core-logic clock / was Re: TTL homebrew CPUs

Thyratron ring counters, in conjunction with the Nixie tubes themselves (the Nixies were specially designed to be an active part of triggering the next Thyratron in the ring) in the first commercially produced electronic calculator, the Anita Mark VII/Mark VIII (the Mk7 sold in Europe for a short time, quickly superceded by the Mk8). These were small Thyratrons, about the size of a pinkie finger from the first joint to the fingertip. There were also used a small number of vacuum tubes, for specific logic elements which required inversion or buffering. The main clock phases were generated by a Dekatron (1 in the Mk8, 3 in the Mk7). There were lots of Thyratrons in this machine. The machine had a capacity of 12 digits, which meant 120 Thyratrons just for the ring counters. There were other boards full of thyratrons for various logic and clocking functions, for a total of over 200 in the machine. It is very cool to see one in operation without the cover on. The thyratrons do emit an orange glow when triggered. While doing calculations, The thyratrons rapidly zip on and off, one at a time, in the ring-counter circuits. Others seem to bounce around in differing patterns, depending on the type of operation being performed. On the subject of making a magnetic core logic clock...it might be possible, but the problem with core is that reading it is destructive. Circuitry would have to be built using a magnetic core flip-flop (which are possible using core logic gating functions) to retain the state of the read bit, and then a cycle initiated to "put the bit back". This would have to be done as each bit is read out of the core array. If it was read four bits at a time (4 planes), then you'd need 4 core flip-flops to retain the state for rewriting back into core. It's very unlikely that the direct output of the core could drive nixie tubes. The core outputs are typically very low voltage, low current pulses that require either pulse transformers, along with further amplification to make into usable logic signals, or all-transistor sense amplifiers. I suppose that some kind of core-based amplifier could be used, but it would be more like a transformer than a pure core, with a small primary winding, and a large number of secondary windings. The logic necessary to handle the timekeeping, the core memory read/modify/write cycle, and BCD-to-1 of 10 decoding to drive the Nixies would take a lot of core-based logic. I think it would be possible to do it, but you'd definitely need some transistorized circuitry for buffering, amplification, and clocking (though a free-running core-based clock generator might be possible). Many early electronic calculators utilized core memory as the main memory for the machines, though surrounded by a lot of diode/transistor logic for the rest of the machine. The core just stored the working registers. As far as I know, no calculator was ever built that utilized core for everything. Dr. An Wang (the inventor of magnetic core memory) did a lot of experimentation with core logic, and had patented it. But, IBM offered him big bucks to assume the patent, and he took the money. Later, Dr. Wang realized that he got ripped off by IBM, and vowed that he'd build a business (Wang Laboratories) that would put IBM out of business. He made a good run for it for a while, but never came even close. Rick Bensene The Old Calculator Web Museum http://oldcalculatormuseum.com