Why do good 4027, 4096 or UPD414 DRAM keep failing?

Last updated: April 28th, 2017

Known good 4027, 4096, UPD414 or equivalent DRAM will pass the first time you test it and then fail on subsequent tests. I've been investigating this issue for some time now and have discovered the following:

There was a bug in V1 code that would clear the bad ram detect flag in the transition from pass 2 to pass 3. Some DRAM would only fail on test 1 or test 2, and pass on the rest and since the flag was being cleared, the net effect is that the DRAM would pass even though it was bad. This bug was fixed in V2.

Additional testing revealed that DRAM would still fail on the short test, but pass on the long test. In ivestigating this issue I discovered that waiting three to five seconds before testing the DRAM again sometimes resolved the issue, my guess is that the DRAM is in a slow power down sequence, a residual charge, and powering it up to soon causes the DRAM to fail the test.

I was ultimately unable to find a firmware solution to this problem so in version 2.1 I have removed the short test capability all together from 4027, 4096 and UPD414.

For the various version of the 4116 blade the following is recommended:

For V 1.x:

1. Don't use the short test at all and stick to the long test.
2. Allow 5 seconds in between single and multiple scans with the 4027, 2096 and UPD414.

For V 2.0

1. Don't use the short test with multiple scans selected, only the long test should be utilized when you have multiple scans configured. On the short test, the DRAM under test may fail after the first scan.
2. Allow 5 seconds in between single and multiple scans with the 4027, 2096 and UPD414.

For V2.1

1. Version 2.1 implements a change so that the IC under test is NOT powered down in between scans.
2. The short test has been removed from the 4027, 4096 and UPD414, only the long test is available on these variatns.

 I am still investigating this issue and will post update in this help article.