PDP-11/44 Restoration

This winter RICM has undertaken the restoration of a PDP 11/44.

The PDP 11/44 was the last PDP 11 designed with discrete logic for the processor.

1/2/11

• • Installed a DD11-DK 9-slot Unibus backplane that was removed from a spare BA-11 chassis.

  • Installed a M9300 Unibus terminator in slot 9 of the DD11-DK backplane.

  • Interconnected the two Unibus backplanes with a M9302 Unibus jumper.

  • Installed G727A jumpers in the Unibus slots where the NPR jumper was intact and installed G7273 boards in the Unibus slots where the NPR jumper was cut.

  • Both RA81 disks power up. The lower one has a flakey circuit breaker. They don't go ready so heads are probably locked.

  • The system powers up and the RUN LED and the LED on the M9076 MFM board both go on for a few seconds. Looks promising.

1/8/11

• • Installed two M8722 MS11-MB 256KW memory boards from MT's collection.

  • Installed the two boards for a UDA50 from MT's collection, the red SDI cables, and the SDI bulkhead from RICM spares.

  • The console cable was missing so we made one from a spare DL11 console cable and the connector housing from an Ethernet cable. We lucked out when we found that the plastic Ethernet connector housing and the one for the console cable were the same. We wired the console cable for DCE and used a male DB-25 connector so we don't need to use a null-modem cable to connect it to the terminal.

  • We tried two VT220 terminals and found that both work. We will use the green screen one for the console.

  • The console baud rate switches on the M9076 MFM board looked wrong so we took the easy path and set them all open for 19,200 baud.

  • We checked the settings of the M7090 Console Interface board. The jumpers look OK, but we found that one of the leads on the transistor Q1 was broken. Transistor Q1 is for the 20mA interface so we gave it a try as is.

  • On power up we saw nothing on the console, but the led on the MFM board indicated that data was being sent to the terminal. We installed an M7090 from MT's collection. On power up we saw some garbage on the screen and then ?22 CP HUNG. Major progress! All of the console commands resulted in the ?22 CP HUNG error. The manual said this means that the console told the CPU to do a Unibus transfer, but it didn't happen.

  • The LEDs on both UDA50 boards were cycling so they are alive.

  • We pulled the UDA50 boards and replaced them with G7273 boards. On power up we saw some garbage on the console and then just the >>> prompt. We could write and read from memory and registers, so the 11/44 CPU must be mostly functional.

  • The LED on the M9076 Console Interface board goes on when writing to the terminal. The other M9076 didn't do that.

  • Reinstalling the UDA50 boards resulted in the ?22 CP HUNG error so the M7485 board must have a problem. Time to look for another one.

  • We followed the procedures in the 11/44 User's Guide to see what boot PROMs were installed. The results were strange so we pulled the M7098 Unibus Interface board to see what was physically installed. Nothing! We will install DL (RL02), DU (RA81), and DY (RX02) boot ROMs.

1/16/11

• • Installed a working UDA-50.

  • Installed boot ROMs for RX02, RL02, and MSCP (RA60, RA80, RA81, RA90).

  • We tried to spin up the two RA81 drives. Both start to spin and then stop. Probably needs replacment starting capacitors. There are more RA81 drives in the warehouse that we can try.

1/22/11

• • I tried to spinup the RA80 drive. It smelled like something was cooking and popped the breaker. I tried the last RA81 in the collection that that has the same spinup problem as the other two drives. At least this drive has one of the newer black HDAs. Time to find some startup capacitors.

  • I found three M8256 RX211 floppy disk controllers in the warehouse and tried to boot the 11/44 from the RX02 that came with the 11/83. It made a lot of strange noises and didn't boot. I tried the RX02 from the 11/70. That sounded more promising, but still didn't boot. I connected that RX02 into the 11/34 and the RY0: could read a floppy. RY1: didn't work at all. After a little surgery I found that the drive belt for RY1: had come off. From the dust coating the belt it had been off for a very long time. I cleaned the heads, but RY0: still would not boot. RY1: seems to work OK. I swapped RX211 boards and verified that all three spare controllers are OK. I connected the RX02 to the 11/44 and got it to boot from RY1. The boot procedure seems a little strange; put the switch in HALT, enter "B DY1" and hit enter, move the switch to CONT.

1/30/11

• • Did some more debugging on the RA81 drives. I got to try the Termiflex that I bought years ago.We have 5 RA81 and 1 RA80 drive that all behave the same; they start to spinup and then stop. I fixed an RA81 that had the same symptom a few years ago by replacing the starting capacitor. We will try to find a replacement starting capacitor this week.

  • Found a bad spot in the cable that connects the RX211 controller to the drives. I will cut off about a foot of damaged cable and crimp on a new connector.

  • The UDA50 controller did not have the cycling LEDs display. We found that wiggling the larger ribbom cable that interconnects the two boards fixed it. I will replace the cable next week.

  • We tried to connect an RA70 disk to the UDA50. The drive works fine at home when it is installed in a VAX-3500, but will not spin up without the operator control panel connected. Maybe I need a jumper on the connector for the operator control panel?

  • Booted RT11 again this week. At least that works.

  • Next week we will try installing the RL02 controller and booting from an RL02 disk.

This is the special special Termiflex that can be used to debug an RA80, RA81, or TU81. With a combination of three shift buttons on the side you effectively have a full keyboard. There is 4k of memory and a scroll switch on the left side.

The RA81 drives pass the basic non-spinning tests.

This is the firmware and hardware version and the serial number of the RA81.

The Master and Slave Diagnostic Status.

The Master and Slave Diagnostic Memory

The fault indication after the drive tries to spin up.

The front panel light indication, the LED indication, and the fault code all say 1; Spin Error or Spindle motor speed transducer timeout.

2/6/11

• • We replaced the starting capacitor in one of the RA81 disk drives. The spinup behavior is the same. Moved the drive next to the power outlet to minimize the AC voltage drop to the drive. That didn't make a difference. Pulled the HDA to inspect the sensor on the spindle. Looks OK. Tried the new starting capacitor in another RA81 and observed the same behavior.

  • We installed the new M7762 RL11 disk controller in the 11/34. Works OK with the existing RL02 drive and RT-11 pack.

  • We found quite a few RL02 and RL01 drives in the warehouse. Some had significant rust on the spindle and near the heads. These will be a parts source. One very clean drive has a broken circuit breaker. We can borrow a replacement from a "parts" drive. We selected the best looking RL02, cleaned just a little rust off the spindle and scrubbed the inside of the drive. We daisy-chained it off the RL02 in the 11/34. It worked sometimes, but worked better with a shorter cable. We found a bent pin in the RL02 terminator that was on the drive in the 11/34 so we need to inspect the connector contacts on the back of the drive. We ran a cable directly from the RL11 controller to the "test" RL02 and it worked perfectly.

  • We installed the RL11 disk controller in the 11/44, cabled the RL02 drive, and it booted XXDP and RT11. Looks like we have a good combination.

2/19/11

• • We tried an RA81 on an AC outlet next to the power panel. The voltage goes from 120VAC to 119VAC when the RA81 tries to spin up, so I guess that low voltage is not a problem. There is a YouTube video showing the debug of the same problem on an RA82. That was a cold solder joint on the spindle sensor, so I will bring a 'scope to the warehouse next time. The RA81 manual says that the operational temperature of the RA81 goes down to 10C. It was probably 2C in the warehouse yesterday. Maybe the RA81 drives don't like the cold?

  • We started pulling the RA90s out of the rack, there are 8x of them. By swapping parts were were able to get one to go ready. By playing with the WP, A, B, and Test buttons we got a display that looks like the drive number. We changed it from 11 to 0 and when we pushed Test it asked if we wanted to save the drive number. So I think that we reset it to 0.

  • The UDA50 was not always displaying the LED pattern. Wiggling the ribbon cables between the two UDA50 boards changed the behavior so we replaced the ribbons. We think that the UDA50 is OK now. We installed a new red cable assembly from the UDA50 to the cabinet kit. On the side of the cabinet kit where the red cables plug on the 0 and 3 stamped in the metal matches the 0 and 3 on the cables. On the other side where the SDI cables plug on the 0 and 3 stamped in the metal is opposite of the red cable side. The large tab cutout in the metal where the connectors plug in is oriented the same on both sides of the cabinet kit. Since we were unsure about the cable orientation we tried a red cable from a KDA50 that does not need a cabinet kit. The behavior was the same.

  • In any case, doing a B DU0 at the >>> prompt gets nothing other than an octal number. We were not sure if the octal number is an error code. Since there is no PDP-11 boot block on the RA90 we were not surprised.

  • We booted RT-11 V5.05 from an RL02. Doing a DIR DU0: gets a "unknown device" or similar message. The du.sys file is on the RT pack. I don't know if this RT pack is generic or has been modified. I don't know enough about RT to tell if it is configured for a UDA50. The sysgen command complains that a file is missing.

  • We have XXDP, I think version 1.?. I need to look for the ZUD* diags on the pack and try those. If the UDA50 diags are not on the pack we have I will have to try making a V2.5 pack with vtserver.

  • We change the serial console speed setting to 9600 so we don't need to configure the VT220 console every time we work on this system.

  • We removed the fan tray so we could swap the broken fan behind the CPU boards with the middle fan that only cools the RL controller. We found that the faston connectors had fallen off the CPU fan. We had to squeeze the fastons so they would stay on the fan contacts. All three fans are working now!

2/26/11

• • One of the DEC experts that we found said that we need an odd number of SDI cables between the UDA50 controller and the controller in the disk drive. We plugged the black SDI cable directly into the RA90 drive, so we had en even number of cables. We will pull one of the wihte SDI cables from the RA90 cabinet so we can connect an odd number of SDI cables to the drive.

  • We did some more sesearch on the XXDP diags for the UDA50 and RAxx drives. Hopefully the XXDP RL02 pack has the necessary diags on it. If not, we will need to make a V2.5 XXDP pack from an image that is available.

To Do:

• • Fix RY0.

  • See what is on the RA81 disks before installing an OS on them. It would be interesting to install BSD 2.11 UNIX ot Ultrix-11. Installing 2.11 Unix would a require a tape drive so we will need to try the TU81+ in the warehouse. We would also need to install a M8739 KLESI-UA controller for the TU81+.

• Install the 1MB memory board.

• Determine how to get an RA7x drive to spin up without an operator console connected.

  • Either change the mounting locations of the drives and processor in the cabinet, or install the RL02 and 11/44 in the cabinet with the newer RA81.

  • Install the 11/44 CPU and an RL02 disk in a short cabinet with one of the RA81 disks.

  • Remove the RX211 controller and put it in the 11/35.

Done:

Find and install an M8256 RX211 floppy disk controller. Done 1/22/11

Find and install an M7485-YA UDA50 disk conroller. Done 1/16/11

Install boot ROMS. Done 1/16/11

Find working RX02 drives in the warehouse. Done 1/22/11

Try to boot RT11 from an RL02 or RX02. RT11 boots from an RX02 1/22/11

Find working RL02 drives in the warehouse. Partially complete 2/6/11

Install the M7762 RL11 disk controller. Done 2/6/11

Replace fan behind CPU boards. It was a wiring problem. Done 2/19/11

Configure the console for 9600 baud. It was 19200. Done 2/19/11