LIBRARY OF THE UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAICN 510.84 COI Digitized by the Internet Archive in 2013 http://archive.org/details/resultsoftestson47wier UNIVERSITY OF ILLINOIS GRADUATE COLLEGE DIGITAL COMPUTER IAEORATORY INTERNAL REFORT HO* !*? RESULTS OF TESTS ON 2£ BUREAU OF 21 IPS - RCA DEVELOPMENTAL WILLIAMS MEMORY TUBES * TYPE C73376B BY J. M„ WIER April 29, 1953 I INTRODUCTION The Bureau of Ships of the Department of the Navy entered into a contract with the Radio Corporation of Africa to produce an improved cathode ray tube for use as a "'illiams type storage tube A lot of 25 of the resulting tubes, RCA developmental type C 733763, is currently being tested by a "round robin" of compti*er groups who are interested in the development of the tubes * The type C73376B is a three inch cathode ray tube with electrical constants quite similar to those o€ the 3KP1 tubes used in the Illiac* However, it is understood that, as compared to 3KP1 tubes, the developmental tubes have a more well defined beam and precautions have been taken to reduce the number of flaws on the screen. A test has been made on the tubes for read-around and flaws. It is understood that the flaw test required that a flaw not reduce the dash amplitude by mora than 30 percent with a beam energy of 2500 volts* This report covers the results of tests performed at Illinois on these 25 tubes and the results of the same tests performed on 25 3KP1 tubes which had previously been selected from a group of 50 stock 3KP1 tubes as being the most suitable 3KPl's for use in a Williams storage system. All of the tests were run using a 102U address raster arranged in a 32 x 32 address array. Every other raster row was displaced one-half of an address spacing horizontally so a s to increase the average distance between addresses. The twitch travels in a vertical direction. II DESCRIPTION OF THE ILLIAC SYSTEM In the two dot Williams system as it was used in the I Iliac when the Illlac was first put into operation, information was sensed when the »2» beam was first turned on. The read-around failures were dots todSashes, For reasons mentioned in Internal Report 1j!>, tlis Illiac memory was altered in order to improve the read-around ratio. This alteration involves the use of a three pulse system instead of the normal two pulse one© The sequence involved in regenerating at an address is the following* The beam is turned on at the address in question for about 1 microsecond. It is then abruptly moved to an adjacent spot about one spot diameter removed, being left on during the process© The output is sensed at a time corresponding to that slightly after the move is made. The beam is left on this spot for a total of 2©£ micro seconds. Then using the ': formation gained from sensing the output, the beam is moved back to the original spot and turned on or not turned on there for 3 microseconds, depending upon the sensed information. The typical wave shapes for the two output signals are shown in Figure 1, along with the sensing time. The signal which initially goes positive will be called a dash. That which initially starts ©ut negative will be called a dot. The dash signal results from leaving the beam off when it is returned to the original spot. The dot signal results from turning the beam on at tiiat tlme Dash D«t Sens/Ajf __ Senses Ltvil Figure 1. -> This mode of operation alters the effect -which reach around ratio has on the infom&tion stored at a spot© The positive portion of the dot signal is caused by a combination of effects, part of which results from the inf oimation stored at tbe first spot and the remainder from the information stored at the second spot. The positive excision is partially caused by the collapse of the space charge about the first spot, which was already charged before the turning on of the beam* Further, since the second spot is uncharged when the first one is charged, an added positive effect is present due to b he charging of the second spoto As for the case of the dash signal, it is negative in this region due to a build up of an equilibrium space charge after the first spot is chargedo The net result of this is that stray secondaries tend to disturb the charges on the first spot in the case of dots and the charges on the second spot in the case of dashes and failures are noted both from dots to dashes and from dashes to dots. In this process, a net improvement is noted in the minimum read-around ratio by a factor varying frcm about three to four« It does mean, however, that read-around failures must be chocked in both directions and this has been done during the testing of the storage tubes It also is to be noted that flaws in general cause failures frcm dots to dashes in this system* III DESCRIPTION OF THE TESTS Each of the tubes was tested in a six digit memory normally used for tube selection and regeneration chassis repair. This memory is provided with circuit sequenced controls for performing various standard tests, among -u- them being the read-around ratio tests given below « After installation in the teat unit, the tubes were adjusted for focus and intensity setting while they were continuously being sequenced from dots to dashes and back again. The focus and intensity settings were made using only the output signals a3 a criterion. No attempt was made to standardize beam currents. Tha intensity was set so as to yeild a "standard" signal f 10m the output of a normal IH5-ac amplifier. After allowing about J> minutes to insure that all tubes under test were operating properly, the following tests were performed. Since originally the circuits in this test unit were designed to test the read-around ratio using a normal Williams system, the read-around ratio test from dashes to dots is somewhat mere elegant and certainly more through. This test proceeds as follows. The entire memory is first cleared to dashes. When started, a dot, which has a longer beam on time than a dash, is written in at address 0, This is folloi-jod by a regeneration at a point removed from that in question. Then a dot is again written in at 0. This process is repeated until the address in question has been bombarded n times, where n is variable in steps of 16 from to 1008. Then the entire memory is automatically checked for dash to dot failures. The bombardment address is then automatically increased by one and the process repeated until the whole memory has been tested. The value of n is advanced manually. Each tube may be checked individually or all six may be checked at once. Once the read-around ratio has reached the point where a failure occurs at some place on the raster, this tube may be ignored by throwing a switch so as not io delay tests on the other tubes at higher read-around ratios* »f^B* The read-around ratio is defined as the maxim-era value n at which no failures occur at any point on the tube. The test of dot to dash failles is not so neat or accurate due to the built-in nature of the test which was da signed to perform the above test* This test starts by filling the entire memory with dots* No automatic checks are allowed to operate « Then bombardments are made n times at each of the 32 points a3ong one row of addresses . The process is then stopped and checks are made visually by the operator. This process is succeeded by a clear of all address^ to dots and the process repeated on the next line. The read-around ratio is defined as before as the maximum number of bombardments at which no failures from dots to dashes occur© It is recognized that this test is not so stringent as the other, but its use has checked, at least qualitatively, with that found ?jhen using the Illiac to test normal 3KF1 tubes. The flaw test performed on the tubes is one which attempts to If s.te flaws by simulating their actual storage use. The test consists of performing a dot to dash failure tost *Aiile the raster is moved slow3y (and manually) through one horizontal and one vertical space so that one address is required to store at all points of a square of unit space size Ifcr moving the raster 3low!ly enough its accurate regeneration is not impaired. Further, by making the scan fine enough, every spot on the raster is used at seme time during the scan for storage. Since flaw failures were found to be from dots to dashes using this system, the entire raster is cleared to dots. Then a slow scan is made. If a flaw of non-storing variety is encountered, a failure is noted by one of the addresses becoming a dash. Since the position will not lose this dash, even after moving off the flaw, at the completion of the scan, there exists on the face of each of the storage tubes a record ,(je» of the ncn-storing flaws* In ordsr to test $v&% how marginal these flaws are or to locate moi'e marginal flaws, the intensity may be reduced so that the marginal flaws will be located. No elaborate attempts in the latter direction were made in the tests Sndicated in this reporto 17 RESUITS OF THE TESTS Table I gives ths results of tests on the 2$ develop® ntal Williams storage tubes. The tubes era of the RCA C73376B type and their serial numbers are given as CK3?19,B- 8 The B» portion only of the serial number is given as the identification of the tube in the table Table H presents ttie results for a similar set of tests of 25 standard 3KP1 type cathode tubes which had already been selected as being the best of approximately $0 tubes. The tubes are numbered in the sequence in which the tests were run. In general It has been found that the worst case seems to occur when the read-around ratio test is conducted by filling the raster with dots (which have longer beam on time than dashes) and surrounding the bombarding address with dots in the case of checking dot to dash failures and dashes in the opposite case. The bombardment is dbne with dots and the failures are then checked in the surrounding locations. This means that the tests in this report are not so severe as they should be for this sensing system, but it is rather difficult to alter the circuitry to do the more rigorous tests. It has been found that in practice, read-aiound ratio values ar e smaller by a factor of two when the more severe test is performed by the Hliac. In any case it is to be noted that there is a marked improvement to be had through the use of the new tubes. In addition to the read-around ratio -,'f. RAH 1 Serial No» Dashes to dots | Dots to dashes 1 Flaws B - 13 7 1008 ^1008 B - 1U >10O8 7IOO8 B - 19 ^1008 ^1008 B - 20 71008 512 B - 22 >10G8 ?10G8 B- 25>* ?lO0& >1008 B - 26 ^1008 >1008 B - 27 7IOO8 71G08 B - 30 7*1008 7100G B • 31 72003 >1008 B - 32 >1008 7-1008 B • 33 7X008 >1008 B - 3U >1008 >1008 B - 35 71008 >1008 B - 36 >2D08 >1008 B - 37 >1008 >1003 B - 38 >1008 $12 B- 39 >1008 >1003 B - U0 >10O8 ,71008 B - i*l >1008 71008 B - 1*2 >10u8 7 1008 B - U3 71008 71003 B - bh >1008 >1003 B- U5 >1008 71008 B - U6 71008 ■71008 TABLE I -8- Number RAR Dashes to Dots "j Dots to Dashes Flaws 1 >1008 x :: > 2 >1008 128 3 >1008 >1008 h 2U0 192 5 >1008 >1008 6 >1008 256 7 38U >1003 8 >1008 68 9 >10O8 >1008 10 >1008 >1008 11 3.60 >1008 12 896 896 13 >1008 >1008 1U 61*0 256 15 >1008 192 16 >1008 256 17 >1D08 320 18 >1008 61* 19 m >1008 20 06 96 21 256 >1008 22 >1008 128 23 >1008 128 21* 61*0 512 25 >1008 128 3 1 5 3 1 1 3 3 1 2 1 TABIE II -9- perfo nuance, the flawless storage surface is a notable gain This i3 especially true when it is remembered that probably more than half of the tube* which were rejected to obtain the 25 tube 3KP1 sample were rejected by reasons of flaws*, Inaddition to these qualities it was noted that the developmental tubes seem to be much less critical of the focus voltage setting than the normal 3KP1, and the amoiznt of adjustment necessary from tube to tube is smallo The grid cutoff point from tube to tube changes somewhat but no difficulties were experienced or are anticipated concerning this* The most undesirable quality of the tubes from the standpoint of the Illiac is their besor The tubes* although electrically and physically almost identical to a 3YP1, have a small shell, duodecal 10 pin socket instead of the medium shell magna! U pin base used on the 3KPlo This seems un* desirable from another standpoint too for it is convenient to mount the tube sockets solidly to the memory frame for easy replacement of tubes There are apparently no sockets available to do this conveniently with the basing at present employed on the developmental tube© Jo Mo Wier ,OUN0.