Radio broadcast .. (1922-30)

.RADIO BROADCAST. in the making of an absolute standard has been passed. Production Starts The engineering department model is then copied, drawings are made, dies con- structed, and parts purchased. The planning and production departments disassemble and reassemble, figuring labor costs and the most economical way of doing things. Minor changes are made to gain savings of time and material, and to con- form with better production methods. Preliminary production work builds a dozen or more samples, each of which is examined and measured, noting how it fits inlo the scheme of things. If these samples prove satisfactory, the 1.20 real production is started. Cau- tiously, for even the engineers sometimes slip, the production increases, the variations and the effect of production expediencies being carefully noted. The pro- duction average is checked against the sample average and against the engineering department stand- ard. Differences are corrected or compromised, and the production grows. As a result of this mass copying, the absolute standard loses its individuality; it becomes an average value. There is no longer any single model which is fhf standard. The absolute stand- ard has been reduced from an in- tangible idea to a physical reality, and back to a synthesis of micro- volts per meter, band width in kilocycles, and percentage re- sponses at 60 cycles. The setting of the absolute standard is com- 0 plete. A second problem which should be coincident with the setting of the absolute standard but is usually subsequent thereto, is the setting of the passing and rejection limits of the individual units. There are both absolute and quantitative rejections, mechanically and electrically. In the case of mechanical details, an absolute rejec- tion might be due to a broken dial, or a defective socket. Mechanical quantitative rejections might be due to play in drive mechanism, scratches, etc. Electrically the absolute rejections might be due to no signal or oscillation. Quantitative rejec- tions electrically might be due to low sensitivity, poor selectivity, bad quality, too much a.c. hum, and low overload capacity. The mechanical defects, both absolute and quantitative, are usually caught by competent inspectors who use their own discretion. Electrical defects are caught by meters which are cajibrated to indicate the values of the relative per- formance characteristics. A failure of the meter to read a predetermined value on any test means a rejection. All of which sounds simple, and it is simple, if these values are correctly predetermined. It is in the determination of these limits that the engineer is prone to err. Rejection Limits Inspection to determine the electrical performance of radio receivers may be viewed from two standpoints. The inspec- tion and the limits may be set to maintain a high quality of product, or to detect and reject a subnormal set. The first view- point is the customary one; the second is the correct view from the economic standpoint. Nor are these two views synonymous as a first impression might indicate. To insure a high quality of product, all units should closely approach the absolute standard in all electrical characteristics. If the rejection limits correspond to the value placed on the absolute standard it is obvious that all sets shipped will be equal to or better than the absolute standard. It is equally obvi- ous that only 50 per cent, of the receivers built will be shipped, as the absolute standard has been chosen as the average of all the sets produced. Such a limit is ridiculous. If the limit is pushed further toward a lower standard, more sets will 10 20 30 40 L- HENRIES Fig. 2 50 be passed and fewer rejected. How far should this " pushing" be carried on? What is a safe basis for setting a rejection limit? There are several methods which might be used to determine what variation in product should be allowable and so determine the limits of rejection. Take as a first method that of customer discrimina- tion. By many experiments it is shown that the average ear cannot detect a difference in sound intensity of less than 15 per cent. (Fletcher: Bell System Technical Journal, Vol. IV, No. 3, p. 376). It might be foolish to maintain limits so close that a single individual could not hear the difference. The average customer expects a sh'ght difference between units. He knows that all receivers cannot be identical. Should Fig. 3 not the manufacturer capitalize this expected difference in units by broaden- ing his limits? Then again, many receivers are used in certain ways where limits do not greatly matter. Certainly a 10 per cent, reduction in sensitivity will not affect the intrinsic value of a receiver used for local reception only. Nor will a reduc- tion of possible power output from 20 60 watts to 15 watts be a cause for rejection if the receiver is to be used in a small home. If a manufacturer can show that 50 per cent, of his receivers are used for local purposes only it will undoubtedly, and justifiably, result in a broadening of his sensitivity limits. Accessory Variations Another method of determining the allowable variation in receiver characteris- tics is to consider the accessory variations. It seems illogical to in- sist on maintaining production limits much closer than the varia- tion resulting from factors over which the manufacturer has no control. A reasonable figure at first guess is to hold the set variation within 50 per cent, of that due to the accessories. In the modern radio set six, seven, or eight tubes are used. While the deviation in these tubes (mutual conductance) approximates 15 per cent., the effect on the set characteristics may be more or less, depending on the product of the stages and the criticalness of design. Measure- ments of the receiver characteris- tics with high and low tubes will show the magnitude of the varia- tion. Line voltage variation,! if not compensated, also causes a per- formance variation. Sometreceivers are quite critical to voltages, others seem to vary in direct pro- portion, while in one instance, over the normal voltage range of 105 to 125 volts, the sensitivity was almost independent of the voltage. Loud speaker variations may also allow the receiver proper to have greater permissible varia- tions.The place where the receiver is operated, as in a large room, or in a location with poor acoustic condi- tions, influences the sensitivity and quality demands. One of the greatest variations lies in the " pick-up" ability of the antenna used. When the apparent sensitivity of a receiver can be quadrupled by merely increasing the size of the antenna, why manufacture and hold it within 10 per cent, of the absolute standard? Combining all of these accessory variations gives a figure which is surprising as a possible variation for an individual unit. While hardly a direct method for deter- mining the rejection limits, a major in- fluence is the repair cost. If the repair cost is high, as is generally the case, it is eco- nomical to put more money into making a uniform product and thus cut the percent- age of rejections. This, based on nonuni- formity of the product, means relatively wider limits. Conversely a low repair cost or a high scrap value tends to narrow the rejection limits. When the cost of rejection and repair becomes appreciable with respect to the original cost estimate, poor engineering, poor production methods, or poor setting of the rejection limits are indicated. As before indicated, a general method of setting the rejection limits is to adjust them so as to reject arbitrarily X per cent, of the sets built. This is fundamentally not a standard at all as it is too easily changed to meet production demands. The X per cent, rejected tends to raise the standard while the 100—X per cent, meets the production demands. While this method is hardly good business, it is un- fortunately quite common. Another Method Any of these methods might be used if the object of the inspection tests is to • DECEMBER 1929 109