ethernet repeater chip?

[James Meyer]

I need a chip to build a 4-port, unmanaged, 10/100 ethernet repeater
hub. AMD has one for 10 base-T only and Intel has one for 10/100 but the Intel
chip is twice as big and more complicated than I need for the application.

I need something in a small package and I don't need any of the extra
ports or functions that sometimes go with switches or stacking hubs.

Jim

 

[dont know]

James Meyer wrote:

I need a chip to build a 4-port, unmanaged, 10/100 ethernet repeater
hub. AMD has one for 10 base-T only and Intel has one for 10/100 but
the Intel chip is twice as big and more complicated than I need for
the application.

I need something in a small package and I don't need any of the extra
ports or functions that sometimes go with switches or stacking hubs.

Jim

new 10/100 5-port hubs sell for 50-60 bucks. is it worth it to design
your own?

 

[Tim Auton]

"dont know" <mojo@nospam_netscape.com> wrote:

James Meyer wrote:

I need a chip to build a 4-port, unmanaged, 10/100 ethernet repeater
hub.
[snip]
new 10/100 5-port hubs sell for 50-60 bucks. is it worth it to design
your own?

$50-60? I bought a 5-port 10/100 *switch* for ~$40 a few weeks ago,
and that was at a real shop not renowned for low prices. I'm sure you
could find a hub for ~$20.


Tim
--
Love is a travelator.

 

[Jeff Liebermann]

On Thu, 01 Apr 2004 13:06:31 GMT, James Meyer <jmeyer@nowhere.com>
wrote:

I need a chip to build a 4-port, unmanaged, 10/100 ethernet repeater
hub. AMD has one for 10 base-T only and Intel has one for 10/100 but the Intel
chip is twice as big and more complicated than I need for the application.

I need something in a small package and I don't need any of the extra
ports or functions that sometimes go with switches or stacking hubs.

Jim

A "repeater" is a hub, What goes into one port, gets repeated out the
others. No intelligence involved.

A "switch" is aware of MAC addresses, which allow it to only send
traffic to the the destination port, thus cutting down drastically on
traffic. Broadcasts and multicast go to all ports.

A "bridge" is just a 2 port switch.

A "range extender" is a two port repeater. It may include bit
regeneration and perhaps a store and forward buffer. It's usually in
the form of a bridge (MAC address aware), but can also be a
non-intelligent cable extender or media converter. Lots of variations
here...

See:
http://www.broadcom.com/products/technology.php?technology_id=15
for chips.

If I had a clue as to how small [numbers], what features you need
[insert itemized list], and what you're doing [insert description], I
might be able to be more specific.


--
Jeff Liebermann 150 Felker St #D Santa Cruz CA 95060
(831)421-6491 pgr (831)336-2558 home
http://www.LearnByDestroying.com AE6KS
jeffl@comix.santa-cruz.ca.us jeffl@cruzio.com

 

[John Woodgate]

I read in sci.electronics.design that Tim Auton <tim.auton@uton.[groupSe
xWithoutTheY]> wrote (in <nfio60tcbegi56kodo6pah2t0rdlvu7qqi@4ax.com&gt
about 'ethernet repeater chip?', on Thu, 1 Apr 2004:

Love is a travelator.

I've resisted asking this since you first posted it. What do you have in
mind?

If you mean the travelator at Bank station in London, I could perhaps
understand the analogy - it's slippery, slow and often breaks down, but
you can take steps to bypass it. (;-)
--
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

 

[Tim Auton]

John Woodgate <jmw@jmwa.demon.contraspam.yuk> wrote:

I read in sci.electronics.design that Tim Auton <tim.auton@uton.[groupSe
xWithoutTheY]> wrote (in <nfio60tcbegi56kodo6pah2t0rdlvu7qqi@4ax.com&gt
about 'ethernet repeater chip?', on Thu, 1 Apr 2004:

Love is a travelator.

I've resisted asking this since you first posted it. What do you have in
mind?

It would appear you have given considerably more thought to my
signature than I have. This is not a good idea. Perhaps it is a
comment from my subconscious on the emotional numbness that depression
brings.


Tim
--
Love is a travelator.

 

[James Meyer]

On Thu, 01 Apr 2004 16:06:37 GMT, "dont know" <mojo@nospam_netscape.com> posted
this:

James Meyer wrote:


I need a chip to build a 4-port, unmanaged, 10/100 ethernet repeater
hub. AMD has one for 10 base-T only and Intel has one for 10/100 but
the Intel chip is twice as big and more complicated than I need for
the application.

I need something in a small package and I don't need any of the extra
ports or functions that sometimes go with switches or stacking hubs.

Jim

new 10/100 5-port hubs sell for 50-60 bucks. is it worth it to design
your own?

When it is part of an embedded system with very tight packaging volume
specs, it is.

Jim

 

[James Meyer]

On Thu, 01 Apr 2004 09:59:57 -0800, Jeff Liebermann
<jeffl@comix.santa-cruz.ca.us> posted this:

On Thu, 01 Apr 2004 13:06:31 GMT, James Meyer <jmeyer@nowhere.com
wrote:

I need a chip to build a 4-port, unmanaged, 10/100 ethernet repeater
hub. AMD has one for 10 base-T only and Intel has one for 10/100 but the Intel
chip is twice as big and more complicated than I need for the application.

I need something in a small package and I don't need any of the extra
ports or functions that sometimes go with switches or stacking hubs.

Jim

If I had a clue as to how small [numbers], what features you need
[insert itemized list], and what you're doing [insert description], I
might be able to be more specific.

If my project manager(mangler) would let me know, I could be more
specific.

So far I've found out that a 100-pin quad flat-pack plus the magnetics
would fit.

It needs to be able come up running without any supervision or
programming even from an EEPROM.

It needs industrial temp specs.

It should be low power.

It should be off-the-shelf in small quantities.

We're building an autonomous underwater vehicle with two ethernet
capable PC-104 computers, an RF modem, and an ethernet equiped microcontroller
to supervise things. They all need access to each other with a fairly high data
rate. The system will probably need to be reconfigured on-the-fly from time to
time. The twisted-pair lines would probably max out at two feet in length.

So far, the Micrel/Kendin KS8995MI looks very likely to be our chip of
choice. I'd hate to find a better one *after* we commit to it though.

Jim

 

[Jeff Liebermann]

On Fri, 02 Apr 2004 00:14:37 GMT, James Meyer <jmeyer@nowhere.com>
wrote:

On Thu, 01 Apr 2004 09:59:57 -0800, Jeff Liebermann
jeffl@comix.santa-cruz.ca.us> posted this:

On Thu, 01 Apr 2004 13:06:31 GMT, James Meyer <jmeyer@nowhere.com
wrote:

I need a chip to build a 4-port, unmanaged, 10/100 ethernet repeater
hub. AMD has one for 10 base-T only and Intel has one for 10/100 but the Intel
chip is twice as big and more complicated than I need for the application.

I need something in a small package and I don't need any of the extra
ports or functions that sometimes go with switches or stacking hubs.

Jim

If I had a clue as to how small [numbers], what features you need
[insert itemized list], and what you're doing [insert description], I
might be able to be more specific.

If my project manager(mangler) would let me know, I could be more
specific.

Indecision is the key to flexibility.

So far I've found out that a 100-pin quad flat-pack plus the magnetics
would fit.

Including the RJ45 receptacle, that's about 2 sq in. Lots of room.

It needs to be able come up running without any supervision or
programming even from an EEPROM.

Groan. You've got a PC-104 computah and you can't be bothered to
initialize your datacomm interfaces? Surely you jest. To the best of
my knowledge, there are no serial datacomm interface chips that have a
parallel input setup system (i.e. dip switches). Anything more
complex than an RS-232 UART requires some type of initialization or
pre-load of addresses.

It needs industrial temp specs.

Huh? Ocean temperatures don't vary that radically unless you're
planning to use this thing under an iceberg or inside a volcano. If
it's liquid water, it isn't much colder than 0C. The upper end may be
a problem if you go fishing inside a volcano.

It should be low power.

Everyone advertises their chips as "low power". Without numbers and
specs, the term has no meaning. For example, does it need a "sleep"
or "green" mode? Is that low power operating, or just in standby?

It should be off-the-shelf in small quantities.

That means Digikey and Mouser should stock it.

We're building an autonomous underwater vehicle with two ethernet
capable PC-104 computers, an RF modem,

I'm curious. What frequency RF are you using? RF doesn't go through
water very well, except at VLF (very low frequencies) as used in
submarine communications. The problem with VLF is that the data rate
is also very low. If you're gonna use any of the ISM frequencies
(915, 2.4GHz, 5.6GHz), you'll find that they were selected for the
frequencies of MAXIMUM water absorption (which is handy for microwave
ovens).

and an ethernet equiped microcontroller
to supervise things.

So, you don't need a switch or hub. You just need an xformer,
transceiver, PAD (packet assembler/disassembler), and probably some
help at the MAC layer. Try one of the Micrel transceivers:
http://www.micrel.com/product-info/fastether_trans.shtml
Also try Realek transceivers:
http://www.realtek.com.tw/products/products1-1.aspx?lineid=4
I don't know if any of these will work without preloading the MAC
address. I suspect not.

They all need access to each other with a fairly high data
rate.

How high a data rate? At what bit error rate? Over what distance?
At a given frequency and power level, each of these can be traded for
the other. It's like "good, fast, cheap. Pick two".

The system will probably need to be reconfigured on-the-fly from time to
time.

I don't understand. Are you building an underwater version of the
DARPA autonimous robot challenge? Duz on-the-fly mean while moving
around under water?

The twisted-pair lines would probably max out at two feet in length.

Then, why bother with ethernet? Ethernet has lots of advantages none
of which apply under water and disconnected from a LAN. Have your
PC014 micros spew whatever protocol and datacomm method is convenient.
Use fiber instead of copper unless you enjoy fixing corrosion
problems. In a point to point environment, with no interference, a
short path, and no need to communicate with the rest of the world,
on-off keying will work just fine.

So far, the Micrel/Kendin KS8995MI looks very likely to be our chip of
choice. I'd hate to find a better one *after* we commit to it though.

That's a 5 port switch. I thought you didn't need all the extra
stuff. Look for an ethernet transceiver and PAD instead.

Good luck.

Jim

--
# Jeff Liebermann 150 Felker St #D Santa Cruz CA 95060
# 831.336.2558 voice http://www.LearnByDestroying.com
# jeffl@comix.santa-cruz.ca.us
# 831.421.6491 digital_pager jeffl@cruzio.com AE6KS

 

[Jeff Liebermann]

On Thu, 01 Apr 2004 18:00:44 +0100, Tim Auton
<tim.auton@uton.[groupSexWithoutTheY]> wrote:

"dont know" <mojo@nospam_netscape.com> wrote:
James Meyer wrote:

I need a chip to build a 4-port, unmanaged, 10/100 ethernet repeater
hub.
[snip]
new 10/100 5-port hubs sell for 50-60 bucks. is it worth it to design
your own?

$50-60? I bought a 5-port 10/100 *switch* for ~$40 a few weeks ago,
and that was at a real shop not renowned for low prices. I'm sure you
could find a hub for ~$20.

Yep, you could probably find one for $20 (after rebates). However,
have you ever priced the small quantity parts prices for the same
ethernet hub or switch? I've never done it for a hub or switch, but
have certainly priced various RF related products. In all cases, I
can't buy the parts for less than what they sell the finished product.
When you're cranking out millions of identical products, and using
robots for assembly, the costs are very very very low. When you're
doing short runs or prototypes, you pay through the nose. For
example, the Micrel KS8995MI 5 port 10/100 chip mentioned elsewhere in
this thread is $21/ea in unit quantities from Mouser.


--
# Jeff Liebermann 150 Felker St #D Santa Cruz CA 95060
# 831.336.2558 voice http://www.LearnByDestroying.com
# jeffl@comix.santa-cruz.ca.us
# 831.421.6491 digital_pager jeffl@cruzio.com AE6KS

 

[John Woodgate]

I read in sci.electronics.design that Tim Auton <tim.auton@uton.[groupSe
xWithoutTheY]> wrote (in <ft7p60dvg54i1c1l38j267jprq1lq2kp42@4ax.com&gt
about 'ethernet repeater chip?', on Fri, 2 Apr 2004:

John Woodgate <jmw@jmwa.demon.contraspam.yuk> wrote:
I read in sci.electronics.design that Tim Auton <tim.auton@uton.[groupSe
xWithoutTheY]> wrote (in <nfio60tcbegi56kodo6pah2t0rdlvu7qqi@4ax.com&gt
about 'ethernet repeater chip?', on Thu, 1 Apr 2004:

Love is a travelator.

I've resisted asking this since you first posted it. What do you have in
mind?

It would appear you have given considerably more thought to my
signature than I have. This is not a good idea. Perhaps it is a
comment from my subconscious on the emotional numbness that depression
brings.

OK, that's a reasonable answer. But I'm very sorry to read that word

'depression'.
--
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

 

[James Meyer]

Top posted to warn lurkers of the length of this reply......


On Fri, 02 Apr 2004 01:37:50 GMT, Jeff Liebermann <jeffl@comix.santa-cruz.ca.us>
posted this:

On Fri, 02 Apr 2004 00:14:37 GMT, James Meyer <jmeyer@nowhere.com
wrote:

On Thu, 01 Apr 2004 09:59:57 -0800, Jeff Liebermann
jeffl@comix.santa-cruz.ca.us> posted this:

On Thu, 01 Apr 2004 13:06:31 GMT, James Meyer <jmeyer@nowhere.com
wrote:

I need a chip to build a 4-port, unmanaged, 10/100 ethernet repeater
hub. AMD has one for 10 base-T only and Intel has one for 10/100 but the Intel
chip is twice as big and more complicated than I need for the application.

I need something in a small package and I don't need any of the extra
ports or functions that sometimes go with switches or stacking hubs.

Jim

If I had a clue as to how small [numbers], what features you need
[insert itemized list], and what you're doing [insert description], I
might be able to be more specific.

If my project manager(mangler) would let me know, I could be more
specific.

Indecision is the key to flexibility.

The guy is so flexible that..... (insert favorite Johnny Carson joke
here).

So far I've found out that a 100-pin quad flat-pack plus the magnetics
would fit.

Including the RJ45 receptacle, that's about 2 sq in. Lots of room.

Even better. We're going to use a small 4-pin connector since the
physical connection doesn't have to be compatable with anything outside our
vehicle.

It needs to be able come up running without any supervision or
programming even from an EEPROM.

Groan. You've got a PC-104 computah and you can't be bothered to
initialize your datacomm interfaces? Surely you jest. To the best of
my knowledge, there are no serial datacomm interface chips that have a
parallel input setup system (i.e. dip switches). Anything more
complex than an RS-232 UART requires some type of initialization or
pre-load of addresses.

The Micrel chip has a pretty good default setup and it "learns"
addresses on the fly.

It needs industrial temp specs.

Huh? Ocean temperatures don't vary that radically unless you're
planning to use this thing under an iceberg or inside a volcano. If
it's liquid water, it isn't much colder than 0C. The upper end may be
a problem if you go fishing inside a volcano.

A long story.... the PM had his last project fail to start at cold
temps when testing out of the water. Rather than find out WHAT was failing, he
made a decree that all chips on this project MUST be industrial temp rated.

It should be low power.

Everyone advertises their chips as "low power". Without numbers and
specs, the term has no meaning. For example, does it need a "sleep"
or "green" mode? Is that low power operating, or just in standby?

The PC-104s, two of them, dissipate so much power that they need
heatsinks the size of great pyramids of Egypt just so they don't catch fire.
And the PM is pushing my part of the project for every milliwatt I can save. Go
figure.

It should be off-the-shelf in small quantities.

That means Digikey and Mouser should stock it.

Nu Horizons, bless them, have stock for the Micrel parts and will sell
as many or as few as we need.

We're building an autonomous underwater vehicle with two ethernet
capable PC-104 computers, an RF modem,

I'm curious. What frequency RF are you using? RF doesn't go through
water very well, except at VLF (very low frequencies) as used in
submarine communications. The problem with VLF is that the data rate
is also very low. If you're gonna use any of the ISM frequencies
(915, 2.4GHz, 5.6GHz), you'll find that they were selected for the
frequencies of MAXIMUM water absorption (which is handy for microwave
ovens).

The radio is 900 MHz spread spectrum and when we use it the vehicle
surfaces to stick the antenna out of the water. That comes in handy because we
can get a GPS fix at the same time.

and an ethernet equiped microcontroller
to supervise things.


So, you don't need a switch or hub. You just need an xformer,
transceiver, PAD (packet assembler/disassembler), and probably some
help at the MAC layer. Try one of the Micrel transceivers:
http://www.micrel.com/product-info/fastether_trans.shtml

That was what I was leaning towards.

Also try Realek transceivers:
http://www.realtek.com.tw/products/products1-1.aspx?lineid=4
I don't know if any of these will work without preloading the MAC
address. I suspect not.

I'll check out RealTek. The Micrel KS8995 series is one pretty smart
chip what with it's ability to learn how to handle packets.

They all need access to each other with a fairly high data
rate.

How high a data rate? At what bit error rate? Over what distance?
At a given frequency and power level, each of these can be traded for
the other. It's like "good, fast, cheap. Pick two".

The prototype is being tested with a 10 base-T hub. But the PC-104s
have 100 base-T capability so that's what the PM says I have to accomodate.

The system will probably need to be reconfigured on-the-fly from time to
time.

I don't understand. Are you building an underwater version of the
DARPA autonimous robot challenge? Duz on-the-fly mean while moving
around under water?

On-the-fly doesn't mean dynamically. But the payload could be
gosh-knows-what and it has to plug-and-play whatever it turns out to be.

The twisted-pair lines would probably max out at two feet in length.

Then, why bother with ethernet? Ethernet has lots of advantages none
of which apply under water and disconnected from a LAN. Have your
PC014 micros spew whatever protocol and datacomm method is convenient.
Use fiber instead of copper unless you enjoy fixing corrosion
problems. In a point to point environment, with no interference, a
short path, and no need to communicate with the rest of the world,
on-off keying will work just fine.

If I was the PM, I'd do things a lot differently. You wouldn't consider
buying the company and promoting me, would you?

So far, the Micrel/Kendin KS8995MI looks very likely to be our chip of
choice. I'd hate to find a better one *after* we commit to it though.

That's a 5 port switch. I thought you didn't need all the extra
stuff. Look for an ethernet transceiver and PAD instead.

You're right, we don't need all the extra stuff. But it's better to
have it and not use it than need it and not have it.

Good luck.

Thanks,

Jim

 

[Jeff Liebermann]

On Sat, 03 Apr 2004 01:35:06 GMT, James Meyer <jmeyer@nowhere.com>
wrote:

Top posted to warn lurkers of the length of this reply......

Something about warning labels...

On Fri, 02 Apr 2004 01:37:50 GMT, Jeff Liebermann <jeffl@comix.santa-cruz.ca.us
posted this:
(...)

Indecision is the key to flexibility.

The guy is so flexible that..... (insert favorite Johnny Carson joke
here).

One project I struggled through had a slogan:
"The quality goes in before the system gets figured out".

The Micrel chip has a pretty good default setup and it "learns"
addresses on the fly.

Yeah, it learns the addresses that it hears on the various ports in
order to determine what devices are out there and to do the switching.
However, it has to be loaded with its own MAC addresses, one per port.
Hopefully, that's done in some kind of NVRAM or built in flash.
Otherwise, you'll find yourself doing a pre-load.

A long story.... the PM had his last project fail to start at cold
temps when testing out of the water. Rather than find out WHAT was failing, he
made a decree that all chips on this project MUST be industrial temp rated.

Dumb(tm). Most of the chips involved probably do not have industrial
temperature range components. Anyway, it probably wasn't a component,
but rather condensation at the cold temperatures shorting something.
Trying to find a non-condensing cold atmosphere in the real world is
almost impossible. I used to design marine radios for Intech Inc. We
just assumed that everything would be soaking wet or coated with a
thin film of condensed water. The trick was to design with low
impedances. That's not terribly difficult but it does have one big
drawback. If done improperly, the power dissipation could become too
high. The heat from your PC104 smoking power hogs should keep the
guts fairly dry, but I wouldn't depend on it. Give the board a blast
of ecologically correct, non-freon, freeze mist until the frost
collects and melts into water on the board. If it still works, you
have a chance. I've done RF designs, where the boards could be
immersed in water and still sorta work (tuning changes are always a
problem).

The PC-104s, two of them, dissipate so much power that they need
heatsinks the size of great pyramids of Egypt just so they don't catch fire.
And the PM is pushing my part of the project for every milliwatt I can save. Go
figure.

Why bother. You have the biggest heatsink available, the ocean. Use
a heat pipe or bar of copper to plumb the heat to the outer surface
and let the water keep it cool. Don't ask me what to do when it's
outside the water. Maybe a removeable water jacket.

(Drivel: My hi-fi for many years was a class-A power amplfier. It
belched so much heat that I had to design a unique water cooled heat
sink system. It was 4ea cola cans bolted to the output xsistors and
filled with water. The water would slowly evaporate, thus cooling the
output xsistors. As long as I remembered to fill the cans, it worked
quite well).

The radio is 900 MHz spread spectrum and when we use it the vehicle
surfaces to stick the antenna out of the water. That comes in handy because we
can get a GPS fix at the same time.

That will work for the GPS but will be fun with the 900MHz antenna.
Unless you're planning an exotic antenna, the heights of the waves
will limit your range. Even a few feet away will be impossible if the
line-o-sight path is blocked by the waves. MBARI uses various
frequencies to supply buoy data back to Mt Toro (elevation 3560ft).
When the wave heights get above the buoy height, the microwave stuff
becomes erratic. Methinks you'll need to stick a lot more than just
the unspecified antenna out of the water.
http://www.mbari.org/moos/
Incidentally, you may wanna look at their AUV (autonimous underwater
vehicle) programs:
http://www.mbari.org/auv/

The prototype is being tested with a 10 base-T hub. But the PC-104s
have 100 base-T capability so that's what the PM says I have to accomodate.

OK. You're stuck with ethernet. I suppose it makes sense. When you
have to deal with trying to find a waterproof RJ45 connector, you'll
probably wish for fiber. Oh, you're using a different 4 pin cable and
connector. That has a chance of working (and later rotting).

On-the-fly doesn't mean dynamically. But the payload could be
gosh-knows-what and it has to plug-and-play whatever it turns out to be.

You might wanna talk to the satellite designers. The birds fly with
only rudimentary instructions in firmware. Usually, it's nothing more
than upload, download, and self protection routines in firmware. The
really complex software is uploaded remotely. For example, the
software for the Mars probes was not done at launch time and was
uploaded after launch.

If I was the PM, I'd do things a lot differently. You wouldn't consider
buying the company and promoting me, would you?

Sorry. I'm busy going broke following my own advice. Besides, you
wouldn't want to work for me. I tend to follow the project management
philosophies of Atilla the Hun, Ghengis Kahn, and Vlad Dracula. For
this maritime project, methinks "sink or swim" would be the
appropriate motivational style. Ever wonder what that plank hung
amidship is for? Avast ye scurvy dogs!

You're right, we don't need all the extra stuff. But it's better to
have it and not use it than need it and not have it.

I've always found that the unused, spare parts, are the ones that bite
me. Unused inputs should be tied either high or low, but never left
floating.


--
# Jeff Liebermann 150 Felker St #D Santa Cruz CA 95060
# 831.336.2558 voice http://www.LearnByDestroying.com
# jeffl@comix.santa-cruz.ca.us
# 831.421.6491 digital_pager jeffl@cruzio.com AE6KS

 

[Marc Randolph]

James Meyer <jmeyer@nowhere.com> wrote in message news:<fa3s605da0pgujva3lt403no3i1t7837q2@4ax.com>...

It needs industrial temp specs.

Huh? Ocean temperatures don't vary that radically unless you're
planning to use this thing under an iceberg or inside a volcano. If
it's liquid water, it isn't much colder than 0C. The upper end may be
a problem if you go fishing inside a volcano.

A long story.... the PM had his last project fail to start at cold
temps when testing out of the water. Rather than find out WHAT was failing,
he made a decree that all chips on this project MUST be industrial temp rated.

Howdy James,

Ahhh... true paranoia at work!

I-temp is going to be your biggest limitation. Most Ethernet vendors
focus on commerial temp, almost to a fault.

Zarlink is known for producing many things for I-temp, and has some
unmanaged switches. If you are willing to sign over your first born
child, Broadcom may have something, although if the Micrel or Zarlink
devices fit the bill, I'd suggest going with them. They'll support
you better than Broadcom. Forget about Marvell.

Good luck,

Marc

 

[James Meyer]

On 3 Apr 2004 11:31:43 -0800, mrand@my-deja.com (Marc Randolph) posted this:

James Meyer <jmeyer@nowhere.com> wrote in message news:<fa3s605da0pgujva3lt403no3i1t7837q2@4ax.com>...

A long story.... the PM had his last project fail to start at cold
temps when testing out of the water. Rather than find out WHAT was failing,
he made a decree that all chips on this project MUST be industrial temp rated.

Howdy James,

Ahhh... true paranoia at work!

I-temp is going to be your biggest limitation. Most Ethernet vendors
focus on commerial temp, almost to a fault.

I'm beginning to find that out.

Zarlink is known for producing many things for I-temp, and has some
unmanaged switches. If you are willing to sign over your first born
child, Broadcom may have something, although if the Micrel or Zarlink
devices fit the bill, I'd suggest going with them. They'll support
you better than Broadcom. Forget about Marvell.

Good luck,

Marc

Thanks Marc. This is my first ethernet design project. Hints from guys
that have been there before me are very valuable.

Jim