Rockbox mail archive
Subject: Re: Getting rid of that damn noise.
From: Andrew Jamieson (ajamiesn_at_mira.net)
> The noise is always present and a constant volume, regardless of whether
> vol is up or down. It's just very hard to hear when the vol is up.
> Also, the noise is not a constant ripple - it exactly matches whatever the
> LCD is actually _doing_ (you can actually control the noise by changing
> the LCD is doing), eg it's whether something is being changed or updated
> onscreen. Thus I'm pretty sure that when the HDD reads and the screen
> freezes, it's silent because the LCD stoped updating, not because of the
> HDD. The HDD doesn't explain all the other links with the LCD.
> I'm guessing that this rules out voltage inverters (they would be linked
> power drain such as HDD, not to the LCD screen), and also seems to rule
> the LCD's power supply, clock, etc, as these things would all produce a
> constant hum or whine, which the noise isn't.
> Turn the volume right down so you can hear the hum, then start browsing
> menus. then go back to the level-meter screen. You'll see what I mean -
> noise is directly tied to what is visibly changing on the LCD, not an
> underlying constant.
> Particularly with the level-meter, which updates constantly at a lowish
> frequency (4hz?), the hum is replaces by a pulsing at that frequency and
> remains constant with the updates until they freeze.
> I don't see how these things can be explained by something outside the LCD
> system. No component (other than the CPU) has it's state affected by what
> actually displaying on the LCD, and you can control the noise by
> what is on the LCD.
> AFAICS, the buck stops at the LCD system.
It sounds to me like it's the data for the sync serial used by the CPU to
control the LCD. The LCD contrast / power is active at all times, so if the
noise can be stopped when the HD is accessed it is probably not this
(although a switchmode power supply going into high freq. to power the HD
would explain it, this would also change if the backlight came on (as the
LEDs would draw more power, increasing the switching frequency) - anyone
test this for us?). The data line would explain why the noise stops when
the HD is active - the CPU stops upodating the LCD (and thus the clock - it
probably polls the HD and thus requires an exclusive state machine,
therefore no LCD updates).
If it is the LCD serial data, then the ultimate cause is a lack of proper
power-supply de-coupling at the CPU. This is quite a common thing, as
complete de-coupling is virtually impossible without seperate supplies (and
even then sudden loading can ripple back through the primary supply to all
sub-supplies ....). Possibly a simple test for this would be to 'update'
the ASCII table in the code (I assume it must have one, as the LCD
controller doco does not mention an embedded one) to contain all blanks.
This is possible given our current level of knowledge. Errrr ...... just
thinking about this for a moment, I'm not sure it will work - the CPU will
probably still send the 'blank' information to the LCD (I doubt they parse
the info for white space), and if the noise does disappear, it could still
be the contrast switcher (which would have no work).
*Sigh* Forget I mentioned it.
Anyway. I think it's the data line.
PS: Sorry I have been a bit slack with my schematics recently (Joachim is
putting me to shame), but I actually found a job again! Bummer, huh? I
will continue to update when I can.
Page was last modified "Jan 10 2012" The Rockbox Crew