Date   : Fri, 10 Dec 2004 12:15:23 +0000
From   : Richard Gellman <splodge@...>
Subject: Re: Video ULA shenanigans

Richard_Talbot-Watkins@... wrote:

>
>
>Richard Gellman wrote on 10/12/2004 11:28:19:
>
>  
>
>>One or two...
>>
>>Firstly, such a thing is impossible, not because of rates, etc.,
>>but because of the way the chip is designed. The chip
>>generates a 40x25 display. No arguments. If you feed weird
>>clock rates at it, at best you *might* get 80x12.5, but my
>>guess is after 40 chars the teletext chip just idles until the
>>next line sync pulse.
>>    
>>
>
>Yeah I wondered if that might be true - since teletext attributes last to
>the end of a 40 column line, there must be a degree of hardwiring for a 40
>column display.
>
>  
>
Control codes are something I hadn't even considered. Internally the 
chip will keep going to "the end of the text line". Now whether the chip 
interprets 40 chars as end of line, or the line sync pulse as end of 
line will determine when control codes "expire". With my guess being the 
former, I would suggest some interesting teletext effects might be 
observed if it did somehow manage an 80 character display.

>>The net effect is you get 40x25, but squashed on the left
>>side of the screen.
>>    
>>
>
>...though I'd still love to see half-width teletext characters on screen,
>if it were possible...
>
>  
>
A thought has occured. The clock input to the TT chip is (from memory) 
6Mhz, with internal stepping down inside the chip. Mode 0-3 use 2Mhz, 
and Mode 4-6 use 1Mhz. Thus if you did switch it to the "higher" 
frequency, it would actually get wider!

>>IIRC, the internal circuitry uses a separate clock for the
>>teletext which is controlled by the teletext bit in the Video
>>ULA. Thus Acorn have effectively prevented such a mode. :(
>>    
>>
>
>That was the other thing I wondered - if the teletext clock speed is fixed,
>perhaps doubling the Video ULA clock speed would result in only half of
>each teletext character being displayed in each half-width cell?  That is,
>unless as you say, the Video ULA clock speed is effectively overridden by
>setting the Teletext bit.  An easy way to find out I guess would be to type
>*FX154,91 in Mode 7 and see if setting the higher clock rate has any
>visible effect at all.  If it did, I would expect at worst an unsynced
>scambled display.  If it had no effect, then it's probably safe to assume
>that Teletext mode ignores the Video ULA clock speed.
>
>  
>
IIRC, the clock selector works thus:

if (teletext) clock = 6, elseif (hiclock) clock=2 else clock=1.

Thus you could set the higher clock speed in the ULA, but if the 
teletext bit is set (which it needs to be to select the TT chip for 
video output) then its clock overrides.

>When I next get my hands on a Beeb, maybe I'll try these things out and see
>what happens.
>  
>
-- Richard Gellman