When we are working on a new project, or debugging one, or just fixing something these items come into play. This begs the question when we make a measurement, how do we know the measurement is accurate?
Take the trusty DVM for example, I have a couple cheap handheld ones, a couple good handheld ones and a couple lab grade bench types. The good ones (Fluke 77 & 87) agree within reason with the lab grade (Agilent HP34401A) DVM on all measurements (Volt, Current, Ohms), the cheap one don't
Without a good DVM for a reference these cheap DVMs could fool you with an inaccurate reading
Suspect we've all "been there done that!"Awhile back we got involved with precision 12 & 16 bit ADCs and DACs and wanted to verify their performance, well a 4 1/2 digit Fluke 87 isn't good enough, so the quest for a more accurate DVM began. This eventually led to a surplus HP34401A and then another (long story). Someday, soon I hope, to get a few minutes with the superb HP/Agilent/Keysight 3458A 8 1/2 digit lab DVM and calibrate my 34401A DVMs, and/or calibrate a portable stable reference.
We've looked a reference standards on-line and the ones you can trust are expensive, justified as they a quality designs with "aged" critical components. Eventually after studying what's available we decided to design a DIY reference standard with goals of excellent temperature and long term stability and be portable operating from batteries or from s standard AC power supply.
The core design is based upon the older National Semi LM399/199 buried zener, which BTW is the voltage reference utilized in the HP34401A DVM!! This chip has a built-in heater to keep the chip temperature at 85C, thus dramatically improving temperature drift to less than 1ppm/C, and the buried zener has exceptional long term stability. A Precision Amplifier is required to boost the nominal 6.95V to 10.000V and should use low TC resistors. Another reference is also used, this is the Analog Devices ADRO1, which is based on the brilliant Widlar Bandgap Reference. This also has good TC and long term stability, but not as good as the LM399 and has a nominal 10.000V output so no Precision Amp is required. Now two independent references are available and depend on different mechanisms for a voltage reference.
To augment the two 10V voltage references a Precision Current reference is included and a few precision resistors. The resistors selected are low TC devices, but not the ultra low TC and expensive Bulk Film types which cost ~$38 per resistor
The LM399 Precision Amplifier design utilizes a differential technique to 1st order cancel any temperature effects due to the resistor TC, so the result should be good over normal room temperature variations. All the potentiometers are multi-turn moderate TC types and their range is just enough to cover the expected trimming range, thus limiting their effects on the precision output voltage.
Input power is from a normal supply or battery of 18~24V and an integral 15V regulator can be employed or bypassed (this is in case the design works at 12V input power which is below the range of the the Precision amplifier output drive capability, but past experience has shown they operate OK). The LM399 heater can be selectively turned off to limit current consumption when under battery power.
Anyway, here's look at the design and PCB for those interested. We'll probably be ordered this PCBs along with the Fast LED Driver PCBs soon.
Best,
