LIBRARY OF THE UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN 510.84 Coi Digitized by the Internet Archive in 2013 http://archive.org/details/selectiontestson34wier ELECTRONIC DIGITAL COMPUTER INTERNAL REPORT NO, 3U SELECTION TESTS ON 3KP1 TYPE CATHODE RA7 TDBSS FOB A WILLIAMS MEMORY JOSEPH M. WIER May 22, 1952 University of Illinois I IHTRQDUCTIQS To provide a suitable number of properly operating cathode ray tubes for the Williams type memory to be used In the University of Illinois computer, a program of tube testing was set up. From previous experience in selecting tubes It was found that the percentage of tubes acceptable to tubes received of standard commercial 3KP1 type was from 25 to 35 percent. This report covers the testing of 139 cathode ray tubes, a sufficient number to acquire the 32 additional tubes needed to tube the machine. II 3DESCRIPTI0H Off TEST It was intended that the test should be conducted so that the tubes were operating under the same conditions as they would be expected to in use. Therefore, the regeneration amplifier and associated circuitry are identical to that which will be found in the completed memory. The tube voltages are the same. The control circuitry necessary to sequence the tests is not identical, bub was so designed as to provide identical signals to the regeneration chassis and cathode ray tube as those In use in the computer. The first test determines the ability of the tube to store Information, at least on a short term basis, and tests the facility with which the machine may write into and read out of the memory. This is done by first adjusting for proper focus, astigmatism and Intensity with the tube in the operating circuit. The output of the amplifier is used a§ the basis for making these adjustments. Then the external control is •et to perform the reading and writing tests by holding a static pattern In half of the memory and continuously reading into and out of the other -2- half, reversing the contents of each spot each time. This type of operation is usually continued for a couple of minutes per tube to determine that it is generally operating quite -well. The main part of the test is done to ascertain a figure for the read-around-ratio and to note the presence of flaws. The read- around-ratio is here defined as the number of times a spot may be written into with a dash before the surrounding dots are so affected by refill that they are sensed as dashes, The control circuitry does this test by writing a dash into an address, which is selected by a set of 10 switches, for n times where n may be set from 1 through 1023 o At the end of this bombardment, the entire raster is regenerated once and the bombardment regeneration sequence is repeated. The figure obtained for the read-around ratio is the maximum value of n for which no failures occur in the surrounding dots for a period of at least five seconds. In order that the 139 tubes might all be tested in a reasonable time, only four points were tested for their read-around- ratio. These points have the following coordinates where the first figure represents the horizontal coordinate on the 32 x 32 raster and the second represents the vertical coordinate. Positive directions are down and to the right from the upper left corner. 0, 8; 2k, 8j 15, 15; 31, 31 The raster configuration is one in which odd columns are displaced downward one-half of a vertical address distance, alternate columns are twitched horizontally in opposite directions and the spacing between -3- alternate columns Is adjusted 80 that failures occur vlth reasonably similar frequency to all the spots surrounding any one spot. The test for the presence of flaws is done by filling the raster with dashes and allowing them to be stored statically while the raster is slowly moved through a one address horizontal and vertical television type scan so that all of the area of the used tube face is required to store a dash at some time. The presence of non-storing flaws will be noted after the scan by the presence of dots on the raster which have been caused by passing over flaws Incapable of storing a dash. This does not indicate every spot which has a secondary emission ratio different from the average value on the screen* but it does in- dicate those spots which would cause trouble in use in the memory as it is built. The criterion for the acceptance of a tvbe was that it have three or less non-storing flaws and have a read-around-ratio at the four selected spots of 30 or over. In a few cases , these conditions have been slightly violated in the case of read-around-ratio when the flaw count was very low. The exceptions probably will not be used in the completed memory but may be useful in an emergency. Ill RESULTS OF THE TES T Of the 139 tubes tested 73 were finally selected as being good enough to consider using in the completed memory. The remaining 66 were discarded for the following reasons; 29 for poor focus, 28 for having greater than three non-storing flaws, 7 for poor read-around-ratio, 1 for an intermittent open, and 1 for being excessively microphonic. The figures for the good tubes selected are Indicated in the attached table. It will be noted that 3^ tubes have no non- storing flaws, 25 have only 1, 8 have 2 flaws, 6 have 3 flaws. The four IBM tubes have been included as a comparison,, It will be noted that none of these have any non-storing flaws and all have high read-around-ratios. It is also worth noting that the IBM tubes with PI phosphor give lower output signals than the 3KPl*s tested by at least 20$. The IBM 2263 has a P5 phosphor and gives signals as large or larger than the standard 3BP1, The smaller signal in the PI type IBM tubes is believed to result from the smaller beam size and the resulting lesser displacement of charge from the reduced spot. Apparently the P5 phosphor has a higher secondary emission ratio than the PI type. A word of caution should be injected about the apparently large read-around-ratios noted even among some of the 3KP1 tubes. Since these tests are conducted by bombarding a spot with a written- in dash and the dot pulse is not present during the boinbardment, these figures might well be lowered somewhat If the test were run on a continuously read- in dash basis where the dot pulse is present since the beam is on longer per cycle under these conditions. In any case, the 73 tubes accepted here should be sufficient to place the University of Illinois machine into operation* GOOD TDBES DESIGHATICST RAK BESKH5ATIOH 0,8 2*,8 15,15 31,31 FLAWS A 6* 6* 96 192 3 B 72 120 256 80 3 C 350 72 80 112 3 D 6* 88 80 26* 1 E 192 288 38* 112 3 P *8 32 56 8 a 320 376 5** 320 1 H 6* 56 80 56 3 z 88 2*0 112 208 2 J 152 120 2*8 880 K 38* 520 280 *8o L 88 72 96 1008 M 2*8 50* 2*6 2U8 2 V 1** 176 128 192 1 120 6*0 888 1023 P 2* 56 96 22* 3 Q 32 96 56 2h8 2 R 352 96 26 768 S 96 128 2*0 *0 T 96 22* U80 128 1 U 112 128 160 192 2 ▼ 216 200 208 80 Page 2 60QD TUBcib CURTIKUED Design. W 96 RAH 224 184 88 Flaws X 120 260 128 512 T 168 128 248 208 2 Z 332 264 376 248 2 AA 320 288 288 12 AB *es 312 496 1023 AC 16 24 32 216 AD 64 72 154 136 AE 32 32 80 H8 AF 192 160 240 520 AO 120 224 120 112 2 AH 56 80 72 40 1 AI 192 112 160 224 1 AJ 376 384 384 892 AK 80 56 104 64 AL 352 80 144 192 1 AH 120 80 104 512 1 AH 64 144 192 160 1 AO 256 240 288 672 1 AP 88 96 128 552 AQ 2*0 192 224 112 1 AR 64 176 240 128 AS 224 192 224 640 AT 56 24C 176 416 AO 120 312 128 512 Page 3 GOOD TUBES CONTIHUED Design. BAR Flaws AV 192 112 272 80 1 AW 56 64 56 144 1 AX 48 320 423 512 1 AY 112 160 168 284 1 AZ 224 352 240 512 BA 4oo 256 208 384 1 BB 16 136 256 128 1 BC 144 2kQ 18* 256 1 BD 192 696 368 464 2 BE 112 128 88 264 1 BP 56 248 280 144 m 192 2#* 248 72 1 BH 232 264 248 248 BI 376 256 408 208 1 BJ 80 96 6k 32 BK 120 184 304 528 1 BL 320 360 488 656 BM 224 192 352 480 BN 16 6k 32 24 1 BO Ikk 288 144 20 BP 22k 112 224 512 » 224 352 160 288 1 BR 92 96 96 128 B3 48 88 160 56 Page k &COD TUBES CONTIHUED Design. BT 208 BO 6k BAR Flav8 kf2 Uk6 kk6 1*0 36 96 1 IBM X263 M6 m 1*80 1023 IBM X300 kl6 kl£ 22k 1*80 IBM N28 ki6 38^ 1023 768 IBM B30 256 272 736 • 381* *OUN 0>