What's the Best Process to Age a Voltage Reference? (in up 3 months)

[Trader]

I'm asking about the Better Aging Process, maybe up to 2,000 hours (~3 months) max. A process to Just Age the reference Not measuring it during the process (only After that, measure it for a couple of days to verify if seems to be more stable).

I know the "perfect" process could take years and some references require different processes, but is there some General Rule of Thumb, like:

1) just power on the reference?, or also put a load to draw 1mA, 5mA, etc (at some intervals)?
2) could be at room temperature? or must be in a heated box?
3) can I use a switching (maybe noisy and not so stable) power supply? or must be linear or even under batteries?
4) should I power it on and power it off at some intervals?, or just keep it on continuously?
5) etc.....

Just to be specific, I want to age some "AD586MN" references, but feel free to explain the process for other voltage references.

I'm inspired by this process, although 7 years is a looong time for me (https://www.febo.com/pipermail/volt-nuts/2011-September/001046.html):

1) Aged at 50 MA for a year,
2) Run in at 7.5 MA for another year,
3) Then No power Aged for another 5 years,
4) Tested for stable and repeatability for better than 1PPM,
5) Tested for low noise for better than 0.1PPM,
6) Has a room temperature 'Zero TC' under of 0.1PPM
7) and the Zero TC current is a under 5 ma.

Thank you so much.

[iMo]

The guy describes his 1N825 aging exercise (it is a TC compensated diode) not the AD586 references you want to age. I would pre-select one with lowest TC at temperature of interest, I would look at its popcorn noise, solder it into the final pcb and let it run for some time. Mind each soldering, powering on/off, mechanical stress, large temperature changes as well as humidity (in case of epoxy packages) could mess up with the reference parameters. I've seen somewhere people heated up that kind of references by running the max 10mA out of it, and the heat (P in Watts = (Vcc-5V)*0.01A) had "speed up" the aging.
I would not consider epoxy packages, no way, it is a waste of time and $.

BTW., there is a lot of threads on this kind of references, ie from Andreas
https://www.eevblog.com/forum/metrology/ad587lw-10v-precision-travel-standard/

[ap]

As stated above, it does not make ANY sense to use this part and age it. You need to have a hermetic part (e.g. CERDIP or metal package) in conjunction with a burried zener reference if you want to apply pre-aging to reduce drift.

[Andreas]

Just Age the reference Not measuring it during the process

I want to age some "AD586MN" references,

Hello,

Ageing in 2 kHrs with a reference: take a hermetic heated reference like LM399 / LTZ1000(A)
But even then the drift will be above the target of 1 ppm/year.
Other references like AD586LQ will take minimum double the time for initial ageing.

In plastic package you will have additionally humidity drift.
On 2 samples (same batch) of LT1027C I have measured 0.5ppm/% rH. (so ~12-15 ppm seasonal change over 1 year in my area).
Lars has mentioned 0.1-0.3 ppm/% rH for Analog Devices (I think it was AD587).
But since I do not know if the relation is really to %rH or absolute air humidity I do not know if the values can be really compared.

What I would not do:
- ageing at much higher temperatures as the following usage. This may introduce hysteresis and a "new start of ageing" after heating.
- ageing without monitoring (you should monitor for "popcorn noise") so you need at least some minutes of monitoring every day.

other notes:
- Switching power supply is no issue as long as you do not measure during ageing
- I have had luck with one AD586LQ reference which was drifting a lot after 1/2 year by a loading with 15 mA (1.5 hours load, 0.5 hours no load) during night. After some weeks/months the ageing suddenly stopped. But on other devices this procedure did not help.
https://www.eevblog.com/forum/metrology/ad587lq-from-aliexpress/msg2317686/#msg2317686

with best regards

Andreas

[GigaJoe]

what i did was:
around 30 zeners with resistors, in thermally isolated enclosure
supply V from transformer and  diode, no caps, half period
so they self heated and feed with current pulses

no idea if it helps, but who knows