Design Con 2015
Breaking News
Memory Designline Blog

A memory problem not to make light of

NO RATINGS
View Comments: Newest First | Oldest First | Threaded View
Danilo
User Rank
Rookie
re: A memory problem not to make light of
Danilo   11/8/2010 2:51:54 AM
NO RATINGS
I do not discard optical interference any more after three different experiences. One FM receiver that changed tuned station every time I flipped the lights on, due to a glass packaged varicap diode. A PABX operator console using an old EPROM based Z8 microcontroller that refused to work in a clear room and a drag race Christmas tree (for the ones who know what is it) that started to get crazy at mid day because bad quality optocouplers. The optocouplers got intereference from the sunlight even enclosed in a metal box with some small vent holes.

Rick_Hille
User Rank
CEO
re: A memory problem not to make light of
Rick_Hille   11/6/2010 2:24:23 AM
NO RATINGS
Charles, I haven't run into this so far but I often wondered if it was just a matter of time before supposedly non-optical parts become optical. I've seen more and more designs with (painfully) tiny packages and I've occasionally quipped to my colleagues my observation that some modern chip scale packages seems to be little more than shellac'ed die! Maybe it'll get to the point when the wave of a "magical" hand makes a problem disappear!

Robotics Developer
User Rank
Rookie
re: A memory problem not to make light of
Robotics Developer   11/6/2010 1:16:43 AM
NO RATINGS
I remember using PALs and GALs in an array processor design mid 80s and had similar wierd experiences with "pin compatible devices". It seems that the particular devices from one vendor worked all the time and the 2nd source type devices would not. The parts were "identical" feature-wise, voltages (as spec'd) and operationally the same. The problem was with the programmer that we were using, it just did not do a good job with the 2nd version of the parts (and because we had stopped verifying the programing, we did not see the problem - it was intermittent and only for certain functions/inputs). The moral of the story is don't scrimp on programming/parts cost without doing a full investigation. When we added the programming verification step we weeded out the few badly programmed parts.

Tom N
User Rank
Rookie
re: A memory problem not to make light of
Tom N   11/5/2010 11:53:21 PM
NO RATINGS
When testing wafers we often need to eliminate light sources, some die designs are sensitive, some are not. And it's especially noticeable at our probing stations with the microscope light source, as one fellow above found out the hard way.

CharlesGlorioso
User Rank
Rookie
re: A memory problem not to make light of
CharlesGlorioso   11/5/2010 10:35:33 PM
NO RATINGS
The problem described hasn't gone away with UV EEPROMS. A few years back we were working with a very small product with lot's of itty-bitty (technical term) surface mount parts including chip scale LDO linear regulators. In debugging the first boards, the engineer needed to use a microscope to locate and probe contact points on the PCBA. But each time he did that, the LDO would fail. Damn that ESD!!!. After breaking way too many boards with his probing, he noticed that if he held his hand a certain way the LDO repaired itself. It turned out that the chip scale package was not opaque, and the bright light source under the microscope was causing the silicon in the LDO to misbehave. So, the gremlin was not ESD. The LDO was light sensitive. I am willing to bet with ever smaller packaging this problem will get worse. After all silicon is inherently photovoltaic.

01830
User Rank
Rookie
re: A memory problem not to make light of
01830   11/5/2010 8:21:39 PM
NO RATINGS
I had a situation with the opposite problem. I was called over to a photoshoot for a new product because the equipment was locking up or going berserk. I get over there and they had the board exposed with an EPROM. Someone had stuck a paper label on it, and when they took a flash photo, it would 'glitch' the system. I took one of the foil 'write protect' stickers from a floppy disk and put over the window and the system worked fine.

Tom VanCourt
User Rank
Rookie
re: A memory problem not to make light of
Tom VanCourt   11/5/2010 5:19:40 PM
NO RATINGS
I remember the UVPROM generation well. Like, when some of our PROMs stopped erasing. They had been working fine. We peeled the labels off the windows to reuse them and set them in the tanning booth for chips about five times longer than necessary. Still, enough old data remained to make the chips unusable. It turned out the stickers on the windows left glue residue behind, and that stuff must have been SPF 100. It kept the PROM perfectly safe from dangerous UV - until we washed it off with acetone. (BTW, only one brand of stickers caused the problem.)

Flash Poll
Like Us on Facebook

Datasheets.com Parts Search

185 million searchable parts
(please enter a part number or hit search to begin)
EE Times on Twitter
EE Times Twitter Feed
Top Comments of the Week