cw@madvax.UUCP (Carl Weidling) (07/13/87)
Yesterday I bought a book called "Bit by Bit, An Illustrated History of Computers", by Stan Augarten,Ticknor & Fields,NY,1984. It seems like a good book from what I've seen glancing through it, but, I was reading about the design of the Mark I in Manchester,England where it says: "He [F.C. Williams, the project's chief engineer] hit upon the idea of employing ordinary cathode ray tubes [to solve the problem of internal memory storage]...Their operating principle was quite simple; "guns" in the bases of the tubes shot positively and negatively charged electrons at the faces of the tubes, thus storing bits in the form of charge spots, which, by the way, were quite visible to the eye." Well, I have a problem with "positively and negatively charged electrons", but also, with how this can be memory. How long did the charges last? were they refreshed? How was the memory read after being stored? If anybody out there knows this stuff, I would be appreciative if you could enlighten me. My limited impression is that the book is better than this little excerpt would make it appear. Regards, Carl Weidling
bobw@wdl1.UUCP (Robert Lee Wilson Jr.) (07/14/87)
The only machine I personally saw using WIlliams tubes for storage was the ORACLE (Oak Ridge Automatic Computer and Logical Engine... They probably spent almost as much time on the name as on the rest of the system :-)). The system there had the tubes covered so I can't speak to the visibility of the bits, but based on how it worked they _ought_ to have been quite visible except that the sense electrode over the screen may have been opaque. Fundamentally what happens is this: Steer the beam (using electrostatic deflection plates) to a spot on the tube face corresponding to memory address. Now turn the beam on or don't turn it on: Let's assume turning it on corresponds to writing a 1. There is now a negatively charged spot wherever we have written a 1. (The charge does leak away. I don't know just how fast, but have the impression refresh was not needed as soon as for DRAM chips.) To read, write a 1 at the location. If there was previously a 0, there will be significant beam current charging the spot, while if there was previously a 1 the beam will be diffused by the negatively charged spot it was aimed at. There is a difference in charging current which is sensed by a conductive plate on the outside of the face of the tube, essentially one plate of a capacitor with the charges on the inside of the face being the other plate(s). Whenever you read a 0 you have to go back and "write a 0" since by reading you wrote a 1. To do a periodic refresh you just scan a whole raster and write back whatever you read. Note: I haven't said how to write a 0 in an already charged location, since I can't remember for sure. No, they didn't shoot positive electrons! I _think_ you sort of write a ring of charge around the spot, "smoothing out the charge hill," but I would be interested in anyone who can say for sure. The tubes were only special in that they were carefully selected for freedom from blemishes in the faceplate. As a kid I always dreamed of getting together a bunch of 5BP1's and building a system. I saw the ORACLE when I was 13, and may have screwed up the description somewhat, but I believe that is about right. I remember an early book _The Design of Electronic Digital Computers_, author I think was Richards or Richardson, which had a chapter on this as well as other storage techniques of the day. (Mercury tank audio delay lines for recirculating bit streams, etc.) I haven't seen a copy lately but assume it must still be in some libraries. Disclaimer: The usual copout.
agn@unh.cs.cmu.edu (Andreas Nowatzyk) (07/14/87)
The Williams tube uses the same principle that is used in storage scopes and some old Tektronix graphics terminals: The screen uses a phosphor with an extremly high resistance, essentially an array of isolated electrodes. This screen is flooded uniformly with low-energy (say 400V) electrons. Due to the emission of secondary electrons, each point of the screen can be in (at least) 2 stable states: a) dark: charged to >= +400V so that no flood-gun electrons can reach the point due to electrostatic repulsion. b) light: charged to a more negative voltage so that flood-gun electrons reach the point. The potential is stabilized at this point due to the emission of secondary electrons so the the net current is 0. You can change the state of each point by using a second electron gun that is focused and uses a higher acceleration voltage (say 2KV). It is easy to see the a->b change, but the reverse is also possible by playing tricks with the screen and flood-gun potentials. Readout is done by scaning the screen with a low-intesity beam and detecting the change is the secondary electron current. Addressing was done with the X/Y deflection of the readout beam. -- Andreas
weaver@prls.UUCP (Michael Gordon Weaver) (07/14/87)
In article <602@madvax.UUCP> cw@madvax.UUCP (Carl Weidling) writes: > > Yesterday I bought a book called "Bit by Bit, An Illustrated History >of Computers", by Stan Augarten,Ticknor & Fields,NY,1984. > It seems like a good book from what I've seen glancing through it, >but, I was reading about the design of the Mark I in Manchester,England >where it says: "He [F.C. Williams, the project's chief engineer] hit upon >the idea of employing ordinary cathode ray tubes [to solve the problem of >internal memory storage]...Their operating principle was quite simple; >"guns" in the bases of the tubes shot positively and negatively charged >electrons at the faces of the tubes, thus storing bits in the form of charge >spots, which, by the way, were quite visible to the eye." > Well, I have a problem with "positively and negatively charged >electrons", but also, with how this can be memory. How long did the charges >last? were they refreshed? How was the memory read after being stored? > If anybody out there knows this stuff, I would be appreciative >if you could enlighten me. My limited impression is that the book is better >than this little excerpt would make it appear. >Regards, >Carl Weidling This is from memory, and also my interpretation of what I have read, so it may not be correct, but here goes. The electron beam of a cathode ray tube is of course negative. When an electron hits the phosphor on the inside of the screen, it knocks out several electrons leaving a positive ions. Eventually, these electrons are replaced, causing the glow that is seen on the other side of the screen. On the wall of the tube between the gun and the screen is a secondary anode used to return to ground excess charge given off by the electron beam so that the net charge does not grow too large. When used as memory, the current to this anode is measured (presumably converted to a voltage via an appropriate resistor). When the beam is writing on to a positive spot of phosphor, slightly less current will be going to this anode, because the positive ions have no valence electrons to be removed by the electron beam. So we can tell which parts of the screen were written to last time. The readout is destructive, and so the bits must be re-written each pass. The charge lasts for about the same time the phosphor glows, which is on the order of 30 milliseconds. I believe the largest number of bits per cathode ray tube was about four thousand. I have been unable to acertain how the regeneration is done, since not more than a few bits of data could be buffered e.g. by tubes. One possibility would be that the data was being written to one tube as it was read from another. The reports I have read suggest that individual bits were written back by the same beam that read them, somehow cleverly using the fact that the positive areas are somewhat larger than the beam that wrote them, but this may be incorrect. -- Michael Gordon Weaver Usenet: ...pyramid!prls!weaver Signetics Microprocessor Division 811 East Arques Avenue Sunnyvale, California USA 94088-3409 Phone: (408) 991-3450
howard@cpocd2.UUCP (Howard A. Landman) (07/14/87)
In article <602@madvax.UUCP> cw@madvax.UUCP (Carl Weidling) writes: >I was reading about the design of the Mark I in Manchester,England >where it says: "He [F.C. Williams, the project's chief engineer] hit upon >the idea of employing ordinary cathode ray tubes [to solve the problem of >internal memory storage]...Their operating principle was quite simple; >"guns" in the bases of the tubes shot positively and negatively charged >electrons at the faces of the tubes, thus storing bits in the form of charge >spots, which, by the way, were quite visible to the eye." > Well, I have a problem with "positively and negatively charged >electrons", but also, with how this can be memory. How long did the charges >last? were they refreshed? How was the memory read after being stored? The description is wrong on a couple of counts, if my memory serves me. First, the tubes were NOT what we would today call an "ordinary cathode ray tube", but rather storage tubes similar to those used on certain Tektronix oscilloscopes about 10 years ago. Once a bit (pixel) has been lit up, it stays lit up. Second, I don't think it was charges but rather sustained currents; this implies no real time limit. (I'm not sure about this, they might have used refresh.) Reading was accomplished by putting a weak signal into the pixel and measuring the output current(?), which varied depending on the state of the bit. I'm not sure how they erased anything! Also, the voltage required to write a bit resulted in a burst of intense light as the bit was written, much brighter than the light from a bit simply staying on. This allowed you to watch your program's memory accesses; random access, linear scan, and infinite loop all have their characteristic appearances. The first real-time, graphical program debugging tool! You can get a similar display today on any 16 bit computer with 2 8-bit D/As and an X-Y oscilloscope. Hook one D/A to the upper 8 bits of the address bus and the Y, and the other to the lower 8 bits and the X. Adjust gain, and voila! -- Howard A. Landman ...!{oliveb,...}!intelca!mipos3!cpocd2!howard howard%cpocd2%sc.intel.com@RELAY.CS.NET "Sometimes a light's all shining on me, other times I can barely see"
news@santra.UUCP (news) (07/15/87)
In article <602@madvax.UUCP> cw@madvax.UUCP (Carl Weidling) writes: ... >... how this [ CRT tubes ] can be memory. How long did the charges >last? were they refreshed? How was the memory read after being stored? >Carl Weidling I once had a tube where characters were visible for several seconds after the screen was blanked... Juha Kuusama, jku@kolvi ( ...!mcvax!enea!tut!kolvi!jku )
bct@its63b.ed.ac.uk (B Tompsett) (07/15/87)
In article <602@madvax.UUCP> cw@madvax.UUCP (Carl Weidling) writes: > > Yesterday I bought a book called "Bit by Bit, An Illustrated History >of Computers", by Stan Augarten,Ticknor & Fields,NY,1984. > > ... I was reading about the design of the Mark I in Manchester,England >where it says: "He [F.C. Williams, the project's chief engineer] hit upon >the idea of employing ordinary cathode ray tubes [to solve the problem of >internal memory storage]..." > If anybody out there knows this stuff, I would be appreciative >if you could enlighten me. From: Nigel Topham <NPT@UK.AC.EDINBURGH.ECSVAX> It was in December 1946 that Williams and Kilburn started work on the CRT store, using commercially available CV1131 12-inch dia. CRTs [1]. The principle of a two-state electrostatic store can be visualised from the following simple experiment. Start with a focussed CRT beam and turn the beam current on (thus producing a charged dot) and off again repeatedly. Negative voltage pulses will be induced by capacitive coupling in a pick-up plate placed close to the outer surface of the CRT screen. Now move the beam whilst it is on so as to write a dash on the screen, then move the beam back whilst the current is off, and then switch the then switch on the current again. This time a positive pulse is induced. With dots and dashes representing logcal O and 1, readable as negative and positive pulses, a binary storage system is available. Although the electrostatic charge leaks away in about 0.2 seconds this can be refreshed in the same way that a modern MOS memory is refreshed. By the Autumn of 1947 the Manchester group had successfully stored 2048 bits of information for a period of several hours, and the way was clear to construct a prototype computer - in Williams words "...to subject the system to the most searching tests possible...". (Editited excerpt from S.H.. Lavington, A History of Manchester Computers, NCC Publications, Manchester, England, 1975, pp. 7-10.) [1] Williams, F.C., and Kilburn, T., "A Storage System for use with Binary Digital Computers", Proc IEE, vol. 96, pt. 2, no 30, 1949, pp. 183ff. -- -- > Brian Tompsett. Department of Computer Science, University of Edinburgh, > JCMB, The King's Buildings, Mayfield Road, EDINBURGH, EH9 3JZ, Scotland, U.K. > Telephone: +44 31 667 1081 x3332. > JANET: bct@uk.ac.ed.ecsvax ARPA: bct%ecsvax.ed.ac.uk@cs.ucl.ac.uk > USENET: bct@ecsvax.ed.ac.uk UUCP: ...!seismo!mcvax!ukc!ecsvax.ed.ac.uk!bct > BITNET: psuvax1!ecsvax.ed.ac.uk!bct or bct%ecsvax.ed.ac.uk@earn.rl.ac.uk
haynes@ucscc.UCSC.EDU.ucsc.edu (99700000) (07/16/87)
Actually the original Williams tube used, I believe, perfectly ordinary CR tubes, no flood gun inside and no special electrode inside. Harry Huskey here worked on these things, but he isn't handy for me to ask just now. Refreshing can be done with the same beam that does the writing because the beam is randomly addressable, not scanning. So you only need to store the one bit per tube that you are currently reading. In fact random access was the whole point, to avoid the time delays inherent in delay-line storage. IBM built its first marketed computers, 701 and 702, with this technology; then quickly switched to cores and brought out the 704 and 705 with the rest of the hardware mostly the same, but with somewhat different architecture. (701 was fixed-point only, 704 added floating point; 701 stored instructions in half words, 704 used full word instructions, added index registers). Williams tubes were also used in SWAC, which is the machine Harry Huskey built for the Bureau of Standards branch on the UCLA campus. They were also used in several other of the early computers built by universities. There's an old book "High Speed Computing Devices" by Electronic Research Associates (a forerunner of Remington Rand Univac) that may be in some libraries. Seems like the flood guns and extra electrodes came later when people realized that it would be possible to make oscilloscope CRTs with arbitrarily-long persistence that way. Hughes Aircraft used to make CRTs and oscilloscopes with these technologies. haynes@ucscc.ucsc.edu haynes@ucscc.bitnet ..ucbvax!ucscc!haynes
aa1@j.cc.purdue.edu (Saul Rosen) (07/16/87)
In article <602@madvax.UUCP> cw@madvax.UUCP (Carl Weidling) writes: > > Yesterday I bought a book called "Bit by Bit, An Illustrated History >of Computers", by Stan Augarten,Ticknor & Fields,NY,1984. > I was reading about the design of the Mark I in Manchester,England >where it says: "He [F.C. Williams, the project's chief engineer] hit upon >the idea of employing ordinary cathode ray tubes [to solve the problem of >internal memory storage]...Their operating principle was quite simple; >"guns" in the bases of the tubes shot positively and negatively charged >electrons at the faces of the tubes, thus storing bits in the form of charge >spots, which, by the way, were quite visible to the eye." > Well, I have a problem with "positively and negatively charged >electrons", but also, with how this can be memory. How long did the charges >last? were they refreshed? How was the memory read after being stored? Of course the statement you quote about Williams' tube memory sounds like nonsense. No positively charged electrons! The charged spot lasted about a fifth of a second. It had to be regenerated at least five times a second, and that provided a real memory. Williams' tube memory is described in all of the old books about computers. One of the more recent references is "A History of Computer Technology" by Michael R. Williams, Prentice-Hall, 1985.
christy@glacier.STANFORD.EDU (Peter Christy) (07/17/87)
This will really date me, but... I learned to program in the summer of 1962 while attending an NSF summer math program for high school students at UCLA. We visited the SWAC (SouthWest Automatic Computer) installation there (one of the '50's one-up machines). It used Williams tube memory. I was offered the chance to program it, but warned that at that time it was so frail that if you slammed the doors on the cabinets it lost bits (really). Instead I learned how to program the 7090 on some interesting number theory tests. I have quite a computer history library here, and if anyone really wants specific references on the technology please let me know by Email and I'll give you some book titles. Essentially Williams tubes, and the memory that Atanasof (sp?) used in his '30's "first" digital computers we're remarkably like today's MOS DRAMs. The nice binary cores were just a short term abberation! Peter Christy christy@glacier.edu.com
martyl@rocksvax.UUCP (Marty Leisner) (07/22/87)
In article <6700@santra.UUCP> jku@kolvi.UUCP (Juha Kuusama) writes: >In article <602@madvax.UUCP> cw@madvax.UUCP (Carl Weidling) writes: > ... >>... how this [ CRT tubes ] can be memory. How long did the charges >>last? were they refreshed? How was the memory read after being stored? >>Carl Weidling > >I once had a tube where characters were visible for several seconds after >the screen was blanked... > I got a leased PC/AT with an IBM green monochrome monitor which worked like this. Thought there was something wrong with it. Then I got a second one which worked the same way!! Those green IBM monochrome monitors can make one seasick!!
steve@nuchat.UUCP (Steve Nuchia) (07/30/87)
In article <3490004@wdl1.UUCP>, bobw@wdl1.UUCP (Robert Lee Wilson Jr.) writes: > The system there had the tubes covered so I can't speak to the > visibility of the bits, but based on how it worked they _ought_ to > have been quite visible except that the sense electrode over the > screen may have been opaque. As I recall from Hodges' biography of Turing (_The_Enigma_) the bits were made visible for debugging and/or entertaining the masses by attaching an ordinary CRT to the williams tube circuitry in parallel.