Yes, I'd see your point depending on what it costs in your area to get calibration and the cost of hardware.
eBay items change the field, so some thoughts.
A well bidded for 3456a is typically ~ $100 and, in USA, commercial calibration another $100; all those prices include shipping a heavy DMM. HP's data gives us a DMM accuracy of ~ < 25 ppm year. You can use it to measure stable but uncalibrated sources to compare your other meters. In a more traditional way, the EDC MV106 reference Dave reviewed has similar accuracy as the $180 reference, and in November, 2011, was as low as $50, uncalibrated. +$100 for formal calibration and you have a calibrator that scale VDC with nV resolution.
But shipping eBay items overseas, can cost more than the item, not to mention your local calibration costs, so a compact, albeit limited source has appeal.
Finally, theoretical concerns. Since we are paying the extra money to obtain higher accuracy, were more uncertainty exists, the technique of calibration has to defined a bit more so we are less uncertain about our uncertainty
![Shocked :o](https://www.eevblog.com/forum/Smileys/default/shocked.gif)
.
For example, at 25ppm accuracy for $180 reference, the preferred reference to calibrate with TUR of 4:1 is ~6 ppm accuracy.
The basic 3458a has accuracy at 10 V: 0.5ppm/24h, 4.1 ppm/90 days, 8ppm/1 year, 14ppm/ 2 years.
To adjust his reference with less uncertainty they need a 3458a at < 1 yr spec for its 10V range. Is this certificate statement for a 1 year or 2 year cycle?
http://www.voltagestandard.com/Home_Page_JO2U.htmlGeller's method of adjusting 10V using a 732a is less uncertain because his source is better. The 1 yr rating of 732a is 3ppm and has a 1 year cal cycle.
http://gellerlabs.com/Voltage%20References.htmFor better results, the standard for calibrating references to 10 ppm is ~ 10x better, or 1 ppm, which is what the Fluke reference can do. Malone's calibration 25 ppm, a bit low for $180 price. The effect of shipping the SVR to an owner unpowered has risks. Lastly, its more common to use 10V in metrology work, not 5V.
The $180 is not a bad deal. It uses a Thaler hybrid reference IC that outperforms all of the monolithic reference ICs currently available. When Thaler owned the company, these chips were far more affordable, but now Cirrus owned Thaler, the prices jumped up many times. Last time I looked, I was seeing prices like $70.
The specs quoted are for 6 month accuracy , and a +/- 6 deg F temperature variation and a 8V input supply variation and I think they are conservative specs. The actual 1000 hour stability of these chips is typically 6ppm, so if you new the temperature and supply voltage that the reference was calibrated at, it would typically be within 7ppm after 6 months. The chips could easily drift a lot less by the end of 2 years.
The price includes recalibration for 2 years, so i think $180 is OK.
5V instead of 10V? not sure if it matters much. If you are calibrating s 59999 full scale meter, 5V is useful. 10V is only 1/6th of full scale - pretty useless. Whatever you have, you will need to scale it eventually to the voltages you need. you just need a starting point.
Richard.