DOE Part III-LED产品测试流程标准翻译：Page-11

888888

a. 间隔性流明输出测量的收集说明  Interval Lumen Output Measurement Collection Instructions
      在2015年7月的SNOPR中，DOE建议所有的间隔流明输出测量满足IES TM-28-14的4.2、4.2.1和4.2.2节中规定的要求。对于大于或等于6000小时的测试时间，能源部建议遵循IES TM-28-14的4.2.1节。IES TM-28-14第4.2.1节规定，用于直接外推的流明维持数据必须在最初收集，此后至少每1000小时收集一次。对于大于或等于3,000小时和小于6,000小时的测试时间，DOE建议遵循IES TM-28-14的4.2.2节，但不收集LED封装和模块的流明维持数据。IES TM-28-14第4.2.2节规定，流明维护数据必须在1,000小时后开始收集，此后至少每500小时收集一次。
      In the July 2015 SNOPR, DOE proposed that all interval lumen output measurements meet the requirements specified in section 4.2, 4.2.1, and 4.2.2 of IES TM–28–14. For test durations greater than or equal to 6,000 hours, DOE proposed that section 4.2.1 of IES TM–28–14 be followed. Section 4.2.1 of IES TM–28–14 specifies that lumen maintenance data used for direct extrapolation must be collected initially and at least once every 1,000 hours thereafter. For test durations greater than or equal to 3,000 hours and less than 6,000 hours, DOE proposed section 4.2.2 of IES TM–28–14 be followed, except that lumen maintenance data of LED packages and modules would not be collected. Section 4.2.2 of IES TM– 28–14 specifies that lumen maintenance data must be collected initially after 1,000 hours, and at least once every 500 hours thereafter.

      不得使用在大于前段规定的间隔时间内收集的流明维持数据，因为这可能会影响到预测结果的准确性。此外，IES TM-28-14的第4.2节指出，必须在每个测量点的±48小时窗口内收集流明维持数据，例如，对于1000小时的间隔，在952小时和1048小时之间，在1952和2048小时之间，等等。这个±48小时的数据收集窗口也适用于小于1000小时的其他间隔。此外，第4.2节规定，用于预测计算的流明维持数据必须在时间上平均分散（在±48小时内），并且在最初的1000小时后，任何两个连续的数据收集间隔的长度都不得超过96小时。因此，如果每1000小时（±48小时）、每500小时（±48小时）等收集的数据可以用于预测计算，但不能每1000小时和偶尔500小时收集一次，因为这将给某些数据点带来过多的统计权重。同上。
      Lumen maintenance data collected at intervals greater than those specified in the previous paragraph must not be used as this may compromise the accuracy of the projection results. In addition, section 4.2 of IES TM–28–14 indicates that lumen maintenance data must be collected within a ± 48 hour window of each measurement point, e.g., for 1000-hour intervals, between 952 hours and 1048 hours, between 1952 and 2048 hours, etc. This ± 48 hour data collection window is also applicable to other intervals smaller than 1,000 hours. Furthermore, section 4.2 specifies that lumen maintenance data used for the projection calculation must be equally dispersed in time (to within ± 48 hours), and that no two consecutive data collection intervals after the initial 1,000 hours shall differ by more than 96 hours in length. Therefore, data may be used in the projection calculation if they are collected every 1,000 hours (± 48 hours), every 500 hours (± 48 hours), etc., but not every 1,000 hours and occasionally at 500 hours, as this will give excessive statistical weight to certain data points. Id.

      CA IOUs和EEAs同意DOE的建议，指出定期收集数据的时间间隔，如1000小时，可以识别早期的灯泡故障。(CA IOUs, No. 44 at p. 4; EEAs, No. 43 at p. 2) 然而，NEMA不同意DOE关于以1000小时为间隔收集流明维护的建议。NEMA指出，1000小时的测试间隔在实践中并不常见，因为工业界正在使用IES LM-80-08，而能源之星的测试收集点在3000和6000小时的间隔内。此外，NEMA评论说，任何改变都会使目前的能源之星认证数据无效，并导致许多产品的重新测试。(NEMA, No. 42 at p. 5) 飞利浦同意NEMA的意见，并补充说FTC通常不会在1000小时的间隔内收集流明维持数据，如果不修改测试程序，由于重新测试和重新认证的成本，制造商的负担将非常大。(Philips, No. 41 at p. 3)
      CA IOUs and EEAs agreed with DOE’s proposal, stating that regular data collection intervals, such as 1,000 hours, allow for the identification of early lamp failures. (CA IOUs, No. 44 at p. 4; EEAs, No. 43 at p. 2) However, NEMA disagreed with DOE’s proposal for lumen maintenance collection at 1,000 hour intervals. NEMA stated that 1,000 hour test intervals are not common in practice because industry isusing IES LM–80–08 and ENERGY STAR has test collection points at the 3,000 and 6,000 hour intervals. Further, NEMA commented that any change would invalidate current ENERGY STAR certification data and result in retesting of many products. (NEMA, No. 42 at p. 5) Philips agreed with NEMA’s comments, adding that FTC also does not typically collect lumen maintenance data at 1,000 hour intervals and that if the test procedure is not modified, manufacturer burden will be significant due to retesting and recertification costs. (Philips, No. 41 at p. 3)

      DOE不同意NEMA的观点，即工业界不熟悉以1000小时的间隔收集数据。NEMA推荐的行业标准IES LM-80-08和TM-21-11要求并鼓励收集1000小时或更少的流明维护间隔。因此，LED光源制造商应该已经在进行测试，至少使用1000小时的间隔。能源部还注意到，灯具制造商对他们的许多灯具进行 "能源之星 "计划的认证，正如NEMA所说，该计划要求对流明维护进行不止一次测量。虽然DOE要求对流明维护进行额外的测量，但DOE指出，一般来说，间隔测量可以提高寿命预测的整体质量。DOE意识到额外的测量可能会增加制造商的负担，并在IV.B节讨论的测试负担计算中考虑了灯具的测试。最后，能源之星计划参考了DOE现有的测试程序，并表示一旦DOE的LED灯测试程序最终确定，将采用该程序。由于这些原因，DOE维持其在所述时间间隔内收集流明输出测量的方法。
      DOE disagrees with NEMA’s point that industry is not familiar with gathering data at 1,000 hour intervals. Industry standards IES LM–80–08 and TM–21–11, recommended by NEMA, require and encourage lumen maintenance collection intervals of 1,000 hours or less. Thus, LED source manufacturers should already be conducting tests using 1,000 hour intervals at a minimum. DOE also notes that lamp manufacturers certify many of their lamps with the ENERGY STAR program, which, as NEMA states, requires more than one measurement of lumen maintenance. While DOE requires additional measurements of lumen maintenance, DOE notes that interval measurements, in general, improve the overall quality of the lifetime projection. DOE is aware that additional measurements may increase the burden on manufacturers and accounted for the testing of lamps in the test burden calculations discussed in section IV.B. Finally, the ENERGY STAR program references DOE’s test procedures where they exist and has stated its intention to adopt DOE’s test procedure for LED lamps once it is finalized.16 Thus, data can be shared between the two programs. For these reasons, DOE maintained its approach to collect lumen output measurements at the described intervals.

b. 预测计算 Projection Calculation
      IES TM-28-14的第5.0节对如何确定集成LED灯的失效时间提供了指导。对于短的测试时间（少于3000小时），IES TM-28-14没有提供推算方法，所以失效时间要用实际测试数据来确定。对于3000小时以上的测试时间，IES TM-28-14提供了两种不同的方法来推算失效时间，这取决于测试时间。第一种是直接外推法，根据整个LED灯的流明维持数据来推算失效时间。第二种是基于整个LED灯和LED光源流明维护数据的综合推断方法。DOE在本节中更详细地讨论了IES TM-28-14的这些规定。
      Section 5.0 of IES TM–28–14 provides guidance for how to determine time to failure for an integrated LED lamp. For short test durations (less than 3,000 hours), IES TM–28–14 does not provide a projection method so time to failure is determined using actual test data. For test durations of 3,000 hours or greater, IES TM–28–14 provides two different methods for projecting time to failure, depending on test duration. The first is a direct extrapolation method for projecting time to failure based on lumen maintenance data of a whole LED lamp. The second is a combined extrapolation method based on both whole LED lamp and LED source lumen maintenance data. DOE discusses these provisions of IES TM–28–14 in more detail in this section.

      如果IES LM-84-14测试完成后的总运行时间少于3000小时，IES TM-28-14没有提供流明维持率预测方法。IES TM-28-14指出，预测可能是不可靠的，因为对于不符合直接或综合推断方法的最低测试时间要求的数据集，预测估计值的范围会大大增加。基于有限的数据集可能会产生不可靠的预测，DOE在2015年7月的SNOPR中提议，对于测试时间少于3000小时的故障时间不做预测。相反，失败的时间将等于测试时间。80 FR at 39653.
      IES TM–28–14 does not provide a lumen maintenance projection method if IES LM–84–14 testing has been completed for a total elapsed operating time of less than 3,000 hours. IES TM– 28–14 indicates that the prediction may be unreliable since the spread of prediction estimates increases significantly for data sets that do not meet the minimum test duration requirements for the either the direct or combined extrapolation methods. On the basis of the limited dataset potentially yielding unreliable projections, DOE proposed in the July 2015 SNOPR no projection of time to failure for test durations less than 3,000 hours. Instead, time to failure would equal the test duration. 80 FR at 39653.

      对于至少6000小时的测试时间，IES TM-28-14程序建议使用直接外推法。直接外推法使用指数最小二乘法曲线拟合，将完整的集成LED灯的流明维持测量值外推到流明维持下降到初始流明输出的70%的时间点。80 FR at 39653-54. IES TM-28-14第5.1节中描述的直接外推法，根据整个LED灯的流明维护数据来预测故障时间，与DOE 2014年6月的SNOPR提案类似。79 FR 32035。然而，当DOE2014年6月的SNOPR根据ENERGY STAR的计划要求产品规范灯（灯泡）1.0版中的基本指数衰减函数预测故障时间时，17 IES TM-28-14根据每个单独的LED灯的数据预测故障时间。因此，在2015年7月的SNOPR中，DOE建议纳入IES TM-28-14第5.1节中提供的直接推断方法，因为这应该导致更准确的预测。80 FR at 39654.
      For test durations of at least 6,000 hours, the IES TM–28–14 procedures recommend use of a direct extrapolation method. The direct extrapolation method uses an exponential least squares curve-fit to extrapolate lumen maintenance measurements of the complete integrated LED lamp to the time point where lumen maintenance decreases to 70 percent of its initial lumen output. 80 FR at 39653–54. The direct extrapolation method described in section 5.1 of IES TM–28– 14 for projecting time to failure based on lumen maintenance data of a whole LED lamp is similar to DOE’s June 2014 SNOPR proposal. 79 FR 32035. However, where DOE’s June 2014 SNOPR projected time to failure based on the underlying exponential decay function in ENERGY STAR’s Program Requirements Product Specification for Lamps (Light Bulbs) Version 1.0,17 IES TM–28–14 projects time to failure based on the data obtained for each individual LED lamp. Thus, in the July 2015 SNOPR, DOE proposed to incorporate the direct extrapolation method provided in section 5.1 of IES TM–28– 14, as this should result in more accurate projections. 80 FR at 39654.

      尽管DOE建议参考IES TM-28-14第5.1节规定的直接外推法来预测LED灯流明维护数据的失效时间（如III.D.1至III.D.3节所述的测试），但2015年7月的SNOPR还提出了以下修改，以与DOE的报告要求保持一致。所有LED灯样品的测量流明维持率数据不能被平均化，也不能使用IES TM-28-14第5.1.2节中规定的平均化程序。相反，能源部建议对每个单独的LED灯完成预测计算，并使用预测的失效时间值来计算样品的寿命，建议的替代程序将在III.F.3节讨论。同上。
      Although DOE proposed referencing the direct extrapolation method specified in section 5.1 of IES TM–28– 14 for projecting time to failure of LED lamp lumen maintenance data (tested as described in sections III.D.1 through III.D.3), the July 2015 SNOPR also proposed the following modification for consistency with DOE’s reporting requirements: Measured lumen maintenance data of all the LED lamp samples must not be averaged, and the averaging procedures specified in section 5.1.2 of IES TM–28–14 must not be used. Instead, DOE proposed that the projection calculation be completed for each individual LED lamp and the projected time to failure values be used to calculate the lifetime of the sample using proposed alternative procedures, which are discussed in section III.F.3. Id.