Retro Fix: Montgomery Ward CyberVision 2001 (Work In Progress)

I recently acquired the all-but-forgotten CyberVision 2001, another late-70s attempt at breaching the home TV-hookup computer market. This system was sold exclusively in the Montgomery Ward catalog, and was featured prominently in a two-page opening spread in its 1978 Spring/Summer edition. The set sported the RCA 1802 CPU and to large capacity was a beefed up rendition of the popular COSMAC Elf kit. The one I received is of multiple levels of despair and with little to no information, schematics and materials to go by online I figure I'll start a blog post to chronicle my experience. This will be a living blog post that I will update each time I'm able to continue troubleshooting this machine. I will fill in with photos and/or videos in time.

Initial Observations and Overview

The set is housed in an all-wooden, blond-tone enclosure. This is unusual compared to the common plastic moldings seen in most systems from the era and gives it much more of a prototype feel than other sets I've worked on. It has a simple woodgrain inspired logo adhered to the front surface, a large volume control potentiometer, simple red power light indicator, and a mini push-button start/reset switch. These are all drilled and affixed directly to the otherwise plain wooden face, with the words VOL, ON and START stamped below them.

As was common at the time including with competing products like the Interact Home Computer, the top of the case included storage space for up to six data cassettes, or as the marketing referred to them — Cybersettes (a term which phonetically garners a very different sort of vision in modern times). The computer has a built-in cassette deck as the only means of data transfer, but sports a 2000 baud speed which was a feat for the time to make for rapid data access compared to the more common 300 baud counterparts. Perhaps most innovative is its use of a stereo head and internal speaker, so that each cassette program can playback studio quality cassette music and narration while simultaneously loading the application data into the computer. Internally the OS checks for a particular byte pattern when the data is loading to know when to stop or proceed, and through various uses of this functionality also enabled retrieving data in real-time off the cassette to continuously update the available 2K of RAM.

There is a hardwired 12V AC power adapter in matching color, with the wires screwed into the face of the adapter. Initial power readings were around 14VAC without load, which was within specs but still inevitably worth replacing with a more modern 12V alternative for the longevity of the machine. The controllers are designated LEFT and RIGHT and are 40-key membrane style rectangles with A-Z, 0-9, ON, OFF, CLR and ENT buttons. They run down a ribbon cable and connect through two separate bulky ports on the back of the machine. I've seen more crude interfaces than these controllers, including the Bally Arcade which only had 24 calculator keys to input and program with. The controllers have a black plastic shell and are embossed with the CyberVision 2001 logo.

The system connects through ordinary RF into a black or white or colored television set. It has a fairly boilerplate tuner internally and you can also toggle between channel 3 and 4.

Basic Hardware Specs

  • 1802 microprocessor
  • 1K ROM
  • 4K RAM (2K program memory and 2K video memory)
  • 2000 Baud Two-Channel Cassette for Syncing Audio and Computer Data
  • 128x96 Color Display

First Layer of Problems

Upon opening the case by removing the four screwed-in base stands, it was immediately observed that the tape belt had decayed over time and was severed. This is extremely common for cassette players from the 70s-80s. A makeshift solution is sometimes to use loom bands or similarly sized rubber bands, but long-term you'll want to replace with a comparable sized cassette belt widely available online. The broken one from the set measured around 25-26cm end-to-end, which would be around 8cm diameter belt. It's pretty easy to tell you have found the proper belt size replacement by playing a tape and listening for audio abnormalities as well as any tape eating symptoms indicative that the belt is not operating to spec.

I found several SRAM chips (P2101 - 256x4) were not well socketed and one was dislodged nearly in half, so reseated these. The RAM is arranged in four logical banks on the board, but I haven't even explored half the ICs yet so am still very unfamiliar with the overall logic board. Since no schematics or service manuals have been shared for this system, it's going to be a lot of manual review to figure out the workings.

There was one large electrolytic hanging off a terminal strip that bridged the cassette interface with the internal amplifier and logic board, I believe a 1000uF but will have to double-check. I replaced it with a new capacitor after poor readings from an ESR meter. Eventually I powered up the machine enough to review its baseline status. A corrupted version of the CyberVision logo appeared on the screen, which was a great indication that the most crucial components all were behaving fine including the CPU and ROM. I suspect there may be a few bad RAM sticks int here, though after manually shorting the reset wires the logo generally would clear up. Also, the reset switch itself was broken off internally, so will need to be swapped.

The cassette initially ate tapes I put in it, and after cleaning and re-adjusting the pinch roller and capstan as well as reworking the belt a bit, I was able to play a cassette to a reasonably decent level of quality and accuracy. But it didn't last long. Suddenly the magic smoke arose from a 22K resistor from the cassette circuit board. I am guessing 22K based on the three bands that weren't incinerated in the poof and checking other areas on the board to find comparably colored ones and then checking the four band color calculators online. I replaced that but another power-up revealed more smoke from the power area of the logic board.  This time I discovered at least one if not several cracked rectifier diodes (WEP170) that make up the bridge rectifier labeled D1-D4 on the board, right near the AC power and large 2200uF capacitor. These diodes have specs of 2.5 Amp 1000V. Era accurate equivalents include HEP170, ECG125, R250. I don't have any 2+ Amp 1000V diodes in my toolbox at the moment so have some replacements on order.

When examining the cassette board closer, I noticed a couple of spliced together wires soldered to a pad that had lifted up along with some of the tracing. I followed the trace and added a bodge wire from the end point back to the spliced wires. Initially I was able to get the audio to play but eventually all I would get is a buzz regardless of playback. I diverted the audio out from the amp board to an auxiliary cable into a dedicated Bose system with similar symptoms, and likewise am not seemingly getting proper readings from the head itself. So this will need to be explored again once I have some more components on hand. I started replacing some of the caps on the cassette board, but there are many of them. The cassette component plugs into the logic board using two three-wire plugs (including ground). One plugs in near the AC power and supplies power to the cassette, while the other is labeled "TAPE TONE" and is where the actual data side of the cassette routes (the right channel). The audio side routes instead to an amplifier and then to the output speaker mounted inside the cabinet.

If I can get the main logic board back in operation, another experiment will be to wire up some feed directly to the input line on the logic board. If I'm ever able to read the data and load a program, that'd be a massive milestone even before fully fixing the cassette side of it. The cassettes themselves are rare and not archived anywhere online. I have a few that I am working to digitize in their true stereo format. Unfortunately my sound card and interface also suffered a malfunction in recent months so I am using work-arounds at the moment but have found success routing from a 1980s Panasonic stereo to my Focusrite 2i2.

A Note About Disassembly

Although it's easy enough to remove the wooden case by simply unscrewing the four rubber bases screwed in, it becomes a bit more tedious to get at the underside of the logic board. You will want to desolder the wires from the LED and reset switch (or from the board side) so that the wooden cabinet doesn't need to tag along. Then there are four bolts in each corner of the board that you can easily unscrew the nuts and washers using your fingers or a pliers. However, the board still won't come out. There are two 7805 voltage regulators on the underside of the board, riveted into the base case itself (originally a NEC 14305 and a NEC 7805H, TO-220-3L). The rivets need to be removed for these to release from the base. In my case a drill bit from inside out to remove the inner core, then a pliers to pull out the remnants of the rivet worked well. Even then, there's also a plastic riser tab in the center that you have to pop either out of the board or out of the case, again a pliers can be handy.

With all of the above taken care of and the two cassette connectors removed, you should be able to isolate the logic board to replace most components. The back accessories will still be connected so if you need absolute access to the underside of the board in ways that you can't manipulate with the peripheral connectors attached, some more work will be necessary. But I had no trouble propping the board up and even upside down with the back peripherals still affixed and could desolder and replace components easily.

Voltage Regulators, Revisited

The Cybervision 2001 uses two of the classic 7805 +5V voltage regulators. As mentioned, they are factory-riveted into steel base plate of the system. This is for good reason as these components are famous for getting blister-scorching hot in a matter of seconds if not properly cooled with an attached heatsink. These ones convert 12-14V input into 5V output.

Unfortunately the permenantly riveted solution makes the entire logic boar inaccessible for repairs. After drilling out the rivets, I found that using 6-32 x 1/2" machine screws with nuts made for an adequate and easily removable alternative. I applied new thermal paste to the base of the regulators before bolting them down.

In the process, I discovered one leg of the 7805 nearest the failed bridge rectifier circuitry was severed under the board. And the other had a loose solder connection on the unregulated-to-center-pin .1uF 12V ceramic cap. I removed both voltage regulators and even the one still in tact seemed brittle and was registering unusually on my multi-tester, so I exchanged both for new 7805s.

Without these regulators bolted down to the base, the heat build-up on these grows drastically fast. Within 5-10 seconds their thermal protection kicks on and immediately the output drops, simultaneously underpowering everything attached. If you're using a device that starts to exhibit strange video artifacts and loss, tearing or other symptoms it is always worth checking these sensitive voltage regulators and make sure the input and the output are steady and within spec.

Modern Alternatives

These days, several incredible drop-in replacements for 78xx series regulators exist. Through the use of buck converters or mini DC-DC adjustable output circuits, we can achieve the same result with much more reliability and none of the heat. Most of these replacements allow you to specify the output voltage anywhere from 1.8V through 12V and offer even greater flexibility with input voltage, so can be used for +5 and +12V rails common on vintage equipment. The sets boast an impressive 97% or higher efficiency rate.

I will be swapping the regulators with these modern counterparts after evaluating several different styles. This includes:

  • R-78B5.0-1.5L (available via Digi-Key): Described as a Linear Regulator Replacement DC DC Converter 1 Output 5V 1.5A 6.5V - 18V Input this is a drop-in, pin-compatible replacement for the 78XX series linear regulators in a similar TO-220 form factor. At the time of this writing, they are $12.97 each, so considerably higher than a pack of 7805s but also a vast improvement.
  • 78xx Replacement (SD8942) - This is a CC-BY-SA 3.0 replacement designed and shared by Stefan Wagner. It mimics the basic regulator style circuitry in the same footprint as a TO-220 linear, but uses a modern DC-DC converter IC. It supports variable voltage by soldering different resistors as detailed in the chart. A 5V converter typically uses a 110k and 15k resistor depending on the voltage reference (typically 0.6). The PCBs can be made via JLCPCB for under $10 for 30 of them using the provided design/Gerber, and I imagine that is similar via PCBWay. The majority of necessary surface mount components will already be in one's toolbox, and even the step down ICs are under $1.50 each. I look forward to constructing these as very cheap DIY converters.
  • 5V Mini Voltage Reducer (example on Amazon): There are endless variants of this sort of component used largely in Arduino and DIY modern electronic projects. They are not form-fitting to be direct replacements for legacy regulators like the ones above are, but still have all the same options and pin-outs so can easily be worked into old projects as well. Most of them have adjustable (but fixed) voltage in similar ranges to the ones above. A 10 pack of these will cost around $12, not much more than a pack of 7805s.

Will update my findings in the future.

External Resources for Further Reading

Blog Update Notes

2023/01/26: Added new section stub about voltage regulators.