| Author |
Message |
John Larkin
Guest
|
 Posted:
Fri Feb 18, 2005 2:19 am Post subject:
Re: Your favorite 10 analog IC's |
|
|
On Thu, 17 Feb 2005 19:02:13 GMT, gwhite <gwhite@deadend.com> wrote:
| Quote: | John Larkin wrote:
On Wed, 16 Feb 2005 09:45:16 -0700, Jim Thompson
thegreatone@example.com> wrote:
If you've never been fired for insubordination, you've not lived. I
got fired from Dickson Electronics (a hybrid house) for redirecting
their business plan from heavily military to a mix between military
and commercial... showing a healthy profit in the process. They went
bust about two years after they fired me and sold out to Siemens. I
think that's even gone now.
I've always taken a peverse pleasure in dancing on the grave of a
company that tooled one around or ignored one's genius. I applaud the
deaths of Metricom,...
I worked there. Why do you say that regarding them? I have my own views of why
they failed.
|
What did you do?
Well, they were originally in the remote-read electric meter business.
I have some semi-cool technology that can do precise metering very
cheaply. I met with the guru Baran and - many times - with the prez,
Dilworth, and the engineers. They had serious doubts that my thing
actually worked - some trig-ignorant guys said flat-out that it
couldn't - so I did some semi-elaborate demos and customized my uP
code for their weird tests. It all worked. I proposed licensing my
stuff to them for 10% of what it would save them and they said, "No,
but how would you like to be a consultant?" Pricks.
They actually never wanted to be in the meter business; all the
engineers were Ham radio guys and seemed to think that measuring 60 Hz
was trivial and annoying. Plus the usual NIH of anybody outside having
a better idea.
The meter they designed was an expensive crock, and nobody would buy
it, so they became a Wireless Internet Company (which the hams
preferred) and burned through a bunch of presidents and hundreds of
megabucks and died.
Funny, I have a friend who had a very similar experience with Cellnet,
or Smellnet as he calls them.
I don't know any stories about Metricom after they switched to the
Ricochet thing; any good gossip about what they did wrong?
I think I actually saw a guy in a cafe using a laptop with a Ricochet.
Once.
John
|
|
| Back to top |
|
 |
John Larkin
Guest
|
| Posted:
Fri Feb 18, 2005 2:26 am Post subject:
Re: Your favorite 10 analog IC's |
|
|
On Thu, 17 Feb 2005 17:22:54 GMT, Bob Stephens
<stephensyomamadigital@earthlink.net> wrote:
| Quote: | On Thu, 17 Feb 2005 09:06:29 -0800, John Larkin wrote:
We just started soldering our own BGAs, some 456-pin, 1 mm pitch
Xilinx parts, $105 each. Everybody was fairly terrified, and surprised
that it just works
I'm sure it doesn't 'just work'.
|
Gosh, they all seem to work to us.
| Quote: | What technique do you use?
|
Just solder stencil, place, run through the oven. Apparently the only
tricky parts are precise placement and the right temperature profile.
We're investigating some expensive optics for inspection, and have a
weird idea for electrical testing, but so far all the boards have
worked.
| Quote: | Scares me and I'm fearless.
|
People who do this full-time say it's easier than fine-pitch TQFPs or
whatever.
John |
|
| Back to top |
|
|
gwhite
Guest
|
| Posted:
Fri Feb 18, 2005 4:03 am Post subject:
Re: Your favorite 10 analog IC's |
|
|
John Larkin wrote:
| Quote: |
On Thu, 17 Feb 2005 19:02:13 GMT, gwhite <gwhite@deadend.com> wrote:
John Larkin wrote:
On Wed, 16 Feb 2005 09:45:16 -0700, Jim Thompson
thegreatone@example.com> wrote:
If you've never been fired for insubordination, you've not lived. I
got fired from Dickson Electronics (a hybrid house) for redirecting
their business plan from heavily military to a mix between military
and commercial... showing a healthy profit in the process. They went
bust about two years after they fired me and sold out to Siemens. I
think that's even gone now.
I've always taken a peverse pleasure in dancing on the grave of a
company that tooled one around or ignored one's genius. I applaud the
deaths of Metricom,...
|
Mostly it was average folks who worked hard that got whacked.
| Quote: | I worked there. Why do you say that regarding them? I have my own views of why
they failed.
What did you do?
|
I was a radio engineer.
| Quote: | Well, they were originally in the remote-read electric meter business.
I have some semi-cool technology that can do precise metering very
cheaply.
|
That is where the name sourced.
Meter Communications == Metricom
| Quote: | I met with the guru Baran and - many times - with the prez,
Dilworth, and the engineers. They had serious doubts that my thing
actually worked - some trig-ignorant guys ...
|
That is funny.
| Quote: | ...said flat-out that it
couldn't - so I did some semi-elaborate demos and customized my uP
code for their weird tests. It all worked. I proposed licensing my
stuff to them for 10% of what it would save them and they said, "No,
but how would you like to be a consultant?" Pricks.
They actually never wanted to be in the meter business; all the
engineers were Ham radio guys and seemed to think that measuring 60 Hz
was trivial and annoying. Plus the usual NIH of anybody outside having
a better idea.
|
The last of the Hams were gone not too long after I got there (early 1998).
| Quote: | The meter they designed was an expensive crock, and nobody would buy
it, so they became a Wireless Internet Company (which the hams
preferred) and burned through a bunch of presidents and hundreds of
megabucks and died.
|
It was Dilworth's idea.
| Quote: | Funny, I have a friend who had a very similar experience with Cellnet,
or Smellnet as he calls them.
I don't know any stories about Metricom after they switched to the
Ricochet thing; any good gossip about what they did wrong?
|
The short story is that they shot themselves in the foot with an impossible
nationwide buildout plan for Ricochet 2 -- they ran out of money.
The major investors, Paul Allen and MCI (to the tune of 700 to 800 $mil,
depending upon how you count it) insisted on a nationwide buildout timeline that
was essentially unrealistic. That is, it appeared to me that they would only
make the investment contingent upon Metricom buying into a rapid buildout. The
aggressive buildout timeline--built into the investment contract--effectively
forced Metricom to burn through tons of cash rapidly. They were making a near
attempt to "light up" the entire nation instantaneously.
One of the major problems with this is they never gave themselves time to figure
how to market Ricochet or how to price it. Again, they ran out of cash just as
they were beginning to figure it out.
Also, they spent way too much on network hardware. I don't think Network Ops
had even 50% of the network hardware installed (much less up and running) by
crash time. When Aerie Networks bought the non-installed hardware and all IP
(for $8.25 mil) they trucked it all to Oklahoma in over 500 semi-trucks. Talk
about excess inventory. The system hardware that was produced en masse never
made it to the so-called "cost cutting phase" of design development because of
the "build out quickly nationwide at any cost" mentality. Put concisely, this
made the hardware cost at least double what it should have been, again draining
money like a sieve.
The bottom line: The directive was to take a small 200 person company in Los
Gatos and build a $1 bil (or more) advanced radio communication system
nationwide in a period of about 1.5 years (maybe 2), and all the marketing and
revenue generating that must necessarily rapidly follow. No one can grow that
fast and be successful for any realistic amount of investment. The "plan" was
crazy and flawed at its root, despite certain executive management problems that
did exist during the massive network inventory over production.
I think this ended up being about a $1.5 bil train wreck.
| Quote: |
I think I actually saw a guy in a cafe using a laptop with a Ricochet.
Once.
|
It actually worked pretty well. A good technology got snuffed out by a crazy
plan.
|
|
| Back to top |
|
|
John Larkin
Guest
|
| Posted:
Fri Feb 18, 2005 4:28 am Post subject:
Re: Your favorite 10 analog IC's |
|
|
On Thu, 17 Feb 2005 22:03:01 GMT, gwhite <gwhite@deadend.com> wrote:
| Quote: | The bottom line: The directive was to take a small 200 person company in Los
Gatos and build a $1 bil (or more) advanced radio communication system
nationwide in a period of about 1.5 years (maybe 2), and all the marketing and
revenue generating that must necessarily rapidly follow. No one can grow that
fast and be successful for any realistic amount of investment. The "plan" was
crazy and flawed at its root, despite certain executive management problems that
did exist during the massive network inventory over production.
I think this ended up being about a $1.5 bil train wreck.
I think I actually saw a guy in a cafe using a laptop with a Ricochet.
Once.
It actually worked pretty well. A good technology got snuffed out by a crazy
plan.
|
Cool; I love biz stories.
But is/was there an actual money-making market for wireless internet
connections? This neighborhood is full of WiFi cafes and all that, and
even when it's free nobody is sitting and surfing, as far as I can
see. People talk or read books or read newspapers in public places. I
think the idea of surfing the net 24 hours a day is over.
John |
|
| Back to top |
|
|
Jim Thompson
Guest
|
| Posted:
Fri Feb 18, 2005 4:46 am Post subject:
Re: Your favorite 10 analog IC's |
|
|
On Thu, 17 Feb 2005 14:28:26 -0800, John Larkin
<jjlarkin@highSNIPlandTHIStechPLEASEnology.com> wrote:
[snip]
| Quote: |
But is/was there an actual money-making market for wireless internet
connections? This neighborhood is full of WiFi cafes and all that, and
even when it's free nobody is sitting and surfing, as far as I can
see. People talk or read books or read newspapers in public places. I
think the idea of surfing the net 24 hours a day is over.
John
|
Yep, Who the hell wants to "surf"?
Anymore, I rarely take a laptop with me when I travel.
Instead I bought _myself_ an IPOD Mini when I bought a bunch of them
for the grandkids at Christmas... loaded it up with "The Da Vinci
Code", and had a relaxing trip last week to Pittsburgh ;-)
...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona Voice:(480)460-2350 | |
| E-mail Address at Website Fax:(480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |
I love to cook with wine. Sometimes I even put it in the food. |
|
| Back to top |
|
|
Tim Shoppa
Guest
|
| Posted:
Fri Feb 18, 2005 6:11 am Post subject:
Re: Your favorite 10 analog IC's |
|
|
| Quote: | I don't know why the 3046 went
|
Well, the CA3046 went, but Intersil still makes NPN transistor arrays
of higher performance. The HFA3046, for example, is in the SOIC
package and the transistors have a ft of 8 GHz. Costs 20 times as much
as a CA3046!
Tim. |
|
| Back to top |
|
|
John Larkin
Guest
|
| Posted:
Fri Feb 18, 2005 11:25 pm Post subject:
Re: Your favorite 10 analog IC's |
|
|
On 17 Feb 2005 17:37:57 -0800, "Tim Shoppa" <shoppa@trailing-edge.com>
wrote:
| Quote: | I don't know why the 3046 went
Well, the CA3046 went, but Intersil still makes NPN transistor arrays
of higher performance. The HFA3046, for example, is in the SOIC
package and the transistors have a ft of 8 GHz. Costs 20 times as much
as a CA3046!
Tim.
|
NEC has some fast arrays. Too fast, often.
John |
|
| Back to top |
|
|
Bob Stephens
Guest
|
| Posted:
Fri Feb 18, 2005 11:39 pm Post subject:
Re: Your favorite 10 analog IC's |
|
|
On Thu, 17 Feb 2005 12:26:39 -0800, John Larkin wrote:
| Quote: | I'm sure it doesn't 'just work'.
Gosh, they all seem to work to us.
|
I wasn't being snotty. What I meant was, it must be non-trivial.
Bob |
|
| Back to top |
|
|
gwhite
Guest
|
| Posted:
Sat Feb 19, 2005 12:26 am Post subject:
Re: Your favorite 10 analog IC's |
|
|
John Larkin wrote:
| Quote: |
On Thu, 17 Feb 2005 22:03:01 GMT, gwhite <gwhite@deadend.com> wrote:
The bottom line: The directive was to take a small 200 person company in Los
Gatos and build a $1 bil (or more) advanced radio communication system
nationwide in a period of about 1.5 years (maybe 2), and all the marketing and
revenue generating that must necessarily rapidly follow. No one can grow that
fast and be successful for any realistic amount of investment. The "plan" was
crazy and flawed at its root, despite certain executive management problems that
did exist during the massive network inventory over production.
I think this ended up being about a $1.5 bil train wreck.
I think I actually saw a guy in a cafe using a laptop with a Ricochet.
Once.
It actually worked pretty well. A good technology got snuffed out by a crazy
plan.
Cool; I love biz stories.
But is/was there an actual money-making market for wireless internet
connections? This neighborhood is full of WiFi cafes and all that, and
even when it's free nobody is sitting and surfing, as far as I can
see. People talk or read books or read newspapers in public places. I
think the idea of surfing the net 24 hours a day is over.
|
The intial idea was not "surfing," although it could obviously be used for
that. At first, they thought the main customers would be businesses. It was to
be nationwide network (really covering all major metro areas, but not rural)
where a subscriber could connect anywhere anytime in the same manner.
For example, suppose a Sales Rep from San Diego was giving a presentation to a
potential customer in a conference room in Boston. The customer has a question
for which the Sales Rep has no local (Boston) information. The Sales Rep,
through VPN, can access the San Diego company servers just as if he/she is in
the San Diego office. The question is answered rapidly. Emails come right
through too, in the normal sense. There is no perceptual difference whether the
user is in San Diego, Boston, or Chicago O'Hare. The user keeps the computer on
standby, finds a moment to open it, and connects in under 10 seconds, the same
way everywhere. The same goes for when the business person goes to their hotel
room for the evening. They are completely insulated from any particular
connection issues of diverse hotels/motels. They can do the same thing for the
next day's sales visit.
That was an example of the mobile business user idea. Many business people
who--by the nature of their business--were on the road in a *single* metro area
absolutely loved the service. It was a very efficient comm link for them, which
they conveniently could move to the home office (or anywhere else) for the
evening, if desired.
Of course, part of the problem was the singular focus on business users. Many
non-business (consumer) users don't have access to DSL or Cable connections. Of
course, these areas are shrinking over time, but nonetheless there was and is
potential revenue from non-business users. The Ricochet connection allows a
home phone line to be dedicated to voice or fax. The idea that the home
consumer user is not physically tethered to a room or even a single neighborhood
could still have some value add even for folks that may not needing mobility
*often*.
Suppose a student shares a house with 7 others and share one phone line.
Ricochet allows independent rapid connectivity for the student, whether at home
or studying in the library at school with their laptop. Police and fire could
use data service too. In fact, *any* sort of data services could be supported.
It is up to the imagination of the *users*. The subscriber device simply needs
to be cheap.
Metricom Marketing had too narrow a focus imo. They actually began figuring out
what it took to break consumer resistance, and how to market Ricochet with some
test programs in the San Diego market. But it was too late--way too much debt,
not enough current revenues, and shrinking cash reserves caused creditors to put
the hammer down.
I also forgot to mention another practical problem with the nationwide plan of
building out the network. They tried to do it all over the country almost
simultaneously. There was not enough network building experience base for
Ricochet 2. Like any complicated process, it is nice to find all the process
hitches in a "test run" (smaller scale) before taking it to higher volume. That
is, I believe they should have built out the SF Bay Area system first because
this would help refine process issues and build personel experience base.
(Metricom was based in the SF Bay Area, so this seems geographically obvious.)
The technical network personel base can then take these learned process
improvements, and their practical learned experience, and diffuse that to
multiple metro areas as appropriate.
As it was, all mistakes were made everywhere, thereby multiplying the cost of
problems. The same goes for marketing, they needed to figure it out in the SF
Bay Area first, before incurring the error multiple of cost. They gave
themselves no chance of success. Executive staff, major investors, and the BoD
were in way over their heads, and it showed.
It was a rather spectacular wreck. To answer your question, I don't know if the
idea could have been ultimately successful. The users who did have it really
seemed to like it. I do know it was never given a reasonable (fair) chance of
discovery due to boneheaded unrealistic planning. If it were destined to fail,
it did not need to fail on that big of a scale. |
|
| Back to top |
|
|
John Larkin
Guest
|
| Posted:
Sat Feb 19, 2005 12:47 am Post subject:
Re: Your favorite 10 analog IC's |
|
|
On Fri, 18 Feb 2005 17:39:32 GMT, Bob Stephens
<stephensyomamadigital@earthlink.net> wrote:
| Quote: | On Thu, 17 Feb 2005 12:26:39 -0800, John Larkin wrote:
I'm sure it doesn't 'just work'.
Gosh, they all seem to work to us.
I wasn't being snotty. What I meant was, it must be non-trivial.
|
Oh, that's OK. I think it turns out to be easier than it looks. We
sent the first few boards out to a pro; we loaded everything but the
BGA, and he added it and xrayed. He told us how he does these add-ons:
smear flux on the pads *with your bare finger*, don't add any solder
paste, place, and heat from above and below.
He claims BGA is easier than regular fine-pitch stuff. He came over to
our place and showed us how to do the proper stencil/place/reflow way,
all the parts at once, and helped us set up our oven profile.
Anyhow, now we can do BGA protos and short runs, although we'll still
send out bigger batches to contract assemblers who have the serious
placement/soldering/inspection gear.
We've just started doing the horrible tiny QFN parts, too. This is a
great equalizer for older guys' eyesight: *nobody* can see this stuff.
John |
|
| Back to top |
|
|
John Woodgate
Guest
|
| Posted:
Sat Feb 19, 2005 1:20 am Post subject:
Re: Your favorite 10 analog IC's |
|
|
I read in sci.electronics.design that gwhite <gwhite@deadend.com> wrote
(in <42163368.8BDCBF76@deadend.com>) about 'Your favorite 10 analog
IC's', on Fri, 18 Feb 2005:
| Quote: | That is, I believe they should have
built out the SF Bay Area system first because this would help refine
process issues and build personel experience base. (Metricom was based
in the SF Bay Area, so this seems geographically obvious.) The technical
network personel base can then take these learned process improvements,
and their practical learned experience, and diffuse that to multiple
metro areas as appropriate.
|
Second phase: change the delighted-user name in the ads to 'Rick O'Shea'
and launch in Boston and New York. (;-)
--
Regards, John Woodgate, OOO - Own Opinions Only.
The good news is that nothing is compulsory.
The bad news is that everything is prohibited.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk |
|
| Back to top |
|
|
Robert
Guest
|
| Posted:
Sat Feb 19, 2005 4:11 am Post subject:
Re: Your favorite 10 analog IC's |
|
|
"John Larkin" <jjlarkin@highSNIPlandTHIStechPLEASEnology.com> wrote in
message
[snip]
| Quote: | Well, the CA3046 went, but Intersil still makes NPN transistor arrays
of higher performance. The HFA3046, for example, is in the SOIC
package and the transistors have a ft of 8 GHz. Costs 20 times as much
as a CA3046!
Tim.
NEC has some fast arrays. Too fast, often.
John
|
Maxim has a couple of precision Bipolar Arrays (QuickChip on 27GHz GST-2
process) they got when they bought the old IC manufacturing line from
Tektronix.
http://www.maxim-ic.com.cn/products/asics/wireless/design_methods.cfm
Robert |
|
| Back to top |
|
|
Jim Thompson
Guest
|
| Posted:
Sat Feb 19, 2005 4:25 am Post subject:
Re: Your favorite 10 analog IC's |
|
|
On Fri, 18 Feb 2005 14:25:56 -0800, John Larkin
<jjlarkin@highSNIPlandTHIStechPLEASEnology.com> wrote:
[snip]
| Quote: |
I was thinking more along the lines of...
http://www.cel.com/pdf/datasheets/upa807t.pdf
These are apparently two discrete transistors in the package. They
peak at over 12 GHz Ft, so are sometimes (usually) twitchy.
John
|
ONLY 12 GHz fT? I've recently been working with 35 GHz fT NPN's ;-)
...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona Voice:(480)460-2350 | |
| E-mail Address at Website Fax:(480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |
I love to cook with wine. Sometimes I even put it in the food. |
|
| Back to top |
|
|
John Larkin
Guest
|
| Posted:
Sat Feb 19, 2005 4:25 am Post subject:
Re: Your favorite 10 analog IC's |
|
|
On Fri, 18 Feb 2005 22:11:08 GMT, "Robert" <Robert@yahoo.com> wrote:
| Quote: |
"John Larkin" <jjlarkin@highSNIPlandTHIStechPLEASEnology.com> wrote in
message
[snip]
Well, the CA3046 went, but Intersil still makes NPN transistor arrays
of higher performance. The HFA3046, for example, is in the SOIC
package and the transistors have a ft of 8 GHz. Costs 20 times as much
as a CA3046!
Tim.
NEC has some fast arrays. Too fast, often.
John
Maxim has a couple of precision Bipolar Arrays (QuickChip on 27GHz GST-2
process) they got when they bought the old IC manufacturing line from
Tektronix.
http://www.maxim-ic.com.cn/products/asics/wireless/design_methods.cfm
Robert
|
I was thinking more along the lines of...
http://www.cel.com/pdf/datasheets/upa807t.pdf
These are apparently two discrete transistors in the package. They
peak at over 12 GHz Ft, so are sometimes (usually) twitchy.
John |
|
| Back to top |
|
|
Nico Coesel
Guest
|
| Posted:
Sat Feb 19, 2005 5:04 am Post subject:
Re: Your favorite 10 analog IC's |
|
|
Rich Grise <richgrise@example.net> wrote:
| Quote: | On Wed, 16 Feb 2005 12:03:09 -0500, Spehro Pefhany wrote:
On Wed, 16 Feb 2005 08:45:47 -0800, the renowned John Larkin
On Wed, 16 Feb 2005 12:12:38 GMT, Fred Bloggs <nospam@nospam.com
Ken Smith wrote:
The "zero power" version on the 22V10 like TICPAL22V10Z is handy in
battery powered stuff.
Now *that* is a good digital catch-all.
The Xilinx Coolrunner flash CPLDs (and similar) are the next gen of
this sort of thing. The smallest ones are something like 4x the logic
of a a 22V10, have a more general architecture, and cost something
like $1.20.
That's a very reasonable price for many applications.
They are surface-mount and have to be programmed on-board,
which can be a small nuisance.
Why do they have to be programmed on board? Can't you get a <$200
programming adapter for the little guys?
Maybe because it's surface-mount? I've never heard of a socket for a
surface-mount part. :-)
|
They exist. Mostly to program and/or copy-protect microcontrollers
prior to assembly.
--
Reply to nico@nctdevpuntnl (punt=.)
Bedrijven en winkels vindt U op www.adresboekje.nl |
|
| Back to top |
|
|
|
|
|
|
FAQ
Memberlist
Register
Profile
Log in to check your private messages
Log in