The Card That Made the Apple II Serious — Wise OwlSkip to main content

Part 1 of 2. This part covers the Videx VideoTerm’s hardware architecture, the MC6845 CRTC, and why slot 3 is architecturally special. Part 2 covers the rendering pipeline, C8 ownership, and the Pascal boot hang.

In 1977, the Apple II shipped with 40 columns of text. That was fine for BASIC programs and games. It was not fine for WordStar, VisiCalc, or the CP/M business software that was reshaping the industry. A serious word processor needed 80 columns. A spreadsheet with real data needed 80 columns. Anyone who wanted to use an Apple II for actual work needed 80 columns.

The Videx VideoTerm, introduced in 1980 and manufactured in Corvallis, Oregon, was the card that solved this problem — and dominated the market until Apple built 80-column support into the IIe in 1983. It was the most widely supported 80-column card for the Apple II and II+, compatible with Apple Pascal, WordStar, and virtually every serious business application of the era. Understanding why it worked the way it did requires understanding a chip that has largely faded from memory: the Motorola MC6845 CRT Controller.

The Display Problem

Before getting to the Videx card itself, it helps to understand what Apple II video actually looked like.

The Apple II+ outputs NTSC composite video — a single analog signal designed for 1970s consumer televisions. It works, in the sense that you can see the screen. But NTSC composite has color bleeding, chroma artifacts on text, and a horizontal resolution that is fundamentally limited by the bandwidth of the signal. The 40-column text mode that the stock Apple II produces is legible on a period-correct TV. On anything else it is a reminder of how much display technology has changed.

The A2FPGA bypasses composite output entirely. Rather than capturing and upscaling the analog signal, it monitors the Apple II’s video memory — the text pages and hires graphics pages that the Apple II’s video circuitry reads to generate its composite output — and generates HDMI directly from those memory contents. The FPGA reads the same bytes the Apple II would use to build its composite signal, and renders them into a clean 720×480 digital frame. The result is text and graphics at correct pixel boundaries, with no analog smear.

Adding the Videx emulation extends this pipeline. The Videx card has its own 2KB of video RAM, separate from the Apple II’s memory map, that stores the 80-column character grid. When 80-column mode is active, the FPGA substitutes its Videx VRAM rendering for the Apple II’s native 40-column text — producing sharper 80-column output on a modern HDMI display than any physical Videx card connected through the Apple II’s composite output ever produced.

The MC6845

The MC6845 is a programmable CRT controller from 1977. Its job is to generate the timing signals that drive a CRT display: horizontal sync, vertical sync, and cursor position. It is not a video chip in the modern sense — it generates no pixel data itself. It tells the host system where the electron beam is and trusts the host to supply the right character data at the right time.

The 6845 exposes 18 registers through a two-port interface. You write a register number to the address register, then read or write the data register. The registers control everything from the number of characters per line to the vertical total to the cursor blink rate. The registers are largely write-only on the original MC6845, though the Hitachi HD6845SP used on the Videx VideoTerm is a Type 1 variant that allows full read-back — a detail that matters for software that checks its own configuration.

If you look at the physical Videx board, the HD46505SP is the largest chip, front and center. The two 2716 EPROMs above it — labeled “INVERSE (C)1981 VIDEX” and “STD. 7X9 E (C)1981 VIDEX” — are the two halves of the character ROM. The “VI-FIM-500 (C)1982 VIDEX” chip is the firmware ROM. A small army of 74LS-series logic chips handles address decoding, bus buffering, and timing.

The FPGA emulation replaces all of it. Every chip on that board — the HD46505SP, both character ROM EPROMs, the firmware ROM, and all the glue logic — lives in about 250 LUTs and 160 registers in the GW2AR-18C fabric.

Why Slot 3 Is Different

The Apple II has 7 expansion slots. Electrically, they are almost identical — each provides the same set of bus signals plus a slot-specific I/O select and device select. But slot 3 has a hardware quirk that makes it unlike every other slot in the machine: it controls the INTCXROM internal ROM mechanism.

When the CPU accesses memory in the $C300–$C3FF range (the slot 3 ROM space), the Apple II hardware checks a soft switch called SLOTC3ROM. If SLOTC3ROM is 0 (the default), accessing $C300 sets a flag called INTC8ROM, which routes the $C800–$CFFF expansion ROM space to the Apple IIe’s internal 80-column firmware instead of external cards. On the Apple ][+, SLOTC3ROM doesn’t exist,...