On Fri, 21 Nov 1997 13:06:23 -0800  Jonathan Stone wrote:

> Can you get the model number (like VR-2xx, or VR19) from the back of
> the monitor and post it? and also the output of
> 
> 	>> cnfg
> 
> with the model number and any version/submodel info of the PMAGB-BA?
> (>> cnfg N, where N is the slot with the sfb, might have more info.)

Open up the box and find out the speeds of the two oscillator cans on
the card, and the state of the oscillator-select jumper (described
further below).

> I seem to remember that the SFB actually comes in two varieties, which
> run at different video rates (60 and 66 Hz, or possibly 66 and 72).
> But the driver in NetBSD only seems to use one speed. Perhaps you have
> a monitor/card combination that Ultrix supports but NetBSD doesn't?

Simon Burge wrote:

> We've also got a number of SFBs here with an extra 60 something MHz
> crystal so that they can drive monitors like VR297s at 1024x864 (you
> select between the two crystals with a jumper).  I think the "offical"
> part number for one of these is a PMAGB-BC, but we just modified the
> PMAGB-BAs ourselves.  Maybe that could be a problem too.  This is where
> the `>> cnfg N' will help greatly - the SFB spits out all sorts of info
> when queried.
> 
> I _think_ that PMAGB-BBs could drive 72 & 66 Hz, and PMAGB-BDs could
> drive 72 & 60 Hz.  This is stretching the memory somewhat!

The SFB chip works with a wide range of clock speeds.  If you read the
original SFB paper from Digital:

  ftp://gatekeeper.dec.com/pub/Digital/WRL/research-reports/WRL-TR-93.1.ps

you'll see a _large_ range of clocks listed.  Based on the speed it
autoconfigures to various resolutions (although the software/firmware
needs to be able to understand what it's doing :-).

The PMAGB-Bx has a 130.808 MHz oscillator soldered onto the board, and
a socket for another (selected by a jumper near the crystals).  This
matches the VRT19's "hi-scan" rate (1280x1024, 72 Hz).
I think this also works with the VR-320?

In the PMAG-BA the 2nd crystal is 119.84 MHz. This matches the VRT19's
"lo-scan" rate (1280x1024, 66 Hz). 

In the PMAG-BC it's 69.196 MHz (the same oscillator used on the PMAG-B). 

As Simon says, if you have any surplus PMAG-B boards you can desolder
the oscillator and install it in a PMAG-Bx.  When I did this (at the
site that Simon & I worked at together) and told Digital about it, they
went away and checked their documentation, and eventually came back
saying that yes the cards were now PMAGB-BCs...

Be aware that you can't remove the 130.808 MHz oscillator - the card
uses this to judge the speed of the 2nd oscillator.

You'll notice that the card itself only says "PMAGB-B", not -BA/-BC/etc.

Hmmm, I've just found a card with a 74.397 MHz crystal on it.  Hadn't
noticed that before...  Here's the list from the Digital paper:

	Typical usage	Resolution	Refresh		Dot clock
	VGA		640x480		60		25.18
	VGA		640x480		72		32.00
	SVGA		800x600		56		36.00
	SVGA		800x600		72		50.35
	1K VGA, ACE	1024x768	60		65.00
	1K VGA, ACE	1024x768	66		66.00
	1K VGA, ACE	1024x768	72		72.80
	1K VGA, ACE	1024x768	72		74.37
	Digital		1024x864	60		69.20
	Sun		1152x900	66		92.98
	Sun		1152x900	72		104.00
	Digital		1280x1024	66		119.84
	Digital		1280x1024	72		130.81
	Digital internal 1600x1200	76		219.00

Getting hold of some oscillators at appropriate speeds and experimenting
to find out which ones work with the PMAGB-B could be an interesting task.

I don't have time to do this, but if anyone wants to take it on I will
offer one PMAGB-B card as a reserve (just in case you're nervous about
blowing up your only PMAGB-B...).
__
David Burren