The Hi-Mound UFO & the Autokeyer Era — Vol 2: From the Bug to the Message Keyer

How electronics took over the timing of Morse — the iambic keyer, the Curtis chip, dot/dash memory and the Mode A/B argument — and the Katsumi MK-1024 that sits next to my UFO and can remember a whole message.

2.1 The problem the electronic keyer solved

A straight key asks your hand to time everything. A bug (the semi-automatic key in its own dive) mechanizes the hardest part — the string of dits — with a vibrating pendulum, but the dahs are still yours and the whole thing imparts a distinctive mechanical swing. The electronic keyer finishes the job: it generates both perfectly-timed dits and dahs from a paddle, at a speed you set with a knob, and it does something no mechanical key can — it remembers.

Two ideas define the electronic keyer, and both matter for understanding the UFO and the Katsumi:

Self-completing elements and dot/dash memory. Tap the dit paddle and the keyer finishes a full, correctly-timed dit even if you let go early; better yet, if you tap the other paddle while an element is still going out, the keyer remembers it and inserts it next. This is what lets a keyer forgive imperfect timing — the machine cleans up your hand.
Iambic (squeeze) keying. With a dual-lever paddle, squeezing both paddles at once makes the keyer send alternating dot-dash-dot-dash automatically. It’s efficient and, once it’s in your fingers, fast — the letter C (dah-di-dah-dit) becomes a single rocking squeeze. (This is exactly the trick the single-lever UFO-001 cannot do — see Volume 1. The Katsumi below can.)

2.2 The chip that made it cheap: Curtis

For most of the 1960s an electronic keyer was a box of discrete logic — real estate and soldering. That changed in 1973, when Curtis Electro Devices (John “Jack” Curtis, K6KU) shipped the 8043, the first keyer-on-a-chip: a single IC that did the dit/dah timing with dit memory. Two years later, in 1975, the pin-compatible 8044 added dah memory, which made true iambic keying practical on one chip. These little Curtis ICs are the reason electronic keyers were everywhere by the late 1970s — they’re the brains inside a whole generation of keyers and, functionally, the same job the UFO’s handful of ICs is doing.

The homebrew world had its own landmark at the same moment: the WB4VVF Accu-Keyer (James Garrett), published in QST in the mid-1970s, a build-it-yourself iambic keyer with dot and dash memory and automatic character spacing that became the reference design a generation of hams soldered up on a kitchen table.

2.3 The Mode A vs. Mode B argument

If you hang around CW operators long enough you’ll hear a mild religious war about iambic Mode A vs. Mode B, and it’s worth knowing what it actually is, because it’s a real behavioral difference, not just jargon:

Mode A (the original Curtis behavior): when you release both paddles, the keyer finishes the element it’s sending and stops.
Mode B (the Accu-Keyer behavior): when you release both paddles, the keyer sends one additional, opposite element beyond what you were squeezing. It started life as another maker’s design quirk and was later adopted as a selectable feature.

Neither is “correct”; operators simply learn one and their muscle memory locks to it. Curtis eventually made it moot: the 8044ABM (1986) let you select A or B and threw in a speed meter. The reason to care is practical — if you learn on a Mode B keyer and sit down at a Mode A one, your machine-sent letters suddenly grow or lose a dit, and it feels like the floor moved.

2.4 The message keyer — remembering a whole transmission

Everything so far still forms the code in real time from your paddle. The last step in this dive’s evolution adds a memory big enough to hold not one element but a whole message: you key “CQ CQ CQ DE N0SWN N0SWN K” once, the keyer records it (WRITE), and thereafter a button press plays it back (READ) at your set speed, perfectly, as many times as you like. For anything repetitive — calling CQ, sending your callsign, a contest exchange, an RST report — a message (memory) keyer turns a paragraph of tedious, error-prone sending into a button. That is the machine sitting on my desk next to the UFO.

2.5 The Katsumi MK-1024 Message Keyer

The Katsumi MK-1024 is the UFO’s perfect foil: where the UFO is a beautiful single-lever wobbler that automates the dit string, the Katsumi is a no-nonsense grey box that is a true iambic keyer and a four-message memory. Everything I can tell you about it below comes from its own instruction manual (a genuine primary source), not from lore.

Figure 1 — The Katsumi MK-1024 Message Keyer — front panel. Top row: the four message buttons A/B/C/D plus ALL, with STOP and PWR. Bottom row: WRITE/READ, AUTO/SEMI, the 6–60 WPM SPEED dial, the squeeze/singl… — Figure 1 — The Katsumi MK-1024 Message Keyer — front panel. Top row: the four message buttons A/B/C/D plus ALL, with STOP and PWR. Bottom row: WRITE/READ, AUTO/SEMI, the 6–60 WPM SPEED dial, the squeeze/single key jacks, and MONITOR. (Katsumi, Japan, late 1970s.)

Reading the panel and the manual together:

The name is the memory. “1024” is the size of its RAM in bits — a single NEC μPD2102 1024-bit static RAM. That memory is organized as four independent 256-bit message slots, A / B / C / D (each holds roughly 30 characters), or combined into one 1024-bit message (~120 characters) via the ALL button. So it stores four messages, not five — “ALL” isn’t a fifth slot, it’s the four chained into one long one.
WRITE / READ. Flip to WRITE, key your message into a slot; flip to READ, and the buttons play it back.
AUTO / SEMI. AUTO is the normal mode — automatic dits and automatic dahs. SEMI gives you automatic dits but manual dahs (and doubles as a tune-up mode). Notice the family resemblance to the UFO’s “dahs auto or manual” switch — same idea, different execution.
A true iambic squeeze keyer. The panel says “With SQUEEZE KEYER” and provides dual key levers; it does full dot and dash memory iambic sending — the thing the single-lever UFO can’t do.
Speed: 6–60 WPM. The dial is marked 6, 20, 30, 40, 50, 60 — so it’s 6 to 60 WPM, not “20–60” (the 20 is just the lowest numbered mark above the 6). This is genuine words-per-minute, unlike the UFO’s characters-per-minute BpM.
Output and power. A 2SB546 transistor switch (150 V / 2 A) plus a built-in high-speed relay rated 700 V / 500 mA for keying a high-voltage cathode line — again the era’s telltale, a physical relay for tube gear. It runs from AC mains (100–240 V) or an external 9–14 V DC supply for mobile use, with a monitor/sidetone speaker and an end-of-message LED.

Figure 2 — The Katsumi MK-1024 — the memory-keyer internals and controls that store and replay the four canned messages. (Katsumi, Japan.)

Figure 3 — The Katsumi MK-1024 — additional view; the μPD2102 RAM at the heart of it is what the "1024" in the model number refers to. (Katsumi, Japan.)

2.6 Where the era went

The message keyer was not the end of the line, just the last purely-dedicated box. The four canned slots of a Katsumi became the dozens of memories in an MFJ or Idiom-Press keyer, then the messages stored in the radio itself, and finally the F-keys of a contest-logging program that sends flawless CW straight out of the computer — the last rung on the ladder from Volume 1. The job never changed from what Morse and Vail set out in 1844; only the question of who holds the timing moved, hand by hand, into the machine.

And that is exactly why I keep the UFO on the desk. It is the moment in that whole march when a Japanese company decided the machine that holds your timing should also be beautiful — a chrome saucer with a glass paddle — and I think they were right.

2.6.1 Sources (Vol 2)

Katsumi MK-1024 Message Keyer instruction manual (primary) — four 256-bit memories A–D (or one 1024-bit via ALL), NEC μPD2102 RAM, WRITE/READ, AUTO/SEMI, 6–60 WPM, squeeze/iambic with full dot & dash memory, 2SB546 (150 V/2 A) + 700 V/500 mA relay output, AC or 9–14 V DC. (Reprint hosted at dl0bn.de/dc7xj.)
History of the Curtis keyer chips (8043, 1973 — dit memory; 8044, 1975 — dah memory / iambic; 8044ABM, 1986 — selectable Mode A/B + speed meter): testparts.blogspot.com “History of Curtis Keyers”; Curtis 8044 application note (users.ox.ac.uk/~malcolm/radio/8044print.pdf).
WB4VVF Accu-Keyer (James Garrett), QST, mid-1970s — the reference homebrew iambic keyer and origin of the “Type B” behavior. ⟨exact QST issue not pinned; commonly cited 1973–75⟩
Iambic / squeeze keying and dot-dash memory concepts: ARRL operating references; hamradioqrp squeeze-keying notes. Cross-references: Keys & Keyers → Paddles & Electronic Keyers (the iambic Mode A/B timing diagram) and → Semi-Automatic Bugs (the mechanical ancestor).