>There are a number of reference boards supposedly removed from 3458As on eBay. >Assuming they are non counterfeit (not necessarily a valid assumption for items on eBay), how easy would they be to integrate into a box to make a stable reference? >The LTZ1000A chip is about $50 new, but to make a stable reference you need quite expensive resistors, so a used 3458A reference is attractive. >But I don't know if problems making low thermal EMF joints might totally ruin the potential performance. David, these 3458A reference boards are also quite expensive, they sell for about 150-200$. They may have problems with high drift and high noise, that's why they show up on the 2nd hand market. It's been demonstrated that it's possible to easily build such stable references with an 40$ LTZ1000 (non A), and 5 precision wire wound resistors ( <10$ each), and an LT1013, so BOM cost < 100$. The emfs are not the biggest problem, can be mitigated by proper design and thermal shielding. The biggest problem with the 3458A and the DIY references is to amplify precisely to around 10V. These resistive dividers usually drift too much over time and temperature to give reasonably stable 10V references. I dfesigned an auto-calibration-divider, similar to the 720As 1st decade, which amplifies the 7,15V exactly by 10/7, which is mostly sufficient to calibrate the 10V range of a DMM. Other solutions require these ultra- expensive Vishay hermetically sealed, oil filled VHP resistors, which easily double the BOM cost. Frank