Restoring a Convergent Technologies NGEN PC/Workstation (Burroughs B25 / Unisys)

In this set of videos, I work on getting a used Convergent Technologies NGEN computer up and working:

Why?

I briefly encountered a few of these Convergent Technologies / Burroughs units back in the 1980s, and I wanted to have one to play with. I had been watching eBay for a while, and when some reasonably priced ones showed up, I decided it was time to acquire one and get it going.

History

Convergent Technologies was a computer manufacturer back in the 1980s, eventually bought out by Unisys. The company produced several different computers over its lifespan, the IWS, AWS, NGEN, and a few mainframe-like computers called the Miniframe and Mightyframe. These machines were marketed by Burroughs at the time, so you might come across these workstations marketed as the Burroughs B20 or B21 (AWS), or the B25 (NGEN).

One thing that separates out the Convergent Technologies NGEN computers from other offerings at the time was the modular design. As shown in the video at the top of the page, the computer is designed based on several different modules that are coupled together electrically by a large header, and mechanically by a latching mechanism. The computer I purchased consisted of a CPU module (CPU, RAM, IO, Video) and a storage module (HDD, FDD). Other modules available at the time would have been dual floppy storage modules, a color graphics module, and disk expansion modules.

Power

Each module is powered by 36 volts DC, and includes one or more built-in DC/DC converters that convert the 36V down to something more useful, such as 12V or 5V. The 36V is distributed from one module to another via the header connectors that couple modules together.

The 36V power is sourced from one or more external power bricks. The bricks connect to the modules using 8-pin SDL connectors. You can power multiple modules from a single power brick if the combined power consumption of the modules is less than the amount of power provided by the bricks. Each module typically has a power requirement printed on the back. For example my CPU module is a 4 and my FDD/HDD module is a 6. This adds up to 10. The single power brick I have is also a 10, so we can power the whole computer from one power brick.

Here is the pinout for the power connector:

Convergent Tech Power Connector

Note that I’ve shown the SDL connector pin-side-up. Beware that if you’ve lost your power cable, you cannot substitute a keyboard cable for a power cable even though they are both SDL cables. 1) They are keyed differently and will not physically plug in, and 2) The keyboard cable has pins 2 and 7 swapped.

Remember that time the power brick caught on fire?

Yes, the power brick caught on fire, we’ll get to that in a moment. In the meantime, here is a picture of the power brick:

The power supply is made by astec. It’s full of lots of beefy transformers, transistors, diodes, and a couple massive heatsinks.

Notice the small white/yellow rectangular part under some wires between four large capacitors and the AC jack on the left. About 10 minutes into my first run of the power supply, it did this.

Power Supply on Fire!

The above picture is a minute after the initial combustion, I didn’t have the camera setup. There really wasn’t a whole lot of fire, just a little bit, but quite a lot of smoke and the office stunk for a day or so. The culprit was an “X2 Safety Capacitor”. These caps go between the line and neutral and are used to reduce AC line interference. Caps from the 1980s are notorious for undergoing spontaneous combustion. Fortunately the rest of the supply was not damaged and it continued to work fine once the cap was removed. It’s not critical, but I’ll make a note to replace it.

Keyboard

At first it looks like your typical PC keyboard, but it’s really quite a bit different. The keyboard has LEDs on some of the function keys. It uses SDL connectors, similar to the ones used by the power cable. The pinout is nothing like a PC connector.

Convergent NGEN Keyboard, with case and foil removed

The CPU in the keyboard is an 8051. The interface communicates to the computer with three IO lines: Reset, Data_In, and Data_Out.

Here is the pinout of the keyboard cable:

Convergent Tech Keyboard Cable Pinout

Video

My CPU module included a monochrome video board. The problem is, this video is output through a DB25 connector on the IO Board, and there is no known pinout for this video connector. The NGEN series has a special purpose-built monitor that connects via the DB25. The keyboard signals are routed through the DB25, and broken out on an SDL connector on the monitor. As of today while I’m writing this, the schematic for the monitor and the schematic for the IO Board and Video Board are unknown to the Internet, so there’s no obvious source to get the data other than experimentation.

I started by pulling the IO Board, so that I could trace what wires connected to the monitor DB25.

Convergent Technologies NGEN IO Board, with metal bracket removed

Here’s what I found by examining all 25 pins of the monitor connector:

SignalPin PinSignal
EC-67 SPEAKER1 25 35 VDC from LM317
EC-29 PWR_ON (Low)2 24G
?3 23Enable LM317
EC-324 22EC-94 – out to kbd
G5 21EC-93 – in from kbd
G6 20G
OUT (74S240N-??) VIDEO7 19G
OUT (74S240N-16) INTEN8 18OUT (74S240N-14) VSYNC
G9 17OUT (74S240N-18) HSYNC
G10 16G
G11 15G
35 VDC from LM31712 14EC-10 – kbd reset
35 VDC from LM317 13  

Note: On the DB25, pin25 is next to pin13 — the breakout board I used kinda flips the rightmost column.

Note: Pins 12, 13, and 25 are connected.

Several of the pins (5, 6, 9, 10, 11, 15, 16, 19, 20, 24) are grounds. This is not unusual. Having many grounds allows them to put power returns separate from signal returns and to give some signals isolated returns. It’s possible that the factory cables might have used twisted pair or coax on the video signals, for example.

Pins 12, 13, and 25 provide some voltage from an LM317 voltage regulator on the IO Board. These pins have approximately 35 VDC on them when enabled. The LM317 appears to be enabled by placing a low signal on pin 23. This could be the monitor’s way of signalling to the PC that it is connect ed and that power should be supplied.

Pins 7, 8, 17, and 18 are all connected to a 74S240N gate on the ioboard. My original hypothesis was that these were for the keyboard circuit, but I can confirm that at least three of them are video! Pin 18 is the Vertical Sync at 60 Hz, ping 17 is the Horizontal Sync at approximately 22 Khz, and pin 7 is Video Out. Pin 8 is Intensity. There is a 19.980 Mhz crystal on the video board, and my hypothesis is that the video bitclock is 19.980 Mhz.

A total of six pins go out to the edge connector on the IO board. Of those, four of them I was able to trace to the edge connector on the video board:

IO PinVideo PinNotes
EC-32EC-32Input to a 74LS240 pin 2 (D1) and
74LS74 pin 11 (B12), 1K pullup
EC-67EC-67Output from 74LS38 pin 3 (F7)
EC-93EC-93Present on video board edge connector,
but unable to trace on video board
EC-94EC-97Present on video board edge connector,
but unable to trace on video board

The video board’s EC-93 and EC-97 pins I was unable to trace to any component on the video board.

Pin 1 is a logic-level audio output. You can use this together with appropriate resistors and a PNP transistor to drive a speaker and the computer will beep as appropriate.

Pin 2 appears to be a power_on signal. It is active low, and pulls itself up to around 4V when inactive.

Viewing on a VGA Monitor

I used an MCE2VGA (https://sites.google.com/site/tandycocoloco/mda-cga-ega-to-vga), which is a small board that converts digital monochrome video to analog vga. It deals with all the grief about converting different refresh rates and different resolutions to something a vga monitor can deal with. In it’s native form, it can handle IBM PC MDA (monochrome), CGA, and EGA.

However, the convergent video output is not the same as IBM PC MDA, so I had to make a few changes:

  • I modified the MCE2VGA dot clock to be 19.980 Mhz instead of 16.257 Mhz
  • I modified the MCE2VGA monochrome vertical sync to be positive instead of negative
  • I disabled the MCE2VGA’s CGA and EGA modes

My custom fork of MCE2VGA is available at https://github.com/sbelectronics/mce2vga-convergent-technologies

Connecting the Keyboard

Combining the tables above, we can establish an interconnect for the keyboard and monitor connectors:

DB-25KBDSignal Name
n/a3, 4+5V
145Reset Keyboard
201, 6, 8GND
217 *Serial – Keyboard to Computer
222 *Serial – Computer to Keyboard

* Important Note: the KBD pins above are on the connector on the keyboard. Keep in mind that the standard coiled keyboard cable swaps pins 2 and 7. If you’re making your own cable, use the pin numbers I specified above. If you’re using the coiled convergent keyboard cable, then make sure to swap these two pins. Refer to the picture further up in the document that shows both ends of the keyboard cable if you have any question.

There is no +5V source on the DB-25. In my prototype I used an external wall wart to power the keyboard. When I make a final board, my plan is to use a switching regulator to step the 35 VDC on pins 12/13/25 down to 5V.

Comments (1)

  1. David Salomon says:

    LOVED reading this post and watching your video! I worked for Burroughs in the early 80s, and another company after that used the B20 series. I had a B26 for some years that I used to support customers when working from home. I wish I hadn’t sold it. Good luck getting yours running again. The B20 series and BTOS were way ahead of their time and were a great platform. Mine was 386 based (B26, if I remember correctly) , had a whopping 20 Meg hard drive, expanded memory (another whopping 256k), a floppy drive, a tape drive and used 4 power supplies.

    I look forward to your next video / post.

Leave a Reply

Your email address will not be published. Required fields are marked *