 The Sol-20 has 2K of onboard static RAM memory beginning at address $C000 and 2K of ROM (the "Personality Module") at $C800. The RAM is divided between a basic "scratch space" in the lower 1K and the video memory in the upper 1K, a size sufficient to store the 16 row by 64 column character display. With this basic configuration, one can run the SOLOS monitor and execute simple commands, like DUMP and ENTER to examine and change the memory. It can also be used as the "intelligent terminal" it was originally designed as. Doing anything else requires additional memory to be installed in the S-100 expansion bus, like any other S-100 computer. This Sol came with a Central Data Corp 64K RAM card. The card had four banks of 4116 dynamic RAM chips installed. This was my best option for covering the entire 64K addressable space of the 8080. Since the card came with the computer, I thought the odds were that it could be configured to deactivate the memory in the 4K space of the system RAM and ROM. Indeed, there is a series of jumpers, or "address selection straps," on the top of the card that can be used to set the address range and any "holes" in the memory space. The first four jumpers are set as: top row 1100, bottom row 0011, which is correct for activating the board over the entire 64K. Next, a "deselect" block is set: 0001. This should set the last 16K for deselecting memory in 2K blocks. The jumpers are 1100 0000 to deselect $C000-C3FF and $C400-C7FF, while keeping $D000-FFFF active. The DIP switches IC15 should be 10001101 (1 = closed, 0 = open) for 8080 processors. The soldered jumpers at IC16 were also correct: 00101010  Thinking that I would probably have to troubleshoot the card, I decided to start the machine with the cover off and the CDC card in the top (vertical) expansion slot of the S-100 bus. This slot is a nice and thoughtful feature that takes the place of a normal expansion card (like the N8VEM one here) for testing and repairing cards. My choice of setup was fortuitous. As soon as I powered on the computer, there was a loud pop from the lower left of the CDC card and a candle-length flame shooting up vertically along the board. I quickly powered down and blew on the flame until it was extinguished. A tantalum capacitor had ”s-ploded.” Did I mention I tested the tantalums for shorts? Such tests are evidently not without their “short-comings.” This one was my first tantalum to explode. It did not disappoint, but the thought of the partially wooden Sol-20 catching fire gave me pause. I replaced the tantalum with an electrolytic rated at 4.7uF, 50V I had on hand from the passive parts bin. The other capacitors, labeled 33uF on the schematic, were actually 4.7uF, 35V. In the meantime, I powered up the Sol-20, which had luckily sustained no damage. Perhaps the RAM board fared as well, too. This time, I took the N8VEM extender card and used that—a bit awkward extending the RAM board two heights above the Sol case, but I wanted to take advantage of the N8VEM's fuses in case another tantalum released its magic smoke.  The next power-up was fine, although I did don a pair of goggles. Using the SOLOS monitor, I entered the memory test the program in the CDC64K manual. The memory seemed to check out fine and I operated the machine for another 30 minutes without problem. I've since installed the CDC card in one of the normal expansion bus slots. The experience has me thinking that I should build a simple rig for testing S-100 cards. There is certainly the danger of harming machines like the Sol-20 and IMSAI with boards of unknown condition. The Sol, with its integrated PC motherboard, may be more susceptible to damage than something like the IMSAI's backplane design. With 64K of memory installed, I've started putting the machine through its paces. The serial port works. With a null modem cable, I have the Sol connected to a Raspberry Pi with an RS-232 shield. Jim Battle's website provides a number of programs that can be transferred over the serial port. Like the Apple II, the Sol can redirect serial data to act as keyboard input. The "ENT" file format on the sol-20.org is simply the format as if the text were entered directly to the monitor. At 1200 baud, it takes a minute or two to load up a ~2K program like Target. Target was originally written by Steven Dompier for the VDM-1 video board, which was designed to provide direct video out for S-100 computers like the IMSAI and Altair. The Sol-20 basically incorporates the VDM hardware. You can see from the splash screen that it "includes sound effects." There is no sound hardware to in the Sol-20. A nearby AM radio provides the sound using methods first reported by Dompier at the Homebrew Computer Club and later in the Peoples Computer Company newsletter and the February 1976 issue of Dr. Dobb's Journal.  |
PTCO Sol-20 >