[SI-LIST] Re: Jitter Characterization on a Tester?

  • From: "Ingraham, Andrew" <Andrew.Ingraham@xxxxxx>
  • To: <si-list@xxxxxxxxxxxxx>
  • Date: Wed, 2 Oct 2002 08:47:09 -0400

There are a few problems with IC production testing.

One is test time.  Jitter testing, and PLL testing, can be slow.  Every
additional second you spend testing an IC, is a second you could have
spent testing another IC.  If you take 5 minutes testing each and every
part, you'll never ship in volume.  So it's not uncommon for IC vendors
to sample-test some of those characteristics that take longer.  Again,
the hope is that the other tests you do on every part, will catch 99.99%
of problems; but the assumption is that most PLL or jitter problems,
also manifest themselves as something else measurable (like out-of-spec
delays or Iddq, etc.).

Off-hand, I would guess that test gear like the Wavecrest are used for
exhaustive tests done on random samples and not every part.

The other problem is at-speed testing.  Often, for various practical
reasons, the testing must be done at slower than normal or maximum
speeds.  As long as the test engineers have properly characterized the
devices so that the slow speed testing catches most speed related
problems, it may be deemed acceptable.  It's a trade-off.  As a simple
example, you may not really need to test a NOR gate that has 50ps
delays, and is capable of handling 2GHz, at 2GHz; when a single
non-repetitive delay test on a rising edge and a falling edge may
suffice (in addition to the usual DC tests).  But unforeseen problems
can happen that don't get caught, especially if they are limited to
narrow frequency ranges only.  What if your IC vendor tested his parts
at maximum and minimum frequencies, but it only has the jitter problem
at nominal?

But if the IC vendor's parts are primarily digital -- and, oh, by the
way, they also have a PLL in them -- they may not have considered
problems like this.  Their test engineers may not be well versed in
analog characterization.

I should also mention that a lot of ICs being bought and used, even by
big companies, are grey market ICs that failed their manufacturer's
testing.  Said parts were probably sold as scrap to be disposed, but the
folks they sold them to were unscrupulous and turned around and sold
them onto the grey market, which these days seems to include normal
distributors and resellers.  This is not a small problem.  You don't
know what you're getting unless you buy direct from the manufacturer.
(Someone on another list is giving away a batch of bad grey market ICs
that he bought through a distributor.)


To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field

or to administer your membership from a web page, go to:

For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field

List archives are viewable at:     
or at our remote archives:
Old (prior to June 6, 2001) list archives are viewable at:

Other related posts: