cw@madvax.UUCP (Carl Weidling) (07/20/87)
Quite a few people responded to my question about how CRT memories worked. Some posted to the net and I presume were read by everyone interested. Some people frankly admitted that they were speculating based upon what they knew about how CRTs work and are used nowadays and some wanted to know what I found out so that they could see how accurate their guesses were. I tried to respond to all the email I got, but this annoying thing called work interrupted me a bit and I may have missed some of you, sorry if I did. Also, some I couldn't figure out how to respond to. Nigel Topham sent me mail, I tried to respond and got my mail bounced back but I see that he posted his information to the net in general. Bert Hutchings @aiva.ed.ac.uk responded privately and included permission to post and I think his info may be of interest. Here is a slightly abbreviated version of what he posted: I was lucky enough to attend a talk by Prof. Williams at Newcastle-on-Tyne in 1963 or 64, when he gave some background to his invention of the CRT memory. Maybe they were using the CRT just as a display device at first, and somebody was monitoring the beam current while the display was constant. That was the essence of it, anyway; they spotted that the current dipped or spiked as the beam approached a previously-displayed bright or dark patch. The development was obvious - detect the fading record, and amplify it. The nicest thing was doing this all in one sweep, switching the beam from detect mode to re-record mode once for every assigned bit position on the screen matrix. Nowadays, we would classify this as a variation on a dynamically-refreshed RAM, I suppose. ---------------- in a second bit of mail from Mr Hutchings: ...but I forgot to answer your question about readout. I'm pretty sure it was bit-serial, from one complete scan line, and this explained the common colloquialism 'line' for 'word', long after CRT memories were defunct; but I can't recall whether you could get a word on demand, or had to wait till the slower vertical sweep reached it. I suspect the latter, and that that was why they only parked about 32 words on each CRT even though they could have packed in an awful lot more. 32 CRT's backed up by a drum would seem to have given a quite acceptable cost and performance in those days. ----------This morning I got the following from Rob Lake, to whom I haven't ----------responded but I presume he will not mind my posting. ---------- The CRT memories worked with 2 electron guns. One gun (read/write) and one "flood" gun. The write gun would illuminate a point on the screen of the CRT, much like the CRT storage tubes in oscilloscopes. The flood gun would provide a lower current sweep (sometimes done with a grid behind the face of the CRT instead of a gun) to sustain the flow of electrons to the previously illuminated point. A resistor ran from the screen to ground. When the read gun (could be the same physical gun as the write gun, but run at a lower current) hit a previously illuminated point, the current thru the resistor would be higher than when the gun hit a non-illuminated point. So --- you got memory. Rob Lake References in some older Tektronix and H-P Technical Journals. --------------------------- Incidentally, I read the whole book "The Illustrated History of Computers", and was, I think, a good job. It covered a lot of ground and I suppose one could not expect it to go into too much detail on the technologies of the different eras, from the abacus and first adding machines, through Babbage's engines, to the modern computer. The writer didn't make any big technical boners that I was aware of, and was for instance, careful to explain why he did not regard ENIAC as a genuine computer in the modern sense because it wasn't a stored program type. Also, he gave attention to the human side, the politicking and misunderstandings and personalities of the people involved in the history. For instance, he almost makes all the murky interactions between government, Moore School of Engineering, Von Neumann, Eckert, Mauchly, Atanosoff, and others, understandable. Regards, Carl Weidling