Mike and Greg,
Here is some interesting comparison data between the Fusion and
the Catalyst specifically as it relates to the specs referenced in
Mike's message:
Maximum Frequency:
The maximum unmuxed frequency of the FusionHF system is 500MHz. If
mux-mode is utilized, even higher frequencies can be achieved
probably limited by the pin electronics and minimum pulse widths. Fusion's
base operating mode frequency is 125MHz but the GaAs timing system
was designed to operate at 250MHz. Basically, all the ATE systems
out there have different modes that provide more flexibility and
formats at the lower frequency modes and less at the higher frequency
modes. Fusion has three modes: 125MHz, 250MHz, and 500MHz.
The interesting question is where the tradeoffs are made.
Mike?? I know the Catalyst has different timing modes with
specific tradeoffs as the frequency mode is increased. Is the
Catalyst a 100MHz base frequency system with modes to go to 200MHz
and 400MHz? If this is correct, what do you lose when going from
100MHz to 200MHz? 200MHz to 400MHz?
Differential Drive and Compare:
This is an area where the Fusion system truly excels over the
Catalyst. Yes, the Teradyne can drive differential signals but they
must use two channels to do it and more importantly they are driving
each of the channels with different timing markers. On the Fusion,
the pin card has modes where you can drive single ended or
differentially using the same timing markers. This eliminates the
skew issues associated with driving a differential signal. Driving
differential signals is really the simple problem, it's receiving
differential signals where the Fusion team really nailed it.
The Fusion
has true differential comparators with a second stage that can measure
the voltage differential between the differential signals. If you've
seen the demos on the Fusion they really drive this point home. This
can be a real problem on a system that has single ended comparators
with 2 channels being used to measure the differential output of the
device because you either can't handle a differential signal that is
offset relative to a static DC level or if you have only a differential comparator, you can't detect if one of the differential outputs has simply gone dead and is running at the common mode level.
Both these problems are solved with the Fusion architecture.
The FusionHF system can have as many as 1024 digital pins and they
don't give up digital pins when they add analog instruments as do
many of the other systems.
I'm not sure how many analog channels can be added to the Fusion.
The FusionHF testhead is the smallest in the industry and more
importantly it has the shortest roundtrip delay of any system out
there. This is crucial for high-speed signal fidelity.
Another area where the Fusion is superior is the APG per pin built
into the timing system. This is supposedly going to give them an
advantage when testing the embedded memory destined for SOC. I'm
not sure this is really that big of an advantage since most SOC
devices have very wide memory widths, i.e. 128 wide buses, and the
trend today is to test these only with simple march patterns in
volume production. This is to save test time. Another trend is to
integrate BIST solutions into the silicon to handle the memory test
problem. Today, these march patterns can be put in standard
vector memory eliminating the need for the APG per pin and the cost
associated with it. BIST will eliminate the need for APG altogether
in volume production.
Impressive specs, right? But can they build it? And will the
customers wait? And will LTX have enough people left to support it?
Mike I think you're right in that ATE stock is not the
place to be for the next six months.
-AL |
