[net.periphs] HP LaserJet, a quick look

henry@utzoo.UUCP (Henry Spencer) (10/10/84)

We've got an HP LaserJet ($4k laser printer, daisywheel emulation with
minor smarts) on demo right now.  It's not bad.  Print quality is very
good.  We don't have much experience with it yet, but do have a few
comments and one major flame.

The built-in fonts are, uh, limited, and we have yet to see real
information on what font cartridges are available.  Mind you, when I
say the built-ins are limited, I mean in style; each is a 256-character
font with every foreign-language letter imaginable, so there can be few
complaints on that score.  Variety of characters aside, though, the
built-in fonts are basically landscape and portrait versions of 10-pitch
Courier, and that's all.  There are mumbles about math fonts, and we're
prodding the local HP man about 12-pitch fonts, but no details yet.  Also,
note that you can plug in *one* font cartridge at a time, and each of
them holds about 3 fonts maximum.

A pseudo-typesetter it's not, but with a few more fonts it looks like it
would make a *dandy* daisywheel replacement.  There is a lot of interest
hereabouts.

Now, for the bad part.

The thing wants to talk 9600 baud, 8 data bits, no parity, XON/XOFF.
PERIOD.  9600 baud, ok.  No parity, straightforward.  But try getting a
V7 (or derivative thereof, like maybe Xenix) to produce an 8-bit data
path with flow control...!?!  The LaserJet *must* have a full 8-bit
path.  If you can manage to get all the top bits zero, then the only
major effect will be that the top half of the font will be inaccessible
and the raster-graphics sequences won't work (they need 8-bit binary data).
But if your hardware or software insists on using that top bit for parity,
then you simply can't talk to the LaserJet.  Being among the lucky ones
with source licences, we put in a quick kernel kludge, and are now
talking to the thing just fine.  But before you sign a PO, make very
very sure that your Unix can give you a full 8-bit output path without
resorting to raw mode (which deprives you of flow control).

<FLAME ON>
Whatever possessed the normally-sensible people at HP to produce such
an abortion of a serial interface?  Don't they know that there are a
lot of systems out there -- not just Unixes, either! -- that insist
that a character is 7 bits only?!?  This is preposterous.  Whoever
imposed this ridiculous constraint should be shot.  At the very least
there should be a feature like the one common on small microcomputer
printers, in which a specific escape character means "act like the top
bit was on in the next character".  Our $400 Geminis will do this, but
the $4k LaserJet won't!  GRRR!!!  [Are you listening, HP?]
<FLAME OFF>
-- 
				Henry Spencer @ U of Toronto Zoology
				{allegra,ihnp4,linus,decvax}!utzoo!henry

toby@gargoyle.UChicago.UUCP (Toby Harness) (10/11/84)

> From: henry@utzoo.UUCP (Henry Spencer)

> The built-in fonts are, uh, limited, and we have yet to see real
> information on what font cartridges are available.

We have the two currently availble font cartridges  ($225 each) -

			Name		orient	pitch	point
(part number 92286A): 
			Courier Bold	port	10	12
			Courier Italic	port	10	12
			Line Printer	land	16.66	8.5

(part number 92286B):
			Helv Bold	port	prop	14.4
			Tms Roman	port	prop	10
			Tms Roman Bold	port	prop	10
			Tms Roman Ital	port	prop	10
			Tms Roman	port	prop	8.0
			Line Printer	land	16.66	8.5

> There are mumbles about math fonts, and we're
> prodding the local HP man about 12-pitch fonts, but no details yet.

I also have heard little more than mumbles about the math font.

> Also, note that you can plug in *one* font cartridge at a time, and each of
> them holds about 3 fonts maximum.

As above, the limt appears to be 6, plus the two that are built in, for a
total of 8.  Plus, with some effort, you can feed it in-line raster-graphics.
An entire font would be a lot of work (for the programer, cpu, and laserjet),
but a few 'extensions' to a standard font would be no real problem.

> A pseudo-typesetter it's not, but with a few more fonts it looks like it
> would make a *dandy* daisywheel replacement.  There is a lot of interest
> hereabouts.

I am sure this is HP`s intention.  Most of the adds I have seen have it
pictured next to an IBM PC.

> The LaserJet *must* have a full 8-bit
> path.  If you can manage to get all the top bits zero, then the only
> major effect will be that the top half of the font will be inaccessible
> and the raster-graphics sequences won't work (they need 8-bit binary data).
> But if your hardware or software insists on using that top bit for parity,
> then you simply can't talk to the LaserJet...
> before you sign a PO, make very
> very sure that your Unix can give you a full 8-bit output path without
> resorting to raw mode (which deprives you of flow control).

But not, of course, with SYS III and later. (Note that HP`s HP-UX is
SYS III/V.)   However, HP did not come up with this to sell to unix sites,
but to offices with a pc or two.  As much as a replacement for dasiywheels,
it is a replacement for dot-matrix, and many of those require 8 bit
data.  In any case, the laserjet has a 56k buffer, and prints at a very
constant 8 ppm, so it should be easy enough to spoon feed it through the
printer daemon if you had to.

> ...Our $400 Geminis will do this, but the $4k LaserJet won't!

Talk to your HP rep; we paid  little more than $2K.

Toby Harness		Ogburn/Stouffer Center, University of Chicago
			...ihnp4!gargoyle!toby

smh@mit-eddie.UUCP (Steven M. Haflich) (10/13/84)

The info posted on this device is interesting and useful, but what
I would really like to know from those with experience is:

1) What is the addressable graphic resolution (PPI) of the device?
2) Is the aspect ratio 1:1?
3) How accurate and reliable is the graphic positioning.

In other words, the device may be a great daisywheel replacement,
but will it also serve for occasional, low-volume-but-high-quality
graphics?  Any opinions?

gert@ttds.UUCP (Gert Svensson) (10/15/84)

In article <217@gargoyle.UChicago.UUCP> toby@gargoyle.UChicago.UUCP
(Toby Harness) writes:
>       .
>  <number of fonts is discussed>
>       .
>
>As above, the limt appears to be 6, plus the two that are built in, for a
>total of 8.  Plus, with some effort, you can feed it in-line raster-graphics.
>An entire font would be a lot of work (for the programer, cpu, and laserjet),
>but a few 'extensions' to a standard font would be no real problem.
>

As far I can understand it must be impossible to download an entire page
in raster-graphics mode with full resolution. The buffer (56k) seem much to
small to hold an entire page.

Gert Svennson      ..!mcvax!enea!ttds!gert

thomas@utah-gr.UUCP (Spencer W. Thomas) (10/15/84)

In article <2884@mit-eddie.UUCP> smh@mit-eddie.UUCP (Steven M. Haflich) writes:
>1) What is the addressable graphic resolution (PPI) of the device?
>2) Is the aspect ratio 1:1?
>3) How accurate and reliable is the graphic positioning.
>

Graphics resolution is 75dpi, 150dpi or 300dpi.  It is supposedly 1:1
aspect.  I assume that the positioning is just as accurate as the
character positioning.  The major flaw is that you can send a max of
56Kb of raster data.  We figured out that this amounts to 5 square
inches at 300dpi and about the whole page (79+ square inches) at 75dpi.
It would seem to be mostly there for printing letterhead logos.

=Spencer

henry@utzoo.UUCP (Henry Spencer) (10/15/84)

> > There are mumbles about math fonts, and we're
> > prodding the local HP man about 12-pitch fonts, but no details yet.
> 
> I also have heard little more than mumbles about the math font.

Latest word:  math fonts expected by January, 12-pitch sooner than that.

> ...  Plus, with some effort, you can feed it in-line raster-graphics.
> An entire font would be a lot of work (for the programer, cpu, and laserjet),
> but a few 'extensions' to a standard font would be no real problem.

Bear in mind that the LaserJet has serious limits on how much raster
graphics can appear on a page.  Sprinklings of non-standard characters
shouldn't be a grave problem, but I doubt very much that you could get
a whole page of text in a non-standard font into the raster memory.

> > The LaserJet *must* have a full 8-bit path...
> > before you sign a PO, make very
> > very sure that your Unix can give you a full 8-bit output path without
> > resorting to raw mode (which deprives you of flow control).
> 
> But not, of course, with SYS III and later. (Note that HP`s HP-UX is
> SYS III/V.)   However, HP did not come up with this to sell to unix sites,
> but to offices with a pc or two.  As much as a replacement for dasiywheels,
> it is a replacement for dot-matrix, and many of those require 8 bit
> data.  In any case, the laserjet has a 56k buffer, and prints at a very
> constant 8 ppm, so it should be easy enough to spoon feed it through the
> printer daemon if you had to.

True, System N will give you independent control of path width and flow
control.  But you miss my point:  there are *many* systems, not just old
Unixes, which can't give you that 8th bit.  Probably including a good
number of micros.  This decision remains stupid.  As I commented, even the
el cheapo dot-matrix jobs have a way to cope with a 7-bit data path.

Spoon-feeding it with a daemon is not an answer.  It is not easy to do
flow control in a daemon -- try it sometime, you need better IPC between
the sender and the listener than standard Unix has.

> > ...Our $400 Geminis will do this, but the $4k LaserJet won't!
> 
> Talk to your HP rep; we paid  little more than $2K.

The price is from our HP rep.  "$" doesn't automatically mean "US$",
although USAnians seem to think so.
-- 
				Henry Spencer @ U of Toronto Zoology
				{allegra,ihnp4,linus,decvax}!utzoo!henry

henry@utzoo.UUCP (Henry Spencer) (10/15/84)

> 1) What is the addressable graphic resolution (PPI) of the device?
> 2) Is the aspect ratio 1:1?

It's 300/inch both ways, like most any laser printer based on the new
Canon printing engine.

> 3) How accurate and reliable is the graphic positioning.

Barring firmware bugs, it should be accurate and reproducible to 1/300
of an inch.  This is raster-scan digital imaging, not character-at-a-time
printing.

> In other words, the device may be a great daisywheel replacement,
> but will it also serve for occasional, low-volume-but-high-quality
> graphics?  Any opinions?

It should work nicely for any graphics that will fit in the 59KB raster
memory limit.  This basically means your pictures can't be excessively
complex, and halftones are pretty much ruled out.
-- 
				Henry Spencer @ U of Toronto Zoology
				{allegra,ihnp4,linus,decvax}!utzoo!henry

toby@gargoyle.UChicago.UUCP (Toby Harness) (10/16/84)

> > > The LaserJet *must* have a full 8-bit path...
> > > before you sign a PO, make very
> > > very sure that your Unix can give you a full 8-bit output path without
> > > resorting to raw mode (which deprives you of flow control).
> > 
> > But not, of course, with SYS III and later. (Note that HP`s HP-UX is
> > SYS III/V.)   However, HP did not come up with this to sell to unix sites,
> > but to offices with a pc or two...

> But you miss my point:  there are *many* systems, not just old
> Unixes, which can't give you that 8th bit.  Probably including a good
> number of micros.  This decision remains stupid.  As I commented, even the
> el cheapo dot-matrix jobs have a way to cope with a 7-bit data path.

OK, let`s put it this way: It seems clear that the LaserJet was designed to sit
next to a MS-DOS machine.  I am not very comfortable with dumping on the
designers of the LaserJet.  I do not think they made very many stupid decisions.
I think they (out-)did what they were told.  The marketing Dept. of HP,
however, has shown its usual short-sightedness in its failure to recognize 
that such a printer would have a much wider market than IBM-PC owners. 

What I am trying to say is "sure, your point is valid, but it`s not the
LaserJet`s fault."  It does what it was designed to do and works with the
equipment it was designed to work with.  That people want it to do other
things or work with other equipment is a favorable reflection on the LaserJet
and a less than favorable reflection on the people who proposed the inital
specs.


> > > ...Our $400 Geminis will do this, but the $4k LaserJet won't!
> > 
> > Talk to your HP rep; we paid  little more than $2K.

> The price is from our HP rep.  "$" doesn't automatically mean "US$",
> although USAnians seem to think so.

$(Canada)4k is still much more than $(USA)2k.

Honest, some of my best friends are Canadian. :-)


Toby Harness		Ogburn/Stouffer Center, University of Chicago
			...ihnp4!gargoyle!toby

henry@utzoo.UUCP (Henry Spencer) (11/02/84)

I've got a bit more info on the LaserJet, from several sources.  If you
get the Service Manual (!), you will find out how to set the baud rate.
The number of choices is modest but includes most of the "reasonable"
speeds.  Note that the Owner's Manual does *not* include this information.

Unfortunately, you are stuck with the need for an 8-bit flow-controlled
path.  Some people have pointed out that the LaserJet is willing to do
flow control via modem-control lines, but this appears to be entirely
undocumented.  Doesn't mean it won't work, just that you can't count
on it working for future planning.  It also means that you need a
serial interface that can do flow control, which many of the cheaper
ones (e.g. DZ11) can't.  Our definitive solution to this one under V7
is to use an unused combination in the tty ioctls (both RAW and CBREAK
bits on) to give us the right behavior.  If you don't have kernel sources,
well, you're screwed.  Lean on your supplier or HP for a fix.

[If anyone actually wants it, our RAW+CBREAK code is simple and available,
although not yet thoroughly tested.]

I am told that downloadable fonts are probably in the LaserJet's future,
and that HP appreciates the desirability of more raster memory.  I have
no time estimate for either of these developments.
-- 
				Henry Spencer @ U of Toronto Zoology
				{allegra,ihnp4,linus,decvax}!utzoo!henry