GSM-SDcard interference?

Tilman Dilo tilman.dilo at gmail.com
Tue Feb 17 12:46:55 CET 2009


On Tue, 17 Feb 2009 09:25:43 +0800, Andy Green wrote:

>> a screen slowly filling with colorful noise 
>
> Ah really... that sounds more like generic Glamo-trashing than
> MCI-specific issue.  All the SD stuff goes through the Glamo so maybe
> it's just pointing at something deeper that's not directly to do with
> SD.

For this to be true, you need to find an explanation why lowering 
sd_max_clock gets rid of all problems whatsoever (higher drive strength 
enables me to use higher clock rates), and why the thresholds are that 
different for both cards (at low drive strength, I still got errors at 
ridiculously low sd clock speeds with the Kingston card).

I've been using my FR with the SanDisk almost three months now (at 12.5 
Mhz) and did not have a single noise-filled screen at that clock rate.

> GPRS uses more slots AIUI so it dents the battery more during TX, maybe
> the issue is around that.  Eg, dodgy Glamo reset circuit.

Do higher sd clock rates use that much more power?

I've had the transfer errors with the FR fully charged and plugged in on 
my computer, but never a noise-filled screen. It is hard to tell though 
if I just did not produce enough occasions for a significant chance of 
that happening.

A screen filling with noise occured at above 80% battery charge iirc. And 
never with lowered sd clock rates.

> What would help is dump the glamo regs while it's normal, and if you see
> that trashed display again, dump it if possible and we can see the diff.

I'll write a script and put it in internal flash, and see if I can get my 
FR to execute it once it happens again.

I suspect that it has something to do with the controller of the sd cards 
- is it possible that interference (on the clock line?) gets it to a 
state where it sends out random junk that irritates the glamo? Filling 
one screen top-down with noise takes approximately five to six seconds 
(iirc), which indicates a transfer rate of 150 KB/s. I'll have to take a 
closer look if I see regular patterns that indicate some structure in the 
data.

Maybe the voltage level of the clock line does not fall below a certain 
threshold in time because of the capacitor with voltage spikes induced by 
GSM interference...

- Tilman










More information about the hardware mailing list