Improved MSV11-P FMPS

[Noel Chiappa]

So I have a busted MSV11-P, and I needed the FMPS to work on it. There is one
online, but on my computers, it was extremely faint, and very hard (almost
impossible) to read.

So I have worked on it some, and I have a new version here:

  http://ana-3.lcs.mit.edu/~jnc/tech/pdp11/MP01239_MSV11-P.pdf

which I feel is much more legigible, and is also a lot smaller (3.5MB instead
of 36MB). 

The default page size is quite small (I have no idea why, I didn't do the
re-conversion to PDF, I don't have Acrobat), but if you blow it up to about
300%, it still has good resolution at that point, and it's quite readable -
all the signal names, pin numbers, package numbers, etc are quite legible.

It doesn't have as much resolution hidden away as the original, but for most
purposes, it seems quite acceptable (in fact, more so in some ways, given the
increased legibility). I was interested in producing something readable, and
easier to manage, and I think this new version succeeds at both.


For those who are interested in the details of what I did (because I've seen
other FMPS scans with this issue, and it might help someone down the road),
after first saving all the pages as individual images (like I said, I don't
have Acrobat, so whatever I did had to be done with something else - and I'm
not sure Acrobat can do what needed to be done to make them legible, anyway),
I started out by trying to simply increase the contrast. That didn't do
anything (at least, with my image tool).

Looking at the page images under magnification, I discovered why: the areas
of ink (lines, letters etc) were actually (in the scanned image) a stipple of
white and black pixels, which together produced the (un-readable) light gray
printing of the original. So the contrast enhancement knob didn't do anything
- each individual pixel was already quite light, or dark.

So I used something called an averaging tool (which takes small groups of
pixels, and averages them together), with a small averaging box size (I used
2x2), which converted the ink areas to a uniform grey; I could then use
contrast enhancement to bring the printing up.

I then reduced most pages to 40% of the original size, since the originals
were scanned at 600 dpi, in 8-bit/pixel grayscale, and were pretty huge. (I
didn't go that far on a few pages - the PCB images - which could use the
higher resolution. Also, if you want the full resolution, the original scan
is of course still available.) In addition to making the images smaller, this
actually increased the crispness of the printing, since the reduction
sampling process got rid of a lot of the jaggedness that the previous steps
had left.

I finally converted the resulting images from 8-bit/pixel grayscale to
1-bit/pixel black-and-white; on inspection of the two side-by-side, this lost
a tiny bit of definition, but I felt that the reduction in size was worth it,
plus to which my image tool has compression for 1-bit/pixel B+W TIFFs, but
not for 8-bit/pixel grayscale TIFFs, so I won doubly on the size.

	Noel