28075Re: vintage SRAMs self healing
- Nov 12, 2012Bill,
I'm seeing this also. At first I thought it was heat related but as Mike said it appears not. I know I am testing the chips individually using Mike's 2102 RAM tester setup for the Apple II SuperProto board on my II plus. I am pretty sure Mike is doing the same on his Apple II.
All the 2102 NOS chips I "tested" before installing worked great, but I have an IMS 8K SRAM card that was populated already and after tracking down two bad chips, I decided to start testing them all outside of the system and off the card using the Mike's test setup. That was when I noticed the problem as well. When I first got my MITS 4k card I had some bad ram also, I'm wondering if those chips would have healed if I tested them a few cycles before I threw them out.
--- In email@example.com, "B. Degnan" <billdeg@...> wrote:
> -------- Original Message --------
> > From: "Mike" <mike@...>
> > Sent: Monday, November 12, 2012 3:29 PM
> > To: firstname.lastname@example.org
> > Subject: [midatlanticretro] vintage SRAMs self healing
> > Hi,
> > I have been testing some vintage Fairchild 2102 SRAMs and have found that
> a small percentage of chips that initially fail a pretty comprehensive
> memory test. However after several reruns of the test, sometimes over
> several sessions, they all eventually start to pass, and can no longer be
> made to fail. This is not due to temperature, since in some cases the
> parts are left for days between test runs. It appears that usage alone is
> what what causes the chips to "heal". Note that the timing margins of the
> test setup are extremely generous, so I doubt that the errors would be
> associated with access speed, changing. Signal quality, power and ground
> look very good.
> > I'm not all that familiar with old SRAM technology, any ideas on what may
> be going on with these chips?
> > regards,
> > Mike W.
> My guess is that the chips are OK, it's the socket or contacts of the chips
> in the socket. The heat loosens up the crud and then until you remove the
> socket they'll be OK. If you deOxit the chips first, you may have a better
> initial success rate?
- << Previous post in topic Next post in topic >>