LIBRARY OF THE UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAICN 510.84 f>o. 22-59 COI Digitized by the Internet Archive in 2013 http://archive.org/details/screenimperfecti22will > EIJSCTRONIC DIGITAL COMPUTER INTERNAL REPORT NO. 22 SCREEN IMPERFECTIONS IN TYPE 3-v?l CATHODE RAY TtfiBES By David C. VlUiMU September 22, 1950 "CRSITY RESEARCH BCAHL UNIVERSITY OF ILLINC : ^^^^^^H X IV\CO>M hh For most conventional uses the presence of small Imperfections in the phosphor screen of a 3KP1 vill cause no noticeable impairment of the tube's serviceability. Such imperfections are visible to the naked eye only upon close inspection of the back of the screen, and not alvaya then. T5ue to the retention of an image upon the retina of the eye, halation effects of the screen, and possibly because the contamination is present upon the inside sur- face of the phosphor and does not completely prevent the electron beam from reaching the phosphorescent material, no effect rer the contamina- tion can be noted open the visible trace across the contaminated area. If the tube is operated in a memory system, however, the presence of an extremely small quantity 3f foreign material 'fficulty if the raster is positioned so that a spot happens to fall on the contaminated area. This is because t' try emisr' rtie of the contaminating material differs frevm that of the surrounding ~. In an operating memory syeten the result of storirg on a ec is a greatly i I DASH output signal, and consequent lack of reliability of the met >rate at that particular address . In some cares imperfections have been found of such a nature that the tube could act be forced to store at that address under any condi- tions. In lieu of any better vay cf describing such imperfections it is convenient to speak of a contaminat* .^ spot in terms of the error llgflAl which a c-ven gpot canses . The error signal is the amovat Ln vults by whJ lard dash signal - 35 volts at the output of an amplifier having a gain of 8U,0C :ld be reduced. Throughout the remainder of this report this terminology is u3ed. tataii «rf «v» ■ , - ■. TOk II It is possible to examine the screen of a cathode ray tube by utilizing the arrangement shown in Figure 1. The tube under test is scanned continuously. When the scanning beam passes over an area having a secondary emission charac- teristic that is different from that of the surrounding phosphor there is a change in the potential level and hence a small flow of current to the pick up screen. The resulting signal is amplified and used to intensity modulate the grid of the 5UP1 viewing tube. The gain of the amplifier is 1^0,000. The sweep for both scopes is obtained from two Hewlett-Packard oscillators set at slightly differ- ent frequencies. As both tubes are swept in parallel there is a one-to-one cor- respondence between points on the viewing tube and the tube under test. Areas causing error signals may thus be marked on the screen of the tube under test. Good results have been obtained with this equipment. Photographs of the viewing tube screen may readily be made and provide a record of the number and approximate size of the imperfections of the tube screen under test. This method is not a quantitative one, but the relative size and intensity of the spots pro- vide a good approximate guide to the size of the error signals that the contami- nated areas will cause. Figure 2 is a photograph made with the equipment Just described. The l«rge, bright spot caused an error signal of 2k volts as measured in an operating memory. Ill The method of examining tube screens which was just described is not en- tirely adequate as it does not give quantitative results which are accurate enough to serve as a guide in ordering tubes. If it is desired to obtain a • ■ ; :■'.' >qsvi mm n ■i-.'-i' certain number of tubes none of which will cause error signals greater than a certain maximum it becomes necessary to knov vith reasonable accuracy what per- centage of tubes, as supplied by the manufacturer will have screen imperfections which are capable of causing larger error signals than the desired maximum. For the purpose of determining the actual size of theerror signals a triggered scope was connected in the circuit as shown in figure 3- After locat- ing an imperfection on the screen of the tube being tested the vertical sweep was shut off and the beam allowed to sweep back and forth over the contaminated area in a straight line. The size of the signal caused by the imperfection could then be determined from the deflection of the triggered scope. » The error signal arising from the largest imperfection on the screen of ■«• each of the shipment of 63 tubes was measured in this manner and the results plotted in figure k. The curve of figure k suggests that if a larger number of tubes had been available the curve might have been very close to a straight line. In any event the small number of tubes available cannot be considered sufficient to serve as a random sample and since they were obtained on one order it is probable that they were manufactured successively, or nearly so. miring the course of the tests small oscillations developed in the video system on two different occasions. The first occurrence was caused by decreased emission of the 6AQ5 cathode- follower output tube the screen was originally tied to the plate and the tube tended to 'take-off as a dynatron oscillator. A 1000 ohm screen resistance and a 0.01 nfd. by- pass were inserted in the circuit, and 1. The original order was 65 tubes. One tube was rejected for mechanical reasons; another tube was placed in service before being measured. ■ hettmz B9flR>3ftd it ohjwj xawr a : . ■ I : -i o.'I t b ■:■"■■'-; UW t *»• soil £>achrc&£«& no norit JOi'iT. 5tf& ./.".V.fl S*X9fc .'•...■uj*at.s sclstf ... no further trouble developed from this source. The second occu- ranee of oscil- lation was an extremely small oscillation having a frequency of approximately 5 mc. This oscillation was probably present during the entire run of the tests up to the time that it was observed. It undoubtedly would have passed unnoticed due to the extremely small size of the oscillation had it not been for the fact that one or two tubes were examined with the signal being applied to the triggered scope through both input amplifiers. These amplifiers have a gain of approximately ten per stage. The cause of this oscillation was coupling between the input and output of the video amplifiers. The feedback path included, in order, the grid structure of the viewing tube, the vertical deflection plates of the viewing tube, the direct connection to the vertical deflection plates of the tube being tested, the pick-up screen of the tube under test, and so to the input of the video ampli- fier. The oscillation was present vhea both cathode ray tubes were turned off. Insertion of a low pass filter having a cut-off frequency of 3.8 mc., in the grid circuit of the 6ak6 amplifier cured the trouble. The results of this series of tests should not be considered as more than a rough guide. It was attempted to keep the accuracy of the measurements as high as possible, but the results cannot be considered as being accurate to within less than 10$. The chief source of inaccuracy is undoubtedly the fact that the tubes tested did not represent an adequate random sample. The error signal volt- ages were measured on a scope having a deflection sensitivity of 22. 5 volts/cm., and the difficulty of estimating the heights of small peaks contributed to the error. The tests took an appreciable length of time to complete and the equip- ment cannot be considered as holding a perfectly constant gain over such a long period - about a week and a half - particularly when the overall gain of the ':■..'■''■ J v; • :.<-•' rfO f •' ■ hnm ?•: it N -A3C M| Mia •>:".''.. ■ I Ml ^ i Ml iMNtfl . ' ■ I li Miffed t )J .'i' hi-»HitI lAua±« tot** tat ft* ■i I bAmm . ■ ; ."■■■ . . .'■ '■ lliKf Jt Jtaft Xamr « Jvods video system was as high as 1^0,000. The gain of the video system was measured at 179,000 at the beginning of the tests, but vas reduced to 1^0,000 when the oscillat ons were removed from the amplifiers. The gain was meaeured each time work was done on the amplifiers, and all error signals were scaled down to the value that would result from a vidoo system amplification of 3^,000. Of the 63 tubes tested only two were without measurable flaws. The largest error signal observed was measured at 21.1 volts in the test equipment. This was the signal originating from the spot shown in figure 2. This tube vas later placed in the memory constructed by George F. Bland and the difference in the dash output from the contaminated area and the normal dash output was measured at 2k volts. The discrepancy here is greater than 10$, but is believed to be due to the difference in beam currents of the two pieces of equipment. The second anode current of the tube in the test equipment was adjusted to 2 \x&., as it is believed that tills figure approximates that of the actual memory of the computer. There has been considerable conjecture as to the actual cause of the imper- fections and the contaminating material. One theory has been that pin holes in the phosphor are the causes of the error signals, or at least one of the causes. At the present time, however, this seems to be definitely contra- indicated. The screens of several tubes were examined by illuminating the screen with a strong light and then looking at the back of the screen through the small band of clear glass between the end of the internal aquadag coating Ml M& :" r < -•:;'.)'•. V&« "to »: 'AOltel and the phosphor. Fin holes, and sons considerably larger holes, in the phos- phor are very easily located in this manner. Several of the more prominent holes were Barked and then the marked area examined in the test equipment. Ho error signals vera observed which could be attributed to holes in the phosphor. The largest such hole found vas approximately l/l6 " x l/32 • There was no measurable error signal originating in the immediate area of this hole . Apparent- ly the difference between the secondary emission characteristics of glass and zinc orthosilicate is too small to give rise tc an appreci* ble error signal. Another theory as to the cause of imperfections seems to be much more likely, although so far it has not been definitely proved. This second theory is that at the time the gun is inserted into the tube the positioning clips scrape off small quantities of the aquadag coating which fall onto the screen, or are shaken onto the screen during subsequent handling. Examination of several tubes has revealed that there is often a small black speck iu the area which is contaminated. At present there is no evidence that every area from which an error signal originates contains such a speck, visible or microscopic, out several have been noted. The tube whose trace is shown in Figure 2 was diseected and the contaminated area examined under a microscope. Upon close examination it was apparent that the origin of the error signal was not one, but several small particles which had fallen onto the screen at some time during manufacture. Figure 5 is a photomicrograph of the area. Conclusion: The chief objective of this work has been to establish a guide which will assist in ordering tubes for use in the computer. With the — "/fx»*** a.''-"i I ^ ■ t vi& y,: ti fl . ■ - .. '• lap h :> dad* - a «1 «xsd£ linn .] OBfJ «M»t too ssi'f • f v ar. :-•■ 9hfL -. Jhcafireo k .saf; X. trfv .- limitation* Mentioned the work haa been successful in this respect A method of evaluating tubes quantitatively for use in a Williams tube m mory has been set up, and a good idea of the quality of tubea as shipped by the manufacturer without special testing for use in Memories, has been gained. Secondarily a good deal has been learned about thenrature of phosphor contaminations, and more vork will be performed along these lines in the f\iture; although it ia doubtful if information thus gained will be of any immediate or direct benefit to the project. It is believed that a 3KP1 may be operated in a memory "without the internal aquadag coating, and It i8 known that the secondary emission characteristic of glass is adequate for storage purposes. At the present time a tube is under construction containing no aquadag or phosphor. Special precautions are being taken to prevent contamination of the screen. An attempt will be made to analyze the contaminant found in the one tube that was dissected, but due to the very small quantity available it Is doubtful if a positive determination can be made. In conclusion I wish to thank the members of the staff for generous help at one time or another, particularly Edward L. Hughes, who has spent considerable time assisting me. The test procedure used has been generally suggested by members of the Computer group at the Institute for Advanced Study. . 'jurw Ttc ■ i m i ■■:. . ■■.:_■*-■»?:«■•■ j - rlti bOn -; il wuf bm Error Signal Tube No. Afflpl. Gain 3^,000 % of Tubes 7* 36 3.18 £.76 I 3 fc.05 6.35 75 M 76 5.k 67 5.* 11.1 57 6. 12.7 58 6.75 56 6.75 61 6.75 63 6.75 73 6.75 81 6.75 82 6.75 51 6.75 25.2 52 8.1 62 8.1 66 8.X 30.2 38 8.U5 *9 8A5 33.3 80 $M 78 9.^5 79 9**5 68 9.^5 39-7 £l.3 UQ 9.5 65 10.12 k2.B kl 10.5 kkM kk 10.6 k6 26 10.8 83 10.8 77 10.8 50.65 25 10.9 32 .9 52.95 39 11.6 *3 11.6 57.1 61; 12.15 58.7 *5 12,7 *7 12.7 61.9 71 13-5 72 13.5 65 29 1^.6 65 -6 69 1A.85 63 .2 31 15.3 69.8 31 15.8 flJk 30 16.3 73 2^ 16 A Jk. 6 5 16.9 16.9 77.8 Tube No. Error Sigaal $ of Tubee 28 17-^ 79.5 23 17.8 27 17.8 3* 17.8 8U.1 ko 17.9 h6 17.9 87.3 33 18. 88.9 60 18.25 90.5 22 IB. 5 92.0 59 18.85 93-6 70 18.9 95.2 35 19.1 50 19.1 96A 53 21.1 100 ' M 8 V3 ^* & ^ *j v^ ta k ^ ^ •V. ^ 1 1 1 ^ V9 ^ 1 > i 1 ^ pk * 5 r* 3 ^ k //\. 1 mi ~ 8 § I K o k k «0 *1 I k 8 Figure 2 Figure 3 N SQ 5 <0 •000 '+9 jo rv/rp as/jy/7— WV 53 <4> Hi l \A/V- # ^<±H 'Q vw- 4> — a^ -*AAA- VW A^ 3l ww- WA/v >AWv ^ U K k $ I ^1 f=f 3 £ ^_ n u n j t *> =1 WSA^ ^ HI- k Ifif k I 1 I ^ t-< ft 3 ^ ii ii DO 00 I k ! k I <4 fUpam 9 V>UN£>^