen Quan-tech. Reproducing the test head isn't enormously difficult but not an afternoon task. That plus a soundcard (192KHz in) and some software will do the job. I would use the internal AGC (uses an internal oscillator) with the FFT to get a direct readout of the linearity of the dut as well as its noise spectrum. You could also use it to screen tubes with the right power supplies. The supplies would be manually set and I'm not sure if we could get the gm readout function, maybe. . .
I would even get the PCB's made if someone steps up to the task of the software and hardware design.
Actually it was not so hard. With JFET's a gate series resitor makes a good noise reference for calibration. Qualitative measurements were good down to the 1nV level, at the bottom line FET's could be selected for pre-amp applications with a pretty simple set up that anyone could duplicate.
Quan-Tech uses that to calibrate as well, but with a very low noise transistor. If the gain of the transistor changes then the calibration needs to be redone.
Which software was suggested? Perhaps a link to the posts/article?
Of course, use what you have. I have a colleague who uses an HP3563 like I have, for sorting fets, and it works better at VERY LOW frequencies such as 1Hz or less, where he needs it very quiet. The Quantech is used by Kirkwood Rough, Linear Systems, Demain Martin, Constellation Audio, and Charles Hansen, to name a few serious users. Not dead or obsolete yet.
Last edited:
Yes, we got a Quantech off of eBay about five or eight years ago. I think we paid $1200 or so. It was nice because it had all of the manuals with it.
We actually had to buy not for FETs, but for some Analog Devices AD844 ICs. These were designed by the great Barrie Gilbert in the early '90s and we used them for years with no problems. Then they started sending noisy ones.
It turned out that they moved them from a 4" fab to a 6" or 8" process. The new fab didn't know what they were doing (it's all done overseas). There are some impurities that cause LF noise.
At first the ADI people would give us refunds on bad parts and were trying to solve the problem. After a year of two they gave up. Now they just say, "It meets the specs. It's not our problem." It's the first and only time I've gotten anything but superlative service from ADI.
So we have to sort all of the parts before we put them onto PCBs. It's a total pain in the @ss. It's also a shame because I will never ever design a product with that part again. And there is no other substitute for it. So it's back to discrete designs.
Oh well, they sound better anyway.
We actually had to buy not for FETs, but for some Analog Devices AD844 ICs. These were designed by the great Barrie Gilbert in the early '90s and we used them for years with no problems. Then they started sending noisy ones.
It turned out that they moved them from a 4" fab to a 6" or 8" process. The new fab didn't know what they were doing (it's all done overseas). There are some impurities that cause LF noise.
At first the ADI people would give us refunds on bad parts and were trying to solve the problem. After a year of two they gave up. Now they just say, "It meets the specs. It's not our problem." It's the first and only time I've gotten anything but superlative service from ADI.
So we have to sort all of the parts before we put them onto PCBs. It's a total pain in the @ss. It's also a shame because I will never ever design a product with that part again. And there is no other substitute for it. So it's back to discrete designs.
Oh well, they sound better anyway.
- Status
- Not open for further replies.