RAM Test Research Articles

Redundancy Techniques for Dynamic RAMs

This paper describes redundancy techniques designed for a 256 K dynamic RAM with polysilicon fuses which are used as electrically programmable ROMs. It shows that four spare rows and columns are suitable for obtaining the maximum yield improvement in the RAM. The first level polysilicon is used for the fuse and a two step etching process is employed to remove the PSG and final passivation layers. Fusing characteristics have been studied and a model based on a thermal equation is presented. A 256 K dynamic RAM test device with redundancy was designed and fabricated using 2 µm lithography. It was found that the yield could be improved by a factor of 5–10, if four spare rows and columns were incorporated.

HMOS III technology

A third-generation single-poly high-performance NMOS technology has been developed. The technology features scaled MOS devices with 250 /spl Aring/ gate and 0.25 pJ speed-power product. The technology integrates advanced processing capabilities in lithography and dry etching to achieve a depletion-load static memory cell size of 0.98 mils/SUP 2/. A 4K static RAM test vehicle has been fabricated with sub-15 ns access time. The technology has a major application in the shrink of existing microcomputers designed on previous HMOS technologies.

High-speed GaAs static random-access memory

An 8-bit fully decoded RAM test circuit has been designed and fabricated using enhancement-mode GaAs-MESFET's with the LPFL circuit approach. Correct operation of the circuit has been observed for a supply voltage varying from 3.5 to 7 v. An access time of 0.6 ns was measured for a total power consumption of 85 mW under nominal operating conditions. This circuit was used to develop and validate both a design strategy and computer-aided design (CAD) tools oriented towards cache or buffer memories of realistic complexity. It is shown that a performance-optimized 1-kbit RAM exhibiting an access time of 1.1 ns for a power dissipation of 850 mW would be feasible with the present fabrication technology.

High-Speed GaAs Static Random-Access Memory

An 8-bit fully decoded RAM test circuit has been designed and fabricated using enhancement-mode GaAs-MESFET's with the LPFL circuit approach. Correct operation of the circuit has been observed for a supply voltage varying from 3.5 to 7 V. An access time of 0.6 ns was measured for a total power consumption of 85 mW under nominal operating conditions. This circuit was used to develop and validate both a design strategy and computer-aided design (CAD) tools oriented towards cache or buffer memories of realistic complexity. It is shown that a performance-optimized 1-kbit RAM exhibiting an access time of 1.1 ns for a power dissipation of 850 mW would be feasible with the present fabrication technology.

Genetic studies on carcass traits in iceland twin ram lambs II. Analysis of principal components and construction of selection indices

The relationship between 20 carcass measures in twin ram lambs at the Experimental Sheep Farm, Hestur, Iceland, was studied by means of principal components. The genetic and phenotypic correlation matrices were modified by the recently developed “bending” procedure in order to reduce certain undesirable effects of sampling errors. The results of the principal component analysis indicated that a reduced number of representative variates can be extracted to form a selection criterion without incurring any great loss in efficiency. The breeding objective is assumed to be the production of compact carcasses having a large muscle to fat ratio. The selection criterion should include measures of fatness, amount of various valuable muscles, and carcass conformation. The ratio of cannonbone circumference to cannonbone length (MC/ML) was found to be a fair measure of the desired conformation. Seven traits were selected for the construction of substitution selection indices for progeny testing of rams. A carcass quality index is proposed, formed by weighting three of the substitution indices with economic values.

A 34 /spl mu/m/SUP 2/ DRAM cell fabricated with a 1 /spl mu/m single-level polycide FET technology

Dynamic RAM test arrays have been fabricated using a single-level polycide FET technology and a cell layout in which the top electrode of a given cell storage capacitor is provided by the adjacent word line. This layout achieves the same density as the conventional double-polysilicon cell, and comparable performance is obtained using a low-resistance polycide word line. Hi-C implants in the storage region provide increased capacitance, better isolation, and reduced transient noise. Design and operation considerations for the cell and arrays are described and measured results are compared to the design values. A cell area of 34 /spl mu/m/SUP 2/ is achieved using a scaled-down n-channel FET technology with a 22.5 nm gate oxide and 1 /spl mu/m minimum mask feature size.

HMOS-CMOS-a low-power high-performance technology

HMOS-CMOS, a new high-performance bulk CMOS technology, is described. This technology builds on HMOS II, and features high resistivity p-substrate, diffused n-well and scaled n- and p-channel devices of 2-/spl mu/m channel length and 400-/spl Aring/ gate oxide thickness. The aggressive scaling of n and p devices results in 350-ps minimum gate delay and 0.04-pJ power delay product. HMOS-CMOS is a single poly technology suitable for microprocessor and static RAM applications. A 4K static RAM test vehicle is described featuring fully CMOS six-transistor memory cell, a chip size of 19600 mil/SUP 2/, 75 /spl mu/W standby power, data retention down to a V/SUB cc/ voltage of 1.5 V and a minimum chip select and address access time of 25 ns.

Interactive software for automatic memory testing

A new interactive software package developed for automatic testing of dynamic RAMs of up to 64 kbits shows how a system can be made ‘user friendly’

Efficient functional testing of RAMS: M. Marinescu. Proc. CNET Conference on Reliability and Maintainability, Tregastel, France, p. 31 (8–12 September 1980). (In French)

Efficient functional testing of RAMS: M. Marinescu. Proc. CNET Conference on Reliability and Maintainability, Tregastel, France, p. 31 (8–12 September 1980). (In French)

Total memory test: A J Borer describes a fast test routine to detect repeatable errors in a 32 kbyte memory

The foundations of RAM testing theory are reviewed and several test procedures designed to cover logical faults are examined in the light of experience of problems most frequently encountered. Widely used test patterns, derived on the basis of simple considerations, are analysed within the framework of the above theory. The results of this analysis, summarised in a table, provide a useful guide for the choice of the test patterns to be used for characterisation or volume testing of RAM chips.

An integrated test concept for switched-capacitor dynamic MOS RAM's

An approach to dynamic MOS RAM testing has been developed. This paper deals in particular with the test problems of switched-capacitor (or single-transistor) MOS RAMs. Test procedures are developed from an understanding of the technology with which the memory circuits are built. The associated design weaknesses and failure modes are first reviewed, and four simple pattern-sensitivity programs are generated. Finally, an efficient test flow is recommended for rigorous functional verification as well as design and yield improvements.

Selecting Test Patterns for 4K RAMS

The selection of 4K RAM test patterns for use in a Go/NoGo or characterization type test program requires consideration of several areas related to test program quality. Among all others, two of the most important of these areas are pattern execution time and failure modes tested for. The most popular test patterns for 4K RAM'S are graphically illustrated with step-by-step instructions on test pattern generation provided.

The use of ultrasonics in phenotype testing of rams

An experiment in developing phenotype testing of rams was begun in 1975. One of the investigated characteristics was the area of the MLD measured ultrasonically. The experiment consisted of 130 rams on two feeding levels. In one group the rams were fed both hay and concentrates ad libitum, while in the other group concentrates were restricted. 124 of the rams were ultrasonically photographed, 61 from the ad libitum and 63 from the restricted feeding group. Weights (5 mths), daily gains and areas of MLD were approximately 39.3 kg (s.d. 4.4), 249 g (42), 11.6 cm2 (1.35) and 33.1 kg (4.6), 182g (32) and 9.4 cm2 (0.96) respectively. The ultrasonic area of the MLD correlated 0.52 and 0.32 with the area measured from carcasss. As a rule the ultrasonic area of the MLD correlated positively with other measures of live animals and slaughter results, but correlations with measures of relative meatiness were almost all nonsignificant. In completing the information concerning dressing weight and quantity of meat given by live weight, the ultrasonic area of the MLD proved to be the best of the measures on live animals. The rams in this experiment came from all over Finland but the next stage will consist mostly of progeny of the rams now selected (the two best and the two medium ones from both groups). With progeny testing the success of selection based on phenotype will be checked.

Progeny testing of Suffolk rams on three maternal breeds

1. Six Suffolk rams (two purchased from each of three breeders) were each progeny tested on 10 Scottish Halfbred (Border Leicester ♂ × North Country Cheviot ♀), 10 Welsh and 10 Clun ewes. The progeny of the Clun ewes were reared at a different location from the progeny of the Halfbred and Welsh ewes.2. Progeny were weighed at birth and every 3 or 4 weeks thereafter so that 56 and 112 day weights could be calculated. Growth rate was considered for the periods 0–8 weeks, 8–16 weeks and 0–16 weeks. The growth rates were corrected for birth type and sex, but not for birth weight. The growth rates of a ram's progeny for a ewe type were converted into a deviation from the mean of all progeny of that ewe type, and measured in standard deviation units. An overall deviation for all ewe types for each ram was also calculated; 0–16 weeks was considered the best period over which to compare ram progeny performance and the difference between the best and worst ram for this trait represented 0·81 of a standard deviation or a 10% increase in growth rate for the progeny of the best ram over the progeny of the worst ram. There was some evidence of ram × ewe-type interaction for this trait.3. A number of progeny of each ram were sent to a central abattoir and dissected into commercial joints. All pieces were weighed and a total carcass value obtained by multiplying each joint weight by an average retail price per pound and summing. The total carcass value was converted to terms of yield in pence per pound. Subjective assessments were made for colour and texture of meat and for tenderness and flavour of cooked meat. There was little variation between ram progeny groups for any of the carcass traits.

The place of ram testing in the sheep industry

The place of ram testing in the sheep industry

Factors Affecting the Estimation of Concentration of Sperm in Ram's Semen by the Photoelectrometric Method

Over 300 ejaculates from about 100 rams were exmained at the U. S. Sheep Experiment Station and Western Sheep Breeding Laboratory during annual Fall testing of rams. The relations between counts of sperm made with a Spencer Brightline hemacytometer and turbidity readings made with a Klett-Sommerson photoelectric colorimeter were analyzed by regression and covariance methods. Diluting fluids had significantly different effects on opacity of the diluted semen. Chlorazene was the diluter of choice, despite its change upon prolonged standing, because it was the only diluter capable of dissolving clumps which were prevalent in rams' semen. A dilution of 1:200 gave the optimum range of readings in the colorimeter scale, lower effect of debris, and lower error of estimate. Scores were established for estimating the amount of visible debris in the samples. Highly significant improvement in colorimetric estimation of sperm concentration was effected by scoring the amount of debris. Non-visible elements of debris, measured as residual turbidity after centrifugation of diluted samples, contributed appreciably to the colorimeter readings. This residual turbidity could not be used as an estimate of total debris. The colorimeter readings of semen with the debris removed gave a regression line with the intercept point not significantly different from zero. Two brands of diluting pipettes showed no differences in variability when rinsed into the diluted sample. They did show significant differences in volume and ability to drain without rinsing. The technique adopted for use (chlorazene at: 1200 with scoring of debris) gave a standard error of estimate of 38.1 (107 sperm per ml.). The error due to workers and instruments was negligible. The standard deviations for the two instruments were similar, 21.4 and 19.7, for sampling and dilution, but the hemacytometer was slightly less accurate, 26.7, because of its higher variance due to the Poisson distributional error of the smaller volume measured. The error of estimate attributable to scoring of samples for amount of debris was 22.7, while the error for debris and heterogeneity which could not be scored was 34.1.