Sad Story

[Al Johnson]

On Monday 15 December 2008, Karthik Kumar wrote:
> On Mon, Dec 15, 2008 at 5:35 AM, Stefan Monnier
>
> <monnier at iro.umontreal.ca> wrote:
> >>>> There are still hardware problems that exist with almost every
> >>>> freerunner out there (GPS signal levels, for one).  I would like to
> >>>> see them fixed by Openmoko, Inc.
> >>
> >> The kernel fix for GPS is a mere quirk. Ideally it should get fixed in
> >> hardware.
> >
> > Huh?  The kernel fix seems to work OK for those people who have old
> > FRs.  Newer FRs have a hardware fix.  E.g. my FR doesn't need any
> > kernel workaround.  No need to wait for GTA03 to get a hardware fix.
>
> So, this kernel fix is a feature, according to Al Johnson? Then why do
> they need a hardware fix?

They don't. Neither of my Freerunners has the hardware fix, and the GPS works 
just fine with any kernel that includes the fix.

> Unless I am missing something, this software  
> fix seems to be a lie. Or the hardware fix seems to be bogus anyway.

You are missing something, so I will try to explain. The 'hardware fix' is a 
small capacitor from the SD clock line to ground, and was a quick response to 
the problem when the SD was discovered to be interfering with the GPS. This 
slows the transition between the digital high and low states, reducing the 
level of interference generated. Reducing the drive strength does the same 
thing, but through a slightly different mechanism, and doesn't require a 
soldering iron to make the change. Note that the hardware fix only reduces 
interference from the clock line. The kernel fix reduces interference from 
the other signal lines too.

To understand the issues more fully I suggest you read up on the analogue 
properties of digital electronics, particularly in regard to the speed of 
transitions and their effect on EMI.

> I have one of the earlier freerunners released, so I'm sure Openmoko
> owes an explanation.

Both hardware and software fixes were discussed extensively on the lists, and 
are in the list archive.