IBM 5151 (MDA display) - common faults / gotchas?

[Chuck Guzis]

On 11/09/2014 10:28 AM, drlegendre . wrote:
> That's nuts.. so not only is the monitor relying on the computer to feed it
> a sane horizontal pulse, but that pulse rate isn't fixed or
> even securely delimited in hardware? Just POKEing the wrong memory location
> or writing bad data to some software register can set a catastrophically
> destructive video mode?

Yup, just some stray port outputs is enough to do it.   You'll note that 
the 5153 doesn't go this route, but has a regular horizontal oscillator 
circuit.

> Ok, yeah, I think I see the merits of having a fail-safe system ala the
> oscillator in the Zenith units we've elsewhere discussed. But of course,
> this does raise yet another obvious question.. why feed the monitor with a
> H-sync from the computer hardware in the first place?

I'm not entirely sure what you mean.  If it's "why not have the monitor 
furnish sync signals instead of the CRT controller>", it's a matter of 
what's available, I guess.  There weren't that many LSI CRTC's back 
then, and none of them could support such a scheme, that I'm aware of.

> In the Zenith / Osborne situation, the only possible answer would seem to
> be "No need to ever adjust (and as such, no need to even implement) a
> manual horizontal hold control".

It's there, but intended to be an "adjust at the factory and then cross 
your fingers".

> But for the IBM PC / 5151, it's a little less clear.. was there a
> perceived or real need to be able to software-switch the hotizontal
> (and/or vertical) scan rates? Some kind of text / framebuffer vs.
> hi-res mode-timing situation as we see in more modern (SVGA etc)
> monitors?

You really do need to tweak the timing.  Witness the arguments to the 
DOS MODE command that allows for shifting the image one way or the other.

> FWIW, way back in the bad old days of manually editing /etc/XF86Config, I
> do recall various warnings about tinkering with / feeding a bad modeline to
> the wrong type of monitor.. could result in major damage, and so on. So I
> guess this sort of thing went on for a very long time, then.. same basic
> situation?

Pretty much.  It's worth noting that the danger pretty much disappeared 
with multisync monitors.

--Chuck