If it's not the wafer, what "new materials supplier" is there? Who's putting the gun to their head to change suppliers on this sole-sourced part.
Logically, continue making it with the old material until the new material gets approved.
Any quality/supply issues will cause big trouble for manufacturers dependent on it.
I'm skeptical. Mergers and acquisitions are never good for products.
AD will surely consolidate fab, remove redundant teams, "improve" things to lower costs. Increasing shareholder value is paramount and great art is never about that.
I'm sure that Agilent and Linear signed a contract for the availability, which is a bounding contract for Keys. and Analog.
I ordered a LTZ1000A 4 months ago directly from Linear's website.. still waiting.
a documentation the size of the holy book
...
I have been noticing thatsomea very tiny percentage of the most recent LTZs have a lot of 1/f noise and popcorn noise. This is related to contamination of the junction layers with metals, (that are normally extracted during processing by infusion of phosphorous on the back side of the wafer, followed by a long bake at 900oC) and/or radioactive particle contamination. I suspect that something in the fab has gone Wr0nG, and they are trying to fix it. I wonder if this is related to the recent contamination in the atmosphere in the northern hemisphere with radioactive contaminants. Maybe it's time to coat the chip with super-pure silicone RTV before sealing the package?
I'm going to shoot a question at LTC and see if they will cough up the information...
-Ken
I have been noticing thatsomea very tiny percentage of the most recent LTZs have a lot of 1/f noise and popcorn noise.
a documentation the size of the holy book
this forum would love to have access to *that* holy book
regards.
I have been noticing thatsomea very tiny percentage of the most recent LTZs have a lot of 1/f noise and popcorn noise.
hello Ken,
can you expound on "a lot of noise", ie; some measurement details? how far off-spec or in-spec?
best regards.
did something get switched on nearby?
I would expect the LTZ1000 to be tested for a little longer than 10 seconds. It already needs some time to get a stable temperature - so no way to do a super fast test anyway. Popcorn noise is likely one of the more important reasons to do the checks and sort out failing units.
However with popcorn noise is can be difficult to detect - especially that type of noise with more infrequent jumps. Chances are one would not notice this in a short interval of maybe 30 seconds. Those very low frequency popcorn cases are the really bad ones, as one never knows and longer averaging also does not help that much. With the more frequent and symmetric jumps, like in TIN's jummpy sample from Ebay one has in theory the chance to detect and even correct for this.
I still fail to see the big LTZ datasheet spec "Oops" in Andreas' data. That shows a part still within the realm of datasheet specs it looks to me: It's showing clearly a P-P value of ~2uV (equipment / experiment noise -floor- limits are not known) which is the datasheet value. Of course the longer you measure, the more likely you'll see more noise. Especially as 1/f creeps in. Double the measure time and you get another octave's worth of 1/f possibilities. If there is -some- popcorn noise, it will probably still be able to pass the LT quality check.
The problem we saw before on just one or two parts was an excursion more like 6 or 8uV P-P across 10 sec - clearly out of spec. Those got replaced.
DigilentKen: Were you thinking about a bigger "oops" deviation from datasheet value? That's what I was curious about. How many parts and what date codes are we talking about??
It already needs some time to get a stable temperature
RTFM