Some time ago I ran across an HP 650c DesignJet color inkjet plotter on Craigslist for $50 from a local firm. I'd been looking for a large format printer for schematic and CAD drawings. That seemed like a deal so, although I was unable to test it, I grabbed it. This one turned out to be a C2859B model.
This model HP 650c is an E/A0 format plotter able to use paper up to 36 inches wide (see pic 1). The plotter can print in black or color ink at 300 dpi and at 600 dpi for black only. It was produced from 5/93 to 10/95; the latter B models supported 64 MB of SIMM memory as well as a Postscript SIMM. Base plot memory was 4 MB and there was a DirectJet network interface available. All of this is managed by an Intel 80960 32 bit RISC processor. The plotter and stand weigh in at just under 70 kg; the plotter itself is 60 kg. In 1994 the list price on the 650c was $9,995 USD.
When I got it, the plotter had the base 4MB of memory and a network interface but no Postscript SIMM.
When I got the plotter back to the lab I immediately found a few problems:
1) The paper refused to load; the plotter erred out on a paper misalignment error.
2) When the carriage fast traversed, it made a thunk-thunk-thunk noise.
3) The VFD display was nearly unreadable.
Given these problems and the press of other projects, I shoved it into a corner. I did run across a DesignJet 600 that I bought for $35. The 600 is a black only plotter, but shares many electrical and mechanical components with the 650c. I shoved that one into another corner.
Recently I had a project come up that really requires large format CAD prints, so it was time to unearth the plotters.
First, I disassembled the 600 donor. These plotters are built like tanks. It took me 4 hours to disassemble the plotter and bag the components. Everything is fastened with torqx screws except the main frame and rails which use an anti-tamper version. The main structural member is a 3 foot long 4 inch aluminum tube with 1/4 inch walls.
After that, I started work on the 650c. The paper loading problem was traced down to a flaw in the Y encoder strip. See pic 2.
Removing the encoder also allowed me to get to the drive belts and the thunk-thunk-thunk was coming from the short Y drive timing belt that had lost a large number of teeth. Both the encoder and the drive belt were replaced with items salvaged from the 600.
The VFD is a Futuba module and can probably be rejuvinated. Having an identical display from the 600, however, I just swapped it as well. It is quite acceptable and readable. The VFD module is in pic 3.
The 600 also yielded 16MB of add-on memory, so I installed that as well. I ultimately found a 64MB upgrade for $20 and eventually installed that to max the plotter out at 68MB.
I have not been able to find an affordable Postscript SIMM, so if anyone has one let me know.
I had a number of inkjet cartridges. These have been out of production for quite a while. The good news is the cartridges are so old they don't have the intelligence to know when they're outdated and they can easily be refilled. Refurbed cartridges are still available as well as some NOS HP cartridges. After a number of cleaning/reseating cycles I managed to get my cartridges more or less working and was able to print the internal demo plots.
Getting the plotter on my network was the next step. The plotter does support bootp, but it was easier to just enter the network ip through the control panel than to struggle with bootp support in my dhcp server.
HP provided Windows 3.1 drivers for the plotter.
Fortunately, I'm a Linux shop and the plotter is more or less supported by Ghostscript and CUPS. Finding a ppd file and entering the plotter into CUPS was trivial. Actually printing things, not so much. Scaling up files resulted in images cropped to letter size.
After much experimentation (and a half roll of paper) I managed to find reasonable solutions.
For Postscript CAD files, I used pstoedit to scale and convert the file to hpgl2. The resulting file could be sent through netcat directly to the plotter.
For Postscript schematic files, the process was a little more involved. Here I used pstops to scale up the file, then added rotate and translate commands to the resulting Postscript file and then sent it through CUPS with a large PageSize option (eg A1 or A0). Using the rotate and translate commands helps minimize the paper use. This was necessary as pstoedit was unable to handle the image file.
The final results can be seen in the last picture. This is a schematic of a Tektronix 213 power supply taken from an excellent scan of the manual provided by Artek. It's draped over the top of the plotter for scale. The brown smear in the lower right is actually from the scanned image and is not a defect from the plotter.
Now the fun is over, time to make some CAD drawings.