Test points - a possible future direction

[Thomas Seiler]

Hi,

I accidentally hit reply instead of reply-all, my apologies for making
this thread rather hard to read...

Werner Almesberner wrote:
> > Thomas Seiler wrote:
> > Currently there is not safe / easy way to get power inside the
> > Freerunner for 3rd party mods :(
>
> You mean to make the Neo provide power to external components ?
> How about hooking them up to the battery terminals ? They're
> reasonably easy to access.

Yes. I can imagine people developing some interesting hardware add-ons
for their freerunners, *if* there is a proper / official way to power them.
Ideally, it would be streamlined enough for commercial add ons, not only DIY.

> Alternatively, how about VDD_3V3 on the WLAN test pad ? It's connected
> to IO_3V3 through a bead that's rated at 300mA. Or VCCA, which goes
> directly to IO_3V3.
>
> The test points may be hidden. See
> http://wiki.openmoko.org/wiki/Image:Gta02a5_pcba_ps.JPG
>
>
> > The temporary fix would be to solder wires on capacitors, but this is
> > obviously not a good idea...
>
> Hah, done it a million times ;-)
>
> > The current debug connector still has a few not connected (not
> > documented?) pins.
>
> There the undocumented ones should all be unused. That's 21 of them.
>
> I'm not sure if I'd like to see more things getting added to the
> debug connector or whether I'd rather have them on that new test
> point connector. The reason is that I'm thinking of ways how we
> could get rid of the debug board (but not its functionality). If
> this succeeds, then the whole debug connector would go.

Well, just thinking out loud here:
What about putting a FTDI in every GTA03 and use a mechanical switch,
accessible from the battery compartiment, that is switching miniUSB
between
Samsung SoC and the FTDI, which would only be powered, when the switch
is in the right position ?
+ no debug board anymore, as all phones include the debug function by default
+ Software upgrading the NOR flash gets easy for everybody
-  additional cost of switch and FTDI :(
- its not possible to JTAG-debug usb drivers, as the miniUSB port is
already taken then :(
But i diverge...

> Of course, there's also a different approach: don't add a new
> connector and fill up the debug connector first. Just for
> monitoring the power rails I would have a dozen lines, though.
> If we add your four audio signals and toss in a pair of GPIOs, we're
> already with only 3 lines left. Ah, and I want all the buttons too,
> so that's one left, maybe zero.
>
> Yet another approach would be to put all the things that are useful
> for testing on a new connector (which could even be removed in mass
> production versions), and put things that are useful for extending
> the device on the debug connector. That would mean that we'd never
> completely get rid of the debug connector.
>
> The common denominator would be the FPC that gives access to things
> without the overhead of the debug board.
>
> > [speaking of the lack of power on the debug connector]
> > I think it would be possible to wire a few of them [pins] in parallel to
> > provide a useful current.
>
> Hmm, only if you need *very* little power. They're rated at 1mA.
> I don't know at what point they really burn up, but almost anything
> you could connect would already be about order of magnitude outside
> the specification, easily by two orders of magnitude.

ok, its a bad idea then.

> > In my dreams :-), the debug connector of GTA03 would provide:
> > 1) vbatt with a current limiter (always on) to allow for wake sources
> > on the 3rd party mod flex... (i.e. rfid readers :-)
>
> I think making VBAT accessible should be no problem (with the
> constraints below). I think some people would bark if I started
> to suggest additional circuits for that, though :)

straight VBAT is better anyway, it would even allow for a LED torchlight mod.

> > 2) io_3v3 that goes to standby together with the freerunner, for
> > simpler things like i2c sensors
>
> Hmm, GTA02 has an unused LDO, but I think we might use them all up
> on GTA03 :-( In any case, I would just put that one on a big
> test pad (e.g., something like the larger ones around the debug
> connector), so that one can easily attach some wire. The FPC
> connectors just aren't made for this sort of things.
>
> I am actually thinking of putting power lines on the test point
> connector, but that would be only for monitoring, not to actually
> drive something.
>
> > 3) add one interruptable gpio.
>
> We might be running low on these. But if there's one that's still
> available, definitely yes.
>
> > 4) put on the unused line in / line out of the wolfson if there are
> > still pins left...
>
> LINE1/2 are on TP3001 and TP3002, but they're pretty hard to access
> (right next to the codec). I'm not sure what kind of noise might get
> coupled into LINE inputs, though. Outputs are low impedance, so they
> should be suitable for the debug connector.

The Line in was tought to allow for FM-Radio mod. If you simply set
the gain to 0
in the mixer shouldn't this take care of the noice when nothing is conected ?

> > 3) use a board 2 board connector (i.e. from molex) to pass higher
> > currents) instead of another FPC
> >
> Hmm, two problems: first, this sounds as if it's getting a bit
> expensive. I think an FPC like the one I described should cost
> about USD 3-5 a piece, consumer price. (Disclaimer: that's just
> from looking at catalog prices of mass-produced FPCs. Could be
> that this being a custom design changes the price dramatically.)
> I think a piece of PCB with a real connector might end up being
> considerably more expensive.
>
> Second, board-to-board connections are rigid. So you have to make
> a much greater effort to stabilize your construction. With the FPC,
> you just tape your contraption down and you're good.
>
> > 6) add all the usb signals ? (including the ID pin)
>
> That might be an option for GTA03, where we'll probably still have
> only Full Speed. After that, we'll almost certainly get High Speed
> (finally !), and I'm not sure how this fits with signal integrity
> demands. Even Full Speed is somewhat picky.

> > All this would allow for battery covers with an fpc circuit to extend the
> > functionallity

> Hmm, besides power, an FPC should be very suitable for this as
> well. Just the "neck" would have to be a little longer than what
>I've drawn. (That ~1cm was really just a random guess. It should
> be reasonably short, particularly if we want to measure things.)
>
> > Advantages:
> > - can provide useful amounts of power to mods
>
> Power is still tricky, yes. A big pad would solve the problem for
> DIY circuits, because you wouldn't mind a little bit of soldering
> anyway, but if you want to make more streamlined extensions, that
> wouldn't be acceptable.
>
> I wish there was a miniature equivalent for the ubiquitous 0.1"
> header.

Hm, good point. what about the type of header found on a 2.5"
Harddrive? (is it 0.05" ?)
These should not cost too much, and be still rather easy to get and
solder for DIY.

You would just need a 2x3 block or so:
VBAT  VIO_3v3  (power these cannot pe but on FPC)
GND   GND
I2C     I2C         (these are nice because a single connector for
most basic mods is sufficient)

Sorry again for messing up the thread...

Cheers,
Thomas

-- 
Excercise 17:
If the human brain was simple enough for us to understand we'd be so
simple we couldn't understand.
Prove this by induction.