plustek: Thread: It's alive! (was: Re: Epson 1250)

Hi,

the lamp on my Perfection just turned on for the first time :)))))

After checking the LM9832 manual and comparing with the source code, I
was sure, that the light should be on theoretically, but it still kept
dark. Then I started wild guessing that Epson might (ab)use one of the
misc i/o lines to gate the lamp. So I configured all misc lines as
outputs and set them to 1 in plustek-devs.c, and the lamp lighted up!

I still have to figure out which of the lines it is, and I still
haven't tested scanning.

Stay tuned....


cu
	Reinhard

Hallo again,

On Thu, 20 Dec 2001 at 17:59, Reinhard Max wrote:

> I still have to figure out which of the lines it is,

The lamp seems to be controlled only by misc I/O pin 5.
It doesn't turn off when reg 29 is set to 0.

> and I still haven't tested scanning.

This is unfortunately not yet working. The image that comes out is
till black (or grey if the brightness is being raised), and sometimes
shows dotted diagonal lines.

I have to stop for today now. Hopefully I can continue somewhat during
my Christmas vacations.

cu
	Reinhard

Hi,

this is not only a wild guess, the UMAX 3400 and 5400 are also
LM983x based devices and it is known that they use  one
of the misc I/O pins. It seems, that this is the better way to keep
the lamps up during the reset phase of the chip which is often
done during the calibration. The sideeffect here on Plustek devices
which use the reg29 is, that the lamp is very often switched on and off...

Anyway - good work. This may help to make the stuff work with
the EPSON...

Gerhard

On Thursday, 20. December 2001 18:44, Reinhard Max wrote:
> Hallo again,
>
> On Thu, 20 Dec 2001 at 17:59, Reinhard Max wrote:
> > I still have to figure out which of the lines it is,
>
> The lamp seems to be controlled only by misc I/O pin 5.
> It doesn't turn off when reg 29 is set to 0.
>
> > and I still haven't tested scanning.
>
> This is unfortunately not yet working. The image that comes out is
> till black (or grey if the brightness is being raised), and sometimes
> shows dotted diagonal lines.
>
> I have to stop for today now. Hopefully I can continue somewhat during
> my Christmas vacations.
>
> cu
> 	Reinhard
>

On Thursday 20 December 2001 05:04 pm, Jaeger, Gerhard wrote:
>Hi,
>
>this is not only a wild guess, the UMAX 3400 and 5400 are also
>LM983x based devices and it is known that they use  one
>of the misc I/O pins. It seems, that this is the better way to
> keep the lamps up during the reset phase of the chip which is
> often done during the calibration. The sideeffect here on
> Plustek devices which use the reg29 is, that the lamp is very
> often switched on and off...
>
>Anyway - good work. This may help to make the stuff work with
>the EPSON...
>
>Gerhard
>
>On Thursday, 20. December 2001 18:44, Reinhard Max wrote:
>> Hallo again,
>>
>> On Thu, 20 Dec 2001 at 17:59, Reinhard Max wrote:
>> > I still have to figure out which of the lines it is,
>>
>> The lamp seems to be controlled only by misc I/O pin 5.
>> It doesn't turn off when reg 29 is set to 0.
>>
>> > and I still haven't tested scanning.
>>
>> This is unfortunately not yet working. The image that comes
>> out is till black (or grey if the brightness is being raised),
>> and sometimes shows dotted diagonal lines.

I haven't gotten the noise lines here, I'm all black, when doing 
a preview, and either can't open device nil. or out of memory 
when I try a real scan instead of a preview, this after setting 
that register to 0x9b, but it did bring the lamp up,  till I 
unplug it to reset it.  So this is going to need some cleanup 
code in the exit routine.

>> I have to stop for today now. Hopefully I can continue
>> somewhat during my Christmas vacations.

I hope so too, its been so real around here I've not had more 
than an hour or so to play with it most evenings.  I'd sent some 
email to epson support inquiring about an updated Software 
Development Guide, but as usual, no answer.  Would it be better 
to get in/on plustek's own development list?  Assuming there is 
such a thing...

-- 
Cheers, Gene

Hi guys,

well first of all - are you on the Plustek list?
If so, there's no need to cc the mails to each other, except
you prefer to get the same mail twice - let me know!


On Friday, 21. December 2001 03:43, Gene Heskett wrote:
> I haven't gotten the noise lines here, I'm all black, when doing
> a preview, and either can't open device nil. or out of memory
> when I try a real scan instead of a preview, this after setting
> that register to 0x9b, but it did bring the lamp up,  till I
> unplug it to reset it.  So this is going to need some cleanup
> code in the exit routine.

I think this all can be done in the code, where the register 0x29
is set or in the lamp functions. It should be easy to extend the
code that way...
What I can't understand is, that the images are completely black.
Maybe there's something out of sync...
Back to the misc I/O 5.
If you have a closer look to the doc, you can see, that this
register defaults to an output - great. By using this output
to toggle the lamp, you should write an 0x19 to turn the lamp
on and a 0x11 to switch the lamp of...
To write an 0x9b makes no sense, as the bits 1 and 2 have
no influence on the register if configured as output. The 9
should also not affect the light - except it also has something
to do with that...

Anyway good luck in finding the trick....
Gerhard

BTW: Reinhard, what's your job at SuSE?

On Friday 21 December 2001 06:29 am, Jaeger, Gerhard wrote:
>Hi guys,
>
>well first of all - are you on the Plustek list?
>If so, there's no need to cc the mails to each other, except
>you prefer to get the same mail twice - let me know!


Well, yes I am on the ####@####.#### but I was hoping that 
plustek themselves might have a user list or development branch, 
anything that might be of help here.

So yes, I seem to be getting 2 copies of these messages, and I see that 
kmail, when I clicked on the reply all button, filled in both the 
address line and the CC: line, so g.jeager, plustek, me, amd max will 
all get copies.  Lets see if I can edit that down to just the list?  
Yup, appears to clean up real nice so only the list gets this reply.

[...]

Cheers & Merry Christmas to all, Gene

On Thu, 20 Dec 2001 at 23:04, Jaeger, Gerhard wrote:

> this is not only a wild guess, the UMAX 3400 and 5400 are also
> LM983x based devices and it is known that they use one of the misc
> I/O pins.

Well, I didn't know that fact, so for me it really was wild guessing
:)

cu
	Reinhard

On Fri, 21 Dec 2001 at 12:29, Jaeger, Gerhard wrote:

> well first of all - are you on the Plustek list?

Yes.

> If so, there's no need to cc the mails to each other,

Right, but that would be easier, if the list would set a Reply-To
header that points to itself. Is it possible for the list maintainer
(is that you, Gerhard?) to configure the list that way, or does
linuxhacker.org not support Reply-To headers?

> I think this all can be done in the code, where the register 0x29
> is set or in the lamp functions. It should be easy to extend the
> code that way...

That was my thought, too.

> What I can't understand is, that the images are completely black.
> Maybe there's something out of sync...

I'll have a look into the other SANE backend which has rudimentary
support for the Epson 1250 (HP scanjet 2200)....

> Back to the misc I/O 5.
> If you have a closer look to the doc, you can see, that this
> register defaults to an output - great. By using this output
> to toggle the lamp, you should write an 0x19 to turn the lamp
> on and a 0x11 to switch the lamp of...
> To write an 0x9b makes no sense, as the bits 1 and 2 have
> no influence on the register if configured as output. The 9
> should also not affect the light - except it also has something
> to do with that...

Sorry, that's my fault. I mixed the high and low halfbyte of Register
5B. So the lamp is controlled by misc I/O 6 which can be set by the
higher half of 5B. 

> BTW: Reinhard, what's your job at SuSE?

I'm in the development departmend, and my main job is maintaining
packages (e.g. Tcl/Tk and PostgreSQL).

My participation here is more of private nature, although I also spent
some working hours on investigation.

cu
	Reinhard

On Thu, 20 Dec 2001 at 21:43, Gene Heskett wrote:

> I haven't gotten the noise lines here, I'm all black, when doing 
> a preview,

Try scanning with brightness >= 10, and 100dpi. I got a grey
background with some coloured noise pixels then. When you scan with
default brightness, the noise lines are so dark, that they can't be
distinguished from black, but you can see them when you make an ASCII
dump of the image using 'pnmnoraw'.

> and either can't open device nil. or out of memory when I try a
> real scan instead of a preview,

The 2.4 kernels (especially 2.4.10, AFAIK) until 2.4.16 had some
problems with the virtual memory engine. There should appear an
official update to 2.4.16 for recent versions of SuSE Linux these
days. You could try if that fixes your memory problems. It should also
work on other distributions.

> I hope so too, its been so real around here I've not had more 
> than an hour or so to play with it most evenings.  I'd sent some 
> email to epson support inquiring about an updated Software 
> Development Guide, but as usual, no answer.

Maybe they are also burried in end-of-year-business, and don't
currently have the time to answer such requersts.

> Would it be better to get in/on plustek's own development list?  
> Assuming there is such a thing...

Plustek probably won't be of much help to get the Epson scanner
working, because the only common thing is, that both manufacturers
build scanners around the same chip from National Semiconductors.

I think only Epson themselves can tell us the details and specialities
about their implementation. If they don't, we'll have to guess or
re-engineer.

Hopefully I can spend some more time on hacking this afternoon.... I
am planning to have a look into the other SANE backend that has
rudimentary support for Epson 1250 (HP Scanjet 2200), and see if I can
learn something from there.

cu
	Reinhard

A small update guys.

I found some code called 'lm9832' on one of the links somebody 
sent me.  This code did control the light, even turning it off 
when you selected the exit from the pulldown.

But it wouldn't run the scanner.  So I've kinda played 
mix-n-match for those motor control values that I thought were 
working in the plustek code.

Reinard: The can't open device stuff you got is group & owner on 
the devices, change group to something that exists as a user.

I'm still getting a bunch of un-initilized stuff warnings from 
the compiler, but it did just now do about half a page in the 
preview mode (which acted like it was some superhigh mode, slow, 
stopped several times before it hung the usb ports, note plural, 
I had to reboot to restore my usb mouse on another port.

But the scanned image as it was painted in the preview window 
appeared to be about the right color albeit over bright and 
bleached.  And the lamp was shut off at some point, its off now, 
but still sitting about halfway down the page where it locked 
everything up.

Now I'm gonna run it again to see it it homes the carriage 
properly.

It did, but the home motion was somewhat slower than I would have 
thought, the motor hummed pretty good and took about 6 seconds to 
reach the home position, but it didn't sound as if it actually 
hit anything when it did.

I need ftp (or cvs maybe)  write perms someplace where I can 
stash what I've got so far (I'm on a dialup) with gftp or cvs so 
others interested in this can poke and prod it too.  Even cvs 
would be ok if I can learn how to cause a cvs merge.

Somebody running a ps2 mouse could probably clean house on a 
lockup without a reboot, but thats what I get here and I *_love_* 
this usb optical mouse.  :-)

-- 
Cheers, Gene