Since it is quite apparent that some users of this forum, particularly metrology, are not familiar with many precision resistor specifications and definitions, I will present industry standard definitions and where deemed appropriate, additional notes will be added. These apply to both precision wire wound and film/foil resistors in most cases, there may be slightly different variations in each case.
Drift: The change in resistance over time, there are sub-categories of short term and long term. There also some variations with resistor types. For short term drift, it can be anything from 24 hours to a few months and the specified conditions can also be different. from sitting on a shelf (unpowered) to continuously powered and under specified ambient conditions, from room temperature @ 23°C to 25°C, to military temperatures of -55°C to +125°C as spelled out in the individual specifications for that type of resistor. Long term drift is usually specified as at least one year or more and also under specified conditions as enumerated above. Some manufacturers will add the short term drift to the long term drift to get the 1 year specification.
Note: Drift is considered unpredictable by manufacturers for many reasons; the drift of an individual resistor cannot be predicted as it is caused by random variables both in manufacturing and subsequent environments. While drift can be predicted somewhat as a group, primarily based on resistor history, both short term and long term drift can only be specified as a general trend as a group. When a long term drift of say ±400 PPM/year is given, that merely says that the vast majority of resistors will stay within that drift area at the end of one year, it does not guarantee that a resistor cannot exceed that drift. The same limits apply to short term drift as specified under a given set of conditions, in some cases short term drift may actually be guaranteed under the specified conditions and will be specifically stated as such, but as a general practice, it is not a guaranteed limit.
Another strong reason why drift is not specifically guaranteed is because the manufacturer has no control over the resistor(s) once they leave the factory, they can be subjected to any kind of stresses while in transit top the customer, for instance very cold temperatures in a cargo hold of a plane or very hot temperatures in a cargo truck transport or delivery van. These stresses can have very real effects on precision resistors, even on resistor standards of high stability such as an SR-104.
Tolerance: This is specified as ±x.xxx% of nominal at a specified temperature. Manufacturers use guard bands when measuring in order to insure that the finished resistor is within tolerance, with precisions it would be a bit out of the ordinary to receive a precision resistor that was right on the limits of the tolerance, the limits are reduced enough to compensate for measurement accuracy and uncertainties in the measurements.
Accuracy: The conformity of a measurement to an accepted standard value, accuracy includes traceability to an appropriate national/international standards organization such as the NIST.
Calibration accuracy: The sum of the uncertainties in the calibration procedure which includes the uncertainties in the references, test instruments, transfers, bridges, etc. Calibration accuracy must be better than or equal to the stated accuracy or initial accuracy of the measurement. The minimum preferred accepted accuracy/uncertainty ratio is 5:1; the preferred ratio is 10:1 when possible. There are circumstances where the 5:1 ratio maybe difficult to achieve, that is usually reflected in the calibration accuracy in use.
Initial accuracy: This is the guaranteed accuracy at the time of shipment; this is the accuracy you will see printed on the resistor, such as ±0.01%, this means that the resistor(s) are within that tolerance/accuracy when they leave the factory. See the note above on drift for reasons why it is not normally guaranteed beyond the shipping doors. This is usually not a factor unless unusually tight tolerance has been requested by the customer, as the tolerance becomes tighter, the likely hood that the resistor may wind up being out of tolerance sooner than expected is unfortunately surprising to some. This is a well know phenomenon which has been around since the early days of resistors. The initial tolerance is not guaranteed forever.
Note: The most common methods used to minimize drift is PMO (post manufacturing operations) and hermetic sealing, it must be noted that not all drift can be eliminated even with lots of time, resistors will drift with age essentially forever at a low level. What is reduced with PMO is the stresses incurred during manufacturing, these stresses will relax with time on their own or some of the relaxation can be hurried up by PMO. PMO does not and cannot relieve all of the stress, it can relieve a lot of it depending on how much PMO is done, this is usually done only for unusual circumstances and of course it does incur extra costs. The resistor will naturally relax the most in the first year (I.e. drift) and normally has decreasing drift with more time. It all depends on how the resistor is used, sitting on a shelf, powered at some wattage, subjected to temperature variations; everything has an effect on the drift rate. Hermetic resistors are normally put through some form of PMO before being sealed inside the can; otherwise the resistor is still going to drift like it would without the hermetic seal. Yes, the resistor’s value will drift with PMO that is the way it works.
Test Conditions: These are the conditions describing how the resistor, measurement device(s) and environment are to be used under. These will include relative humidity, power, temperature, frequency, ect. In most instances with precision resistors, the measurement is done with minimal power and a specific temperature and power, depends on the rating of the resistor, can be as low as 10mW….keeping in mind that more power will cause more self heating and a larger change in resistance. Resistors are normally specified with minimal power for measurement. Other effects such as different temperature and higher power must be considered by the customer and specified if required for the manufacturer in order to meet specifications.
Hysteresis: Hysteresis is defined as a change in resistance in response to an external stimulus, such as temperature cycling or power cycling. The response depends on how relaxed the resistor is, the residual stresses in a resistor respond to the external stimulus by changing resistance and depending on how ‘relaxed’ the resistor is determines how close it comes to its original resistance. Hysteresis depends on the type of resistor, how it was manufactured and treated and its intended use. Like drift, hysteresis tends to decrease with use and age but may never completely go away.
Humidity: This is highly variable depending on the type of resistor and how it was designed. Some resistor types are quite sensitive to humidity while others are not, of course hermetic mostly eliminates this from consideration at some expense. Humidity is often cited as a problem but is often incorrectly measured in that each parameter must be isolated in order to achieve an accurate measurement. Humidity tends to be difficult since it cannot be measured as precisely as many other parameters and isolating a measurement of humidity’s effect on a resistor are not easy nor quick. For an accurate measurement, all other parameters must be controlled precisely; it is not possible to measure humidity while other parameters are varying at the same time. Normally, humidity is tested in a humidity chamber with specified power applied and at a specific temperature for a specified period of time. A measurement of humidity on an unpowered resistor will result in a significantly different result which is not a normal use of the resistor; they are usually under power in the real world and should be tested as such as close as is possible.
These are just some of the common parameters involving resistors and their measurement, perhaps more to come later.
When comparing tight tolerances, the instruments in question must be calibrated and accurate enough to make the required measurement, the instrument’s accuracy and uncertainties must be known precisely if one is going to attempt PPM comparisons. For a tolerance of say ±0.01%, you could possibly get away with unknown accuracy and uncertainty of 10 PPM to 20 PPM, the problem there is if the resistor in question is close to the outer tolerance (let’s say it is +90 PPM) that within the question of accuracy/uncertainty, then it is very possible that the measurement will indicate the resistor is out of tolerance when in fact it is not or the resistor could be slightly out of tolerance and the measurement says it is within, it works both ways. ±100 PPM is relatively easy, it gets exponentially more difficult as the required accuracy increases, when approaching ±50 PPM, ±25 PPM or ±10 PPM, the instrument in question cannot be a guessing game, either it is precisely known to be accurate enough (by comparison to standards) or it isn’t and if it isn’t, it cannot be accepted to execute the required accuracy. This may be a big inconvenience but if you are going to chase after PPMs, then you have no choice but to play by the rules. Calibration was invented exactly for this purpose, to make measurements on different instruments more agreeable, that is why labs like the NIST exist, without traceability, and there would be endless arguments over whose measurements were correct.
Home
Help
Search
Login
Register
