Understanding N6700C Measurement Accuracy
The N6700C series of DC power analyzers support a wide variety of power modules, electronic loads, and source/measure units (SMUs). The datasheet lists the measurement accuracy for each supported module. However, it is important to understand that Keysight determined this data by performing 4,883 samples of 20.48 microseconds each and then averaging this data to arrive at a single number. The following calculation clarifies the reason for this choice.
4,883 x 20.48 microseconds = 0.1 seconds
The interval of 0.1 seconds corresponds to 5 power line cycles (PLCs) of integration for a 50 Hz power supply line, and 6 PLCs of integration for a 60 Hz power supply line. Integrating over multiple PLCs eliminates AC power noise, thus providing greater measurement accuracy.
The datasheet specifies the base measurement accuracy uncertainty in each measurement range for each module as a percentage of the measured value plus an offset value. Clearly, when making time sampled measurements faster than 0.1 seconds the measurement accuracy will decrease from the datasheet values. The N6700 Specifications Guide (publication number N6700-90001) provides detailed information on all N6700 modules, and it has tables that specify how much additional measurement offset you must add to the base value when sampling at faster speeds. Although minimum sampling times depend on the specific module, the N6781A SMU modules can sample as fast as every 5.12 microseconds (which is the fastest sample rate for all the N6700 modules). In datalogging mode, the fastest sampling rate for the N6781A SMU modules is four times this or 20.48 microseconds.
The following table on page 6 of the N6780 source/measure unit data sheet (publication number 5990-5829EN) lists the measurement accuracy for the N6781A/N6782A SMUs (first column), N6784A SMU (second column), and N6785A/N6786A SMUs (third column).
Suppose we are doing a datalog measurement at the maximum sampling rate of 20.48 microseconds. In this case the table on page 52 of the N6700 Specifications Guide tells us that in the 10 microamp measurement range for measurements with sampling times between 5.12 microseconds and 102.4 microseconds the measurement accuracy uncertainty increases by 15 to 20 nanoamps. To be conservative we should take the worst cast number and add this to the base value we just calculated.
As you can see, in the lower measurement ranges the measurement accuracy is dominated by the offset uncertainty rather than the measurement percentage uncertainty.
Using the lowest measurement range obviously always provides the best measurement accuracy, but what happens if some of the measurement points exceed the range limit (10 microamps in this example)? When the N6705C is operating in a fixed measurement range and the quantity under measurement exceeds the fixed range limit, then the instrument returns a value of 9.9E+37. This unrealistically large measurement value is present simply to notify the user that an out-of-range measurement occurred. If you are observing a measurement on the front panel (using either the scope or datalog modes), then the sample points that include the out-of-range data will appear as red lines. Again, this is there to make the user aware that these measurement points were out of range and no valid measurement data is available.
If you are taking a long datalog of low-level currents and your measurement data only occasionally exceeds the fixed measurement range that you are using, then it might be acceptable to just ignore or filter out those measurement points that are out of range. However, it is more often the case that you want to know with reasonable accuracy the values of both high and low levels in a datalog. Keysight’s patented seamless measurement ranging capability allows you to do this. Seamless measurement ranging checks multiple measurement ranges simultaneously and selects the best measurement range for each point in a measurement trace or datalog. The following images illustrate the advantages of seamless measurement ranging when datalogging current on a device with large dynamic changes in its current draw.
Determining Measurement Accuracy When Using Seamless Ranging
When using an N6700 module that supports seamless measurement ranging (such as the N6781A), determining the measurement accuracy when using seamless ranging is more complicated. In this case, you need to perform a post-measurement analysis of the data using the cumulative current distribution function (CCDF). You then need to determine the percentage of time that the measurement samples spent in each of the three measurement ranges that seamless measurement ranging uses (3 A, 100 mA, and 1 mA). Next, based on the sample period used, for each of these three ranges you need to determine the worst-case measurement accuracy for that range using the tables in the N6700 Specifications Guide. With these two sets of data you can determine the total measurement accuracy by adding together the squared values of the weighted uncertainty for each range and taking the square root of that calculation.
Summary
In conclusion, it is important to understand that the accuracy data in the various N6700 data sheets corresponds to very specific measurement time sampling conditions. It is likely that most measurements you will perform using the N6700 series will be at sampling rates faster than those used in the datasheets. To determine the accuracy of those measurements you must use the tables in the N6700 Specifications Guide, which tell you how much additional measurement uncertainty you must add to the base values shown in the N6700 datasheets. When using seamless measurement ranging, you must weight the amount of time spent in each of the three supported measurement ranges by the maximum measurement uncertainty experienced in each range.