[sdiy] On the Internet, talk is cheap

Paul Schreiber synth1 at airmail.net
Thu Feb 7 01:44:18 CET 2013


[about the SSM2164 and any other IC that is going away]

a) you would figure CEM ICs would be a hot commodity in 1997-98, since only
a few people had any (Dave Longo......cough). When I struck the deal with
Doug Curtis for his NOS, I announced it, set up a website (which was VERY
expensive back then), spammed everyone. I asked people that  *IF THEY WERE
TRULY INTERESTED* in an *IMMEDIATE PURCHASE*, please send me an email.

So, the day before the chips arrived, I had ~ 800 emails.

Starting interest = 800 emails
Bulls**t factor = 0.5

So say 400 emails were what I would consider 'serious' and then from those
maybe 300 were 'immediate' and the others would be 'when I deliver a few
more pizzas'.

I bought like 400 Uline boxes, bags, ESD labels, Avery labels and foam. I
formally announced and waited.

The incoming orders went something like this:

Day 1: 2
Day2: 0
Day3 :1
Day4: 0
Day5: 0
Day6 : 4 >>>got those pizzas out the door
........

Day 30: at total of about 14 orders. In a MONTH. It took me *3 years* to
*break even*. And this was for (supposedly) the hottest, hardest to find ICs
on the planet.

b) the analog chips of yore all on high-voltage (meaning 40V) process. Last
time I checked, the iCrap doesn't run on +-15V. In fact, ever since Maxim
invented the MAX232 RS-232 converter with the built-in charge pumps, there
hasn't been a negative supply used in any PC and that was like 1988. Also,
these 40V chips need a LOT of masks (like 50) versus the generic 0.18u TSMC
digital parts (like 22) and remember, you pay *per wafer*, makes NO
DIFFERENCE what the hell is on it or how many. Each technology has a fixed
wafer cost. The typical MPU today on 0.18u has a die cost of probably 15
cents. The 6u 40V die for an op amp is probably 4 cents. But if I am
Qualcomm I can charge HTC $17 for a "baseband processor" in the cell phone,
but if I am TI or ADI what the hell can I get for an op amp? 12 cents?

Like was said before: it's NOT THE CHIP that gets obsoleted, it's the
PROCESS FOR THE WAFERS. If the wafers can't attract business, then the
wafers dry up and therefore the ICs associated with that wafer go away. I
used to work for a military electronic company that used a special high-temp
(125C operating) analog switch on a 12u process (100 times larger than
modern stuff) and that process went tits-up in like 1994. But they did a
last-time buy of like 100,000 die (which basically fits in 2 4-drawer file
cabinets) and that was a *100 year supply*. In the IC industry, anytime you
see 'last-time buy' that means you are on the die reserve, that process is
LONG GONE.

Paul S.




More information about the Synth-diy mailing list