[mythtv-users] LircChannelChanger / HD Channels - Going to wrong channel

[HP-mini]

On Wed, 2014-04-02 at 14:01 +0000, Gary Buhrmaster wrote:
> Debounce compensation compensation.
> 
> Real remotes actually send a constant stream of codes
> while you are pressing the key (not just a one time code),
> so that the receiving device can catch the code.  The
> receiver sees that stream of (say) '0' codes, and determines
> that one has pressed (and is holding) the '0' key.  With
> real world (IR) remote controls, people's fingers "bounce"
> a bit on the keyboard, or wave the remote around a little
> (there is good human factor research to show you cannot
> hold the remote absolutely steady).  So STBs that see a
> stream of '0' codes, with a very small interruptions, just
> presume a bouncy finger, or a waving of the hand.  They
> will not acknowledge that you actually intended to stop
> pressing the '0' key until some time has elapsed, *unless*
> they see a different code (so, in generally, you can press
> '1234' as fast as (humanly) possible, but '0001' may only be
> seen as '01'.  This is often called debounce compensation,
> since the key is considered to be "bouncing".  Since humans
> tend to be slower than computers, and they are watching the
> screen accepting the different '0's, this is rarely seen when
> using the manual remotes.  But blasters are fast.  So
> LircChannelChanger has a rdelay to deal with insuring that
> if a '0' is sent immediately after a previous '0' (a repeated
> character delay) an additional delay is added.
> 
> 
The real switch debounce is handled in remote before any IR Tx.
IMO That debounce is for unintended button fumbling/rollover.
Real remotes also use sequence numbers to indicate unique button
presses. This sequence number increments (modulo 16) for every unique
button press.
This is why optically interrupted button press does not cause key/button
repeats.
Unless LIRC generates these sequence numbers the inter-digit delay would
have to be greater than the repeat timeout of IR Rx.
Recording raw remote codes could have problem with not having unique
sequence numbers (this is prob unlikely for small number buttons).

AGC
All (approp. designed) IR Rx have AGC with varying time constants, the
AGC gets pumped by IR code preamble (& potentially by external signals).
So placing a high powered Tx emitter very close to Rx LED can cause
problems over-driving the Rx &/or not allowing enough time for AGC to
settle.