The past couple months I've been asked to diagnose three ~25-35 year old pieces of audio equipment/test gear that were malfunctioning and the problem has traced back to the same fault - dead or dying 25-35 year old CMOS 4051 analog multiplexer. Two were from the same mfgr and one not.
What gives? Just random chance or is something killing these things now?
bitrex wrote
Where those in input circuits that could have been exposed to outside [high] voltages?
Nope. Only common things is all driven fast and continually to switch DC voltages when operating, and two were in sample/hold circuits feeding hold capacitors. All Japanese-brands with date codes from the mid-late
80s, Toshiba, Hitachi.I wonder if in the S&H rig-up they could be drawing deceptively large currents, not apparent with pen n paper design, when charging up the hold caps and electromigrating the interconnects to destruction over time. The failure mode is dead channels. Not sure the designers intended the chip to be used that way...
4051s are a known failure point in Tektronix 24x5 scopes, of similar vintage. They continuously multiplex front panel pots and demux a single DAC to set related internal voltages.
I have used CD4016, CD4051, CD4053 and the HC versions many times, also in sample and hold, mostly the HC ones as video switch. All Philips. Oldest one I can find was scrapped 8 years back, had real life of 2010-1984 = 16 years without any problems, on almost 24/7, CD4016. but had 10k resistor in series with the signal path.... no peak currents. Yes electromigration can be a cause. Peak currents are limited in those chips, what was it, on resistance 150 Ohm or so? No minimum Ron specified for the HC (looking at datasheet). For the CD4016 Ron is much much higher
So, maybe the peak current is sufficiently limited in s/h in the old chips.
Was not there some problem with dual in line plastic packages long ago?
Tin Whiskers?
-- This message may be freely reproduced without limit or charge only via the Usenet protocol. Reproduction in whole or part through other protocols, whether for profit or not, is conditional upon a charge of GBP10.00 per reproduction. Publication in this manner via non-Usenet protocols constitutes acceptance of this condition.
I have used those chips for many decades. I've had no failures, but worried about vulnerability, from the beginning I've used DIP packages with sockets.
--
Thanks,
- Win
Doesn't electromigration require current flow sustained for long periods, I'd have thought s/h flows would be extremely short duration?
piglet
Was pure tin used on lead-frames then?
piglet
Is this three instances of the same gear or three separate designs?
Is Vdd high, as in 15 or 18V?
piglet
If they are CD-series (the original high voltage parts) they are prone to destructive SCR latchup, which can be triggered by small current spikes into the ESD diodes. Most HC-series parts are protected from that.
Power supply sequencing can trip the latchup too, especially in a mixed-signal circuit.
I'd guess that a s/h wouldn't move enough charge to cause electromigration.
Or maybe they are just bad parts. ICs weren't as reliable back then, compared to now. Early plastic-encapsulated chips were horrible.
-- John Larkin Highland Technology, Inc lunatic fringe electronics
the latter seems like the most probable explanation, these are all American-made products but filled with Japanese mfgr Toshiba and Hitachi logic, no Texas Instruments or nothin' to be found here. Hard to service too because the PCB manufacturing "quality" of these double-sided thru hole boards seems pretty sketchy as well. Built on a budget
Holes are not plated through and after 30 years even moderate amounts of heat applied for a little too long will easily cause the board to drop pads, so swapping even a single defective IC is an anxiety-provoking chore
The usually-quoted electromigration limit for top-level Al/Cu metal is
1E6 A/cm**2. Metal-gate CMOS was/is made on a 6-um litho process, and guessing that the metal is 100 nm thick, that's about 6 mA for a minimum-width line.Electromigration has an activation energy, so for the same total charge, it's worse for short pulses and high temperatures. So it's not totally implausible that it might be electromigration, depending on the litho, at least in the DC case. I agree with JL that latchup is a more likely cause.
What were the symptoms?
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics Briarcliff Manor NY 10510 http://electrooptical.net http://hobbs-eo.com
Interesting. Why were you particularly worried about that specific chip being vulnerable?
In my ignorance I'd have been more concerned about the sockets.
Fortunately I've never had to worry about 30+ year reliability.
On a sunny day (Sun, 9 Sep 2018 17:42:10 +0100) it happened Tom Gardner wrote in :
Gold plated turned sockets for DIL I have always used, and never a problem. Makes life so much easier...
PICs are on same sockets too.
Some stuff with sockets here from the eighties still works fine today:
Even the nicad is still OK... Will program 8049 and EPROMs too. The potcore is part of the program voltage generator.
2018 - 1985 = 33 years.
Neither do I ;-)
Symptoms on all are one or more "dead" channels, failed open-circuit. e.g. a clock is feeding a binary counter or somesuch and the appropriate bits feeding the three strobe inputs of a chip and it goes switch switch switch - channel dead sitting at ground due to the output pull-down - switch switch switch... etc.
Two of the pieces were old synthesizers that use an array of 4051s to multiplex the output of one (costly at the time) 12 bit DAC into the ~40 or so DC control voltages required to drive the rest of the circuitry.
Back in the day SSM made a chip specifically for this purpose the 2300 with the S&H capacitors integrated on the die. The S&H caps in this discrete version are relatively large, 0.022uF. The off-state leakage in the integrated variant could probably be better controlled.
For whatever reason 4051 of this vintage failing in the S&H arrangement seems to be a common prob I've read others online talking about this type of failure and I've seen two fail in that configuration myself.
If I had to guess I think maybe it has to do with sizing the hold cap, the precise value may be critical to long-term reliability. Also in this arrangement the analog switch is being reverse-biased and passing leakage current for much of its operating life
I snip off all the pins first, then desolder them one at a time.
-- John Larkin Highland Technology, Inc lunatic fringe electronics
Yep, after I bodged one of the pads trying to de-solder the pins one by one with a sucker and then extract the IC I've done the same. It feels a bit barbaric but with the failed chip still in the leads wick too much heat, and if you apply heat for too long on these old pads too long it's sure to drop.
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.