home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.ibm.pc.hardware
- Path: sparky!uunet!munnari.oz.au!spool.mu.edu!torn!nott!bnrgate!bmers95!bnr.ca!khor
- From: khor@bnr.ca (Kyle Hor)
- Subject: Re: What's the deal? My chip says "SX-25"; Norton says "SX-33"
- Message-ID: <1993Jan5.150757.9425@bmers95.bnr.ca>
- Sender: usenet@bmers95.bnr.ca
- Reply-To: khor@bnr.ca
- Organization: Bell-Northern Research Ltd.
- References: <5870140@pollux.svale.hp.com> <C01LuI.542@news.cso.uiuc.edu> <sheldon.725671411@pv141b.vincent.iastate.edu> <C0Co70.1ID@usenet.ucs.indiana.edu>
- Date: Tue, 5 Jan 1993 15:07:57 GMT
- Lines: 46
-
- In article <C0Co70.1ID@usenet.ucs.indiana.edu>, ntaib@silver.ucs.indiana.edu (Iskandar Taib) writes:
- |> In article <sheldon.725671411@pv141b.vincent.iastate.edu> sheldon@iastate.edu (Steve Sheldon) writes:
- |> > But since these are targeted for commercial usage, they're most likely
- |> >only testing 1 in every 100 or so, as Mike said, a few out of each batch.
- |>
- |> > Chances are also good that if any of the sample batch fails the 25 Mhz, and
- |> >depending on the percent failed, they'll test all chips in that batch to see
- |> >if any of them are salvageable before they go throwing them all out the
- |> >door. Of course this last one depends on the economics of how expensive it
- |> >is to test all the chips vs. just making new ones.
- |>
- I beg your pardon. Ever since the Japanese IC 'revolution', *ALL* IC
- chips (no matter their size, complexity or commercial grade) are
- tested before they are shipped. The tests (electrical) may vary in
- completeness and stringent-ness (sp?) but the tests are performed.
-
- |> Just wondering: at what point do they test the chips, and how? Do they
- |> do it while at the wafer stage, before they encase it in ceramic? How
- |> are the connections made?
- |>
- Tests are performed at various stages to reduce the cost of buidling
- an IC that is bad in the 1st place. Many of these are sampling test.
- But at final test 100% of the chips are tested. And it is usually
- during this stage that grading is done.
-
- The normal practice is to test against the highest grade, lets say 50MHz.
- If it fails that, then it will be downgraded to the next and so on. If
- it fails the lowest grade, 20MHz (?) then it will be scraped.
-
- For a 33MHz grade, the chip is usually tested to pass at, say 35MHz.
- The industry calls this the guard band. Another factor, is the
- market demand at the point in time. If the manufacturer happen to
- come out with a large batch of 33MHz chips but there is great demand
- for 25MHz chips, they will simply re-label these to 25MHz and ship it.
- This is referred to as downgrading. (Somewhat confusing as there are
- real downgrades during final electrical test).
-
- Anyway I do not represent any IC manufacturer. Enuf said.
-
- ............................................................................
- . Kyle Hor . / / _ . Email khor@bnr.ca .
- . Bell-Northern Research Ltd. . /__\ / / /_\ . Phone (613) 763-2488 .
- . Ottawa, Ontario, Canada . / \ \/ / \__ . ESN 393-2488 .
- ....................................__/.....................................
- ................ standard disclaimer, 2 cents opinion ....................
- ............. searching forever, for the perfect solution ................
-