Friday, October 6, 2017

HP 1000 terminal (2645A) working, Alto work accomplished


2645A terminal

I brought the 2645A terminal and keyboard to Marc's house, along with the box of spare boards, with the aim of putting together a logically consistent system then diagnosing any problems until it worked. Fortunately, Marc has eight 264x terminals of various flavors and some spare parts.
The 2645A, boards shown on left side of the ad
Using the documentation, cross checking it against the combinations found in Marc's terminals, I came up with the set of boards that should give me a useful working terminal. First, I installed the Keyboard Interface card and the Extended Serial Communications card which hooks this to the HP 1000 controller and the keyboard. 

Next, I set up the display group - the display control, display memory adapter, display timing, and display enhancement boards - were installed side by side with a top side connector tying the four together. The enhancement board had the multiple character sets available for the terminal, such as greek/math and graphic shapes. 

Adjacent to the display group, I installed a processor group - the 8080A processor board and a 32K Control ROM board, with a top side connector tying the two together. To the right of the processor I stuck in a universal RAM board with 16KB configured. 

Since the terminal had dual cartridge tape drives installed, I had to put in the tape controller board and the tape read/write board. The rubber capstan has turned to goo, so that must be replaced before I can think about putting cartridges into them. 

Marc has quite a bit of experience with these and recommends a mod to the write logic to allow use of a much more reliable alternate tape cartridge from Athana. The only downside is that the blank cartridges are pretty expensive - $50 each - but they appear to work very reliably. With the tape cartridges I can boot the HP 1000 from the reader in the terminal. 

When I powered up the terminal, I got garbage on the screen and/or strident error beeping.  I did diagnostic work as well as swapping boards known to work from other terminals, until I had identified the Universal Ram Board (60171) as the culprit. I had two of them, one 16K and one 32K, but neither worked properly.

Marc had a 16K 60171 which worked well, plus it was a good template for setting the 16 dip switches (called "strapping" by HP) to make it work properly in the system. 

While I could have begun testing the individual 16K RAM chips on the boards, Marc had a pile of spare RAM chips of the right type on hand. We just populated my board with eight 16K chips (making it a 16KB board), plugged it in, and everything worked perfectly.

The terminal is now back in my workshop ready to connect to the HP 1000. Once we figure out the replacement method for the capstan rubber, I can apply that to my two drives and make the slight mod to the R/W board to work with the Athana tape cartridges. 

2262D terminal

I had tested the monitor unit of the terminal previously and it appears to be working fine. It sails through its power on test, but a more comprehensive self-test must wait for a keyboard to be attached. My keyboard is missing the plastic bottom and the cable that connects it to the monitor. The monitor itself is missing its pedestal and base. 

I will build a wood stand for the monitor and a bottom for the keyboard, but first I need to build a cable. I did find schematics on bitsavers and know the pin-out for the connection on the keyboard side. Since I also know the pinout for the monitor side, the cable will be easy to fabricate. 


Diablo drive filter foam replacement

The air supply in the Diablo comes from a blower and goes through a HEPA filter, then is routed into the disk cartridge by a foam seal that fits around the metal flap door on the cartridge. The foam, however, has deteriorated to a partially friable, partially gooey mess. 

I removed the old foam and cleaned the area, before cutting and gluing on some soft rubber weatherstrip material. It fits well and won't introduce and particles in the clean air stream. 
That drive is now ready to begin archiving the remaining cartridges from the Xerox PARC archives. 

Ethernet connection box

Ken was chasing a troubling bug in the file transfer process in his ethernet connection box. This unit hooks directly to the Alto unit connector, thus not requiring the Ethernet Adapter Unit or 3 Mbit ethernet cabline. It makes use of the IFS code developed by Living Computers Museum + Lab
but is implemented on a BeagleBone.

The network box loaned by the LCM has to go back to CHM where it is needed for their Alto work, thus we have a deadline to complete the connection box. Marc will build a nice enclosure once the functionality is completely tested and solid.

Ken used today to track down the flaw, which he discovered by the afternoon and soon had a fix that allowed the box to completely support the FTP activity. He has some more testing and improvements to accomplish but is well on the way to our end of the month target. 

No comments:

Post a Comment