From A Friend>
Hi Mike,
>if Spinco does in fact get the GSM x-licenses why indeed >could there not be a 1x
>CDMA/GSM chip by next summer or sooner.
>Here's why not.
>1. x-license buys the right to us the technology once it >is created. One has to create the
>technology first.
Good point. The author should then conclude that if SpinCo is to produce GSM based
chips in the very near future then SpinCo could conceivably have started on this a long
time ago? Recall from Dr.J's own statement regarding other folks doing
WCDMA/cdma2000 at the moment - 'anybody can develop anything in the labs.
Royalties, licensing etc. come into play when they try to sell it' - Why wouldn't that apply
to SpinCo too?
>2. Good CDMA RF engineers don't transition into good GSM >engineers overnight.
>They should be in shape with prototypes by next summer.
Very good point. Does the author know of any GSM RF engineers SpinCo could hire? ;)
Through Globalstar and elsehwere Qualcomm/SpinCo does have some GSM experience.
>3. Even if the RF side is sorted - I know the software >side isn't. A multimode chipset
>will need 4 (arguably 5) major components
>a) Access substrata
>QCOM/SpinCo are historically good at this
>b) The network substrata
>This is pure GSM - SpinCo would be nuts to try to write >this from scratch, and they
>wont. QCOM has been touring the UK in an acquisition hunt. >They don't have a deal
>with anyone (yet)
>c) Application programmers interface.
>The API shields the handset software chaps (the MMI >developers) from the
>complexities of the protocol.
>This is a weak area for QCOM, who made a dogs dinner out >of this component for
>IS-95.
Quite on the contrary. Look for the APIs and software description at:
qualcomm.com
>A GSM API is much more complex.
>d) GSM Call Control state machine.
>This component could either form part of the MMI or the >API. In my experience, most
>GSM chipset vendors don't supply this component, but >nearly all Handset OEM's,
>particularly those in Taiwan (and the putative ones in >China) want the chipset guys to
>supply it.
What's silly to re-invent should be presumably acquired :)
>By the time WCDMA gets to market, I think this will be >offered as part of the ASIC
>software
>e) Reference design for an MMI
>A bigger job than you think, bu could be outsourced.
Again, look at the link to MSM software above.
>a) .. e) if written from scratch, would take 100+ software >guys two years to get into a
>state fit for type approval.
>b) realistically requires a partnership deal with a stack >vendor, who could also help
>with c) and d)
This is needed for CDMA anyway - can be extended over to GSM.
>After the software is written and integrated on chip, then
>the ASIC's have to be type approved. The process from >start to finish typically takes 9
>months. If SpinCo was working with a partnership company >like TTP, it could be done
>quite a bit quicker.
This is a valid point too - Type Approval process is lengthy - then good to acquire access
to technolgies that have already gone through the process?
>Presumably there will be a lead customer at this point, >and a lengthy series of field trials
>will start. These can last anywhere from 3 months to a >year, then you can start serious
>production.
Yeah. Qualcomm/Spinco has done similar things with IS-95A, IS-95B, now 1x and
1xEV.
I think the author has a fairly good understanding of the chip engineering process but let
that not deter anyone from thinking that Qualcomm/Spinco doesn't - these things and a
lot, lot more are surely being taken into account. |
