I saw this and thought of C.A.F.

http://www.altera.com/corporate/news_room/releases/2010/products/nr-innovating-at-28-nm.ht
ml

Fuck-a-doodle-do, 28 Gbps transceivers. Maybe now we will see who can 
really design a PCB and who is a blowhard. Bring. It. On!!

Oh, there are some other interesting things also.

Cheers, Syms.

On Feb 14, 9:05=A0am, Symon
<symon_bre...@hotmail.com> wrote:
 28 Gbps transceivers. Maybe now we will see who can
> really design a PCB and who is a blowhard. Bring. It. On!!

Yes, the 28Gbps was more interesting than some 'another fab process
notch' release.

As for PCB design, doesn't that actually get easier, as you have less
and less freedom of length ;) ?

Symon wrote:
> Fuck-a-doodle-do, 28 Gbps transceivers. Maybe now we will see who can 
> really design a PCB and who is a blowhard. Bring. It. On!!
damnit, the fundamental wavelength of a 28GHz signal
is smaller than the pin's pitch... or close...

> Oh, there are some other interesting things also.
like... hummm... the price tag ?
oh, and the software tools that we'll have to
pay to be able to debug for Altera ?

> Cheers, Syms.
happy hacking,
yg
-- 
http://ygdes.com / http://yasep.org

On 2/13/2010 8:37 PM, -jg wrote:
> On Feb 14, 9:05 am, Symon<symon_bre...@hotmail.com>  wrote:
>   28 Gbps transceivers. Maybe now we will see who can
>> really design a PCB and who is a blowhard. Bring. It. On!!
>
> Yes, the 28Gbps was more interesting than some 'another fab process
> notch' release.
>
> As for PCB design, doesn't that actually get easier, as you have less
> and less freedom of length ;) ?

You may be right. But if you are, my microwave buddies are gonna have to 
take a big pay cut!

Hedging, I will still be taking them out for beer to steal their secrets.

Shhhh...

Syms.

On 2/13/2010 8:57 PM, whygee wrote:
> Symon wrote:
>> Fuck-a-doodle-do, 28 Gbps transceivers. Maybe now we will see who can
>> really design a PCB and who is a blowhard. Bring. It. On!!
> damnit, the fundamental wavelength of a 28GHz signal
> is smaller than the pin's pitch... or close...
>
Right! We will see who paid attention in their maths lessons!

à bientôt, Syms.

On Feb 13, 2:57=A0pm, whygee <y...@yg.yg>
wrote:
> Symon wrote:
> damnit, the fundamental wavelength of a 28GHz signal
> is smaller than the pin's pitch... or close...

Well it's late, and I haven't looked at the announcement yet, but
1/28th of a ns should be slightly over a centimeter for light in a
vacuum, probably around half an inch for a signal on a board --
exactly how far apart are the pins on this new chip :-)

BTW, if you're talking about the wavelength, you can probably double
that to around an inch, because 28Gbps using 1s and 0s is probably
only a 14 GHz signal, right?

Pat

On Sat, 13 Feb 2010 21:28:54 -0800 (PST), Patrick
Maupin
<p...@gmail.com> wrote:

>On Feb 13, 2:57 pm, whygee <y...@yg.yg> wrote:
>> Symon wrote:
>> damnit, the fundamental wavelength of a 28GHz signal
>> is smaller than the pin's pitch... or close...
>
>Well it's late, and I haven't looked at the announcement yet, but
>1/28th of a ns should be slightly over a centimeter for light in a
>vacuum, probably around half an inch for a signal on a board --

eh? 0.5in==1.27cm; but group velocity on a typical PCB is 
something around c/2 so surely you meant "under a centimeter".
OTOH I really haven't got the first idea what happens
on PCBs at >2GHz frequencies; presumably FR4 won't cut it
any more, and more exotic substrates are required to
avoid excessive loss and dispersion?

>exactly how far apart are the pins on this new chip :-)

tee hee.  Please let's forgive the poor chap his
one-order-of-magnitude error.  The newspaper I usually
read has a nasty habit of telling me that a power station
has an output of a few hundred "mw", which is about 1E9
times smaller than I would expect :-)

>BTW, if you're talking about the wavelength, you can probably double
>that to around an inch, because 28Gbps using 1s and 0s is probably
>only a 14 GHz signal, right?

Ah, now it gets interesting.  Nothing in the original
press release said anything about GHz.  Are we *sure* that 
it's NRZ coding?
-- 
Jonathan Bromley

On Feb 14, 5:02=A0am, Jonathan Bromley
<jonathan.brom...@MYCOMPANY.com>
wrote:
<snip>
> Ah, now it gets interesting. =A0Nothing in the original
> press release said anything about GHz. =A0Are we *sure* that
> it's NRZ coding?
> --
> Jonathan Bromley

Oooooohhh.....  QAM?  OFDM?  I need to find my notes on the Costas
loop, don't I?

On Feb 14, 8:32=A0am, John_H
<newsgr...@johnhandwork.com> wrote:
> On Feb 14, 5:02=A0am, Jonathan Bromley <jonathan.brom...@MYCOMPANY.com>
> wrote:
> <snip>
>
> > Ah, now it gets interesting. =A0Nothing in the original
> > press release said anything about GHz. =A0Are we *sure* that
> > it's NRZ coding?
> > --
> > Jonathan Bromley
>
> Oooooohhh..... =A0QAM? =A0OFDM? =A0I need to find my notes on the Costas
> loop, don't I?

The standard marketing gigabits probably comes from the
total bandwidth of all transceivers on the part.  And if it
really means 28 Gbps on a single wire pair, NRZ, then
a bit period would represent about .4" of FR4 using the
old 1 ns/ft estimate.  Still a large pin pitch but no doubt
a real pain to work with.  I hope the transceiver pins are
on the edge of the package, lined up to properly connect
to a standard connector / fiber transceiver without
any cross-overs.

- GLS

Patrick Maupin wrote:
> Well it's late, and I haven't looked at the announcement yet, but
> 1/28th of a ns should be slightly over a centimeter for light in a
> vacuum, probably around half an inch for a signal on a board --
> exactly how far apart are the pins on this new chip :-)
hmmm I had a 10 fold error in my rought estimate, but
it scares me anyway :-)

> BTW, if you're talking about the wavelength, you can probably double
> that to around an inch, because 28Gbps using 1s and 0s is probably
> only a 14 GHz signal, right?
oooopps right, I get caught in this old trap again and again...

> Pat
yg

-- 
http://ygdes.com / http://yasep.org