I am looking for NPO capacitor pairs that match each other's capacitance value closely. Their values can be 10% but the difference between the parts in the pair must be much closer.
100 Volt parts around a few nF. Does anybody make these?
Will 2% matching work for your application?
Using a Digikey search reveals several 1% tolerance 100 volt NPO capacitors around a few nF. The 399-9811-ND is in stock and is a leaded 4700 pF, 100 V, NPO ceramic having a tolerance of 1%.
0.1% is pretty tight for capacitors and is +/-4 pF for a 4000 pF capacitor. To match 4000 pF to this tolerance will require good measurement repeatability. A mechanical fixture can be built to hold the caps and if the caps are surface mount a tweezer probe can be used.
Meter resolution and short term repeatability (drift) are all important to matching these caps. The caps can be measured over a period of minutes and binned into 8 pF wide bins. The "matched" caps can then be compared one right after the other to further reduce the effect of meter drift. These types of caps may have a temperature coefficient of 30 ppm/deg C and so a 33 deg C difference between two caps will consume the 0.1% (1000 ppm) tolerance. So, they should not be set into the fixture with fingers and should be handled with tweezers.
Starting with 1% tolerance caps having a Gaussian distribution and covering two standard deviations about 2/3 of the caps will fall within 0.35% of the nominal value. But if the 1% caps have been selected from a wider distribution (such as 5 or 10%) the distribution will be rectangular. If this is the case the values will be essentially evenly distributed within 1% of the nominal value. With a rectangular distribution 100 caps having a tolerance of 1% should fall into 5 bins and with a Gaussian distribution 70 of 100 caps should fall into two bins.
But it's not always this simple. Take a look at this application note from Maxim.
When using a thermistor as the temperature sensor, it becomes the most important component for the temperature regulating system. To achieve high system performance, the thermistors should have low short term and long term drifts in the R-T relationship, the resistance vs. the temperature, compact size, and reliable. Some low cost thermistors are coated with epoxy materials and have large short and long term resistance drift, i.e., when sensing the same temperature, the resistance changes with time, causing the system with a wrong sensed temperature, thus, the target object temperature is drifted. Glass coated thermistors, on the other hand, have almost zero short and long term drift, all our thermistors are of this type. When choosing the thermistor for a temperature control system, these are the main specifications to consider: operating temperature range, sensitivity, drift, size, mounting approach, etc. Our thermistors and thermistor assemblies are mainly NTC thermistors, and can be used with our TEC controllers and TEC modules for building high stability low drift temperature control systems. These NTC thermistors and thermistor assemblies have high precision (+/-0.25C@25C), compact size (bare version down to 0.8mm diameter), and are coated with glass material on all the bare thermistors and the thermistor of which the thermistor assemblies are made, resulting in low both short and long term drift. The NTC thermistors come in bare and lugged forms, the former is for point sensing to achieve high accuracy for specific points, such as laser diode die, etc.; the latter is for bulk temperature sensing, such as optical components, etc. If you have any questions about choosing and using thermistors and thermistor assemblies, let us know. We try our best to help. If you need our products ,please see here http://www.analogtechnologies.com/thermistor.html