How good are these parts if I build a voltage reference out of it?

[macboy]

Scored a few used LM399 at US$2.5/each, end of discussion. I am going to design my standard project around these.

Congratulations -- good to see that your project is making progress. With that choice it will make sense to have your voltage reference calibrated once it's built, since the LM399 has great stability, but a rather large 2% initial tolerance.

So your suggestion here is go ahead and build the voltage reference out of LM399, and send it on a trip to the cal lab? Will its status being a used part affect its accuracy?

Yes, the status as used affects it... for the better. It is well known that these heated zeners drift the most in the first few hundred hours of use, then level off to a much better long-term stability. A used device is actually best, otherwise you need to 'burn in' the new device for several months in order to get over that initial period of drift.

The LM399 has excellent stability, but poor initial tolerance. That means that it any two of them could be at significantly (a few %) different voltages, but they will each be extremely stable, over a range of temperature, and in the short term and long term. After building your device, you will need to get it calibrated so that it is putting out exactly some value that you choose. Then (if you did it right) it should remain quite stable over time.

Only you can decide what amount of accuracy is needed, and you need to decide this before starting the design of the project. After you go beyond about 0.01% (100 ppm), you get into volt-nut territory where everything matters. Choice of components becomes critical: low drift, precision DC op-amps, resistors need to be special expensive low tempco ones. And the layout of the PCB becomes important too.

If you want to make it adjustable, you might try building a couple of decades of Kelvin-Varley dividers. You could use relays in place of the switches, but I'd suggest using latching relays (less self-heating) and make sure that they are signal (or telecom type) not power relays. Depending on your specifications, you might look for ones with low thermoelectric contacts.

There are various threads here on eevblog and other forums discussing building precision references. Search and read, and try not to lose your mind in all the volt-nuttery.

[technix]

Scored a few used LM399 at US$2.5/each, end of discussion. I am going to design my standard project around these.

Congratulations -- good to see that your project is making progress. With that choice it will make sense to have your voltage reference calibrated once it's built, since the LM399 has great stability, but a rather large 2% initial tolerance.

So your suggestion here is go ahead and build the voltage reference out of LM399, and send it on a trip to the cal lab? Will its status being a used part affect its accuracy?

Yes, the status as used affects it... for the better. It is well known that these heated zeners drift the most in the first few hundred hours of use, then level off to a much better long-term stability. A used device is actually best, otherwise you need to 'burn in' the new device for several months in order to get over that initial period of drift.

The LM399 has excellent stability, but poor initial tolerance. That means that it any two of them could be at significantly (a few %) different voltages, but they will each be extremely stable, over a range of temperature, and in the short term and long term. After building your device, you will need to get it calibrated so that it is putting out exactly some value that you choose. Then (if you did it right) it should remain quite stable over time.

Only you can decide what amount of accuracy is needed, and you need to decide this before starting the design of the project. After you go beyond about 0.01% (100 ppm), you get into volt-nut territory where everything matters. Choice of components becomes critical: low drift, precision DC op-amps, resistors need to be special expensive low tempco ones. And the layout of the PCB becomes important too.

If you want to make it adjustable, you might try building a couple of decades of Kelvin-Varley dividers. You could use relays in place of the switches, but I'd suggest using latching relays (less self-heating) and make sure that they are signal (or telecom type) not power relays. Depending on your specifications, you might look for ones with low thermoelectric contacts.

There are various threads here on eevblog and other forums discussing building precision references. Search and read, and try not to lose your mind in all the volt-nuttery.

I am creating a LM399 "VoltCard" (voltage reference with a 5-pin card edge connector) and whatever precision I can get out of that is going to be my target precision. The op amp is OP07C and my PCB house have limitations, so no volt nutting allowed here.

[slurry]

I found two new LM399H so i decided to make a PCB for them, as i'm no volt nut the PCB is rather simple.
I used OP77 as buffer, i can add the link to OSH park PCB when i get home if anyone is interested in a PCB?