soi cmos for rfic

[jason]

Hello All

Anyone who deal with or understand the challenges of using soi cmos for
rfic design?
Is the trend good and how does it go?

Kindly share with me

Thank you

Jason

 

[Keith Williams]

Hello All

Anyone who deal with or understand the challenges of using soi cmos for
rfic design?

SOI CMOS for RF? Why not GaAs or SiGe? SOI CMOS is more of a logic
process.

 

[Jim Thompson]

Hello All

Anyone who deal with or understand the challenges of using soi cmos for
rfic design?
Yes.

Is the trend good and how does it go?

Yes, it's good.

Kindly share with me

That's what I get paid to do ;-)

Thank you

Jason

...Jim Thompson

 

[Kiviranta, Mikko]

[email protected] says...

SOI CMOS for RF? Why not GaAs or SiGe? SOI CMOS is more of a logic
process.

CMOS is dirt cheap, and incredibly pliable even without SOI.
See the paper by Doan et al, in the 2004 IEEE Solid State Conference.
His slides are at
bwrc.eecs.berkeley.edu/Research/RF/Publication/pubs/Doan_ISSCC04_slides.pdf

Regards,
Mikko

 

[Keith Williams]

CMOS is dirt cheap, and incredibly pliable even without SOI.
See the paper by Doan et al, in the 2004 IEEE Solid State Conference.
His slides are at
bwrc.eecs.berkeley.edu/Research/RF/Publication/pubs/Doan_ISSCC04_slides.pdf

The question was specifically about SOI. AFAIK, SOI is more of a logic
process. Is anyone doing serious RF in SOI? I know there is some
analog (PLLs etc.).

 

[[email protected]]

Keith Williams skrev:

The question was specifically about SOI. AFAIK, SOI is more of a logic
process. Is anyone doing serious RF in SOI? I know there is some
analog (PLLs etc.).

I believe SiliconWave does (did?) a Bluetooth transciever in SOI,
siw1502

-Lasse

 

[Jim Thompson]

The question was specifically about SOI. AFAIK, SOI is more of a logic
process. Is anyone doing serious RF in SOI? I know there is some
analog (PLLs etc.).

Honeywell in Maryland did, though I think that operation is now shut
down.

...Jim Thompson

 

[Phil Hobbs]

Honeywell in Maryland did, though I think that operation is now shut
down.

...Jim Thompson

We do, I believe. Back in the dim distant days when we made laptops
(i.e. before May 2005) we did all the wifi stuff ourselves. Some was
SiGe, but not all. We're still supplying Lenovo, as well as a lot of
other customers.

Cheers,

Phil Hobbs

IBM T. J. Watson Research Center
Yorktown Heights NY

 

[jason]

Thank you All for the comment.
But why there are comments on Good things of SOi while some even shut
down their plant or production?
Will cmos soi become mainstream of cmos roadmap ?
How does it benefit analog , Rf and digital domain respectively?

Anyone knows?
By the way, Jim, do you design soi cmos circuit for analog or rf
circuit?
Why did you choose soi for? Low noise or ?

Hear from you all

Thank you

Jason

 

[Jim Thompson]

Thank you All for the comment.
But why there are comments on Good things of SOi while some even shut
down their plant or production?

Yield. Reproducibility.

Will cmos soi become mainstream of cmos roadmap ?

I don't think so. I've yet to see a stable process.

How does it benefit analog , Rf and digital domain respectively?

Low stray capacitance to ground. Low crosstalk.

Anyone knows?
By the way, Jim, do you design soi cmos circuit for analog or rf
circuit?
Yes.

Why did you choose soi for? Low noise or ?

High frequency. 1.6GHz GPS receiver.

Hear from you all

Thank you

Jason

...Jim Thompson

 

[Keith Williams]

We do, I believe. Back in the dim distant days when we made laptops
(i.e. before May 2005) we did all the wifi stuff ourselves. Some was
SiGe, but not all. We're still supplying Lenovo, as well as a lot of
other customers.

Many (most?) of the ThinkPads use "Centrino Technology", thus the WiFi
is Intel's. A quick look shows the T40s, X31s, and R40s are Centrinos,
anyway.

http://www.pc.ibm.com/us/thinkpad/mobiletechnology/

 

[Phil Hobbs]

Many (most?) of the ThinkPads use "Centrino Technology", thus the WiFi
is Intel's. A quick look shows the T40s, X31s, and R40s are Centrinos,
anyway.

http://www.pc.ibm.com/us/thinkpad/mobiletechnology/

Right, the 2 GHz ones are mostly Intel now--the challenges are at higher
frequency.

I know the guy who designs them--they started out with a 900 MHz
cordless connection for a modem, way back before wifi. Now they have a
60 GHz transceiver chip working, and 94 GHz in the works. They talked
about it at ISSC '05, and are going to again at CICC in September, but I
don't have the reference handy.

Cheers,

Phil Hobbs

 

[jason]

Phil

The 60GHz transceiver is made in SOI technology or bulk cmos or SIGe?

Thanks

Jason

 

[Mikko Kiviranta]

The 60GHz transceiver is made in SOI technology or bulk cmos or SIGe?

If Phil refers to the same Berkeley group I referred to
earlier:
http://bwrc.eecs.berkeley.edu/Research/RF/Publication/pubs/Doan_ISSCC04_slides.pdf
that is bulk CMOS.
As I am not collaborating with those guys (my colleagues
here at VTT do) I'm not up-the-date with the details, though.

Quoting from Doan et al "Design considerations for 60GHz
CMOS radios" by Doan et al in the December 2004 issue of
the IEEE Communications Magazine (p137):
"A wideband general-purpose 60GHz amplifier has been designed
and fabricated in a 130nm digital CMOS process with no special
analog or RF oprions".

Regards,
Mikko

 

[jason]

Thanks Mikko
So it is a bulk cmos process.

By the way, anyone here deals with device physics for a SOI device
before?
I need some advice.

By the way Jim, how do you decide if you would go for fully or
partially depleted soi?

Hear from you all

best regards
Jason

 

[Jim Thompson]

Thanks Mikko
So it is a bulk cmos process.

By the way, anyone here deals with device physics for a SOI device
before?
I need some advice.

By the way Jim, how do you decide if you would go for fully or
partially depleted soi?

I didn't. As is often the case, I was stuck with the client-selected
process.

Hear from you all

best regards
Jason

...Jim Thompson

 

[Phil Hobbs]

Phil

The 60GHz transceiver is made in SOI technology or bulk cmos or SIGe?

Thanks

Jason

I misspoke, sorry. The 60 GHz transceiver is SiGe bipolar. It was done
by Brian Gaucher's group here at IBM Watson.

Cheers,

Phil Hobbs

 

[jason]

Thanks Jim



Jason