Ltz1000 burn in

[sahko123]

I've got 2 ltz1000s and an ltz1000a that are unused and non-aged. I want to age them or burn them in but I don't have an oven to do so. Would it not work fine to just set up the heater section to burn in the ltz1000/a? Would I need to have current to the zener as well as heater control? And for faster burn in should I set the temp very high? (within max specs of course). Is it also known whether or not the early life drift is accelerated by higher temps? Or is this just zener based rather than burn in?

Thanks for any and all responses!

[branadic]

The only public available recipe is the one provided by Cern: https://xdevs.com/doc/CERN/1101699_V1_Burn_in_of_LTZ1000.pdf

I've aged LTZ references myself and used a styrofoam box, insulated with heat resistant material inside with a heated Fischer case and the aging board sitting inside of that to keep the references at temperature 1, that rises to temperature 2 if the heater of the references are turned on. The reference heater is then switched on to push it to temperature2 and off to get back to elavated temperature 1.

-branadic-

[Andreas]

Hello,

the most important question for me is: did it really accelerate the ageing?
From my experiments with the successor ADR1000 I get more and more the impresion that a burn in gives nearly no advantage on stabilisation time.
(at least for my 2 samples #1 and #2)

https://www.eevblog.com/forum/metrology/lowest-drift-lowest-noise-voltage-reference/msg4189336/#msg4189336

My LTZ1000(A) #3-#9 usually stabilized during the T.C. adjustments (1-2 kHrs) to near final value of 1-2 ppm/year.
(Except LTZ#8 which had ~4 ppm during the first year and ~2ppm in the 2nd year and is now stabilizing).

My used final burn in cirquit is here:
https://www.eevblog.com/forum/metrology/lowest-drift-lowest-noise-voltage-reference/msg3839447/#msg3839447

The PCB is shown here:
https://www.eevblog.com/forum/metrology/lowest-drift-lowest-noise-voltage-reference/msg3837242/#msg3837242

The heater and the thermal isolation with a coffee mug is shown here:
https://www.eevblog.com/forum/metrology/lowest-drift-lowest-noise-voltage-reference/msg3788051/#msg3788051
https://www.eevblog.com/forum/metrology/lowest-drift-lowest-noise-voltage-reference/msg3812672/#msg3812672

with best regards

Andreas

[branadic]

I can only speak for the LTZs that I've pre-aged and they stabilized within 2 weeks of operation after that burn-in, but for ADR1000 

-branadic-

[Dr. Frank]

Supporting the statement of Andreas, I don't recommend a high temperature burn-in (as CERN method) on the LTZ1000/A for volt-nuts use.

If you simply chose a lower oven temperature of 45..55°C, and take care not to create temperature shocks during soldering, and no low temperature storage of the unheated chip (below 10°C), then probably each of your assembled references will have a drift of less than - 1ppm/year, after maybe an initial drop of -1 .. -2ppm within a few weeks after assembly.

All of my 7 references showed this behavior.

It's much more important, not to create any hysteretic effects also during the operation of the reference assembly, by such events as described above. These might cause much bigger deviations of up to 5ppm.
 
That's the reason, why the Wavetek / Fluke 7000 references feature that temperature cycle mechanism.

Frank

[NaxFM]

It's amazing how I opened the forum to ask how to age a voltage reference and the first topic I see is about aging voltage references...
Well, i may just use this topic for my question then. Maybe an answer to my question will help you with your own problem

I want to experiment with less noble voltage references which don't use ovens. How can I rapidly age them to simulate months of powering on in just few days?
I know that manufacturers use some sort of aging process for many things, but I don't know how they do it. I have a temperature controlled oven. Would it be sufficient to cook the boards at a set temperature for few days, maybe while still powered on?

[bdunham7]

I want to experiment with less noble voltage references which don't use ovens. How can I rapidly age them to simulate months of powering on in just few days?
I know that manufacturers use some sort of aging process for many things, but I don't know how they do it.

I suppose it depends on the model, but I'm fairly sure that A) cooking them won't guarantee anything and B) many of those references are selected for stability, not aged to stablity.  IOW, they set up reference farm with a whole bunch of references and then bin them according to how stable they are.

[branadic]

If you simply chose a lower oven temperature of 45..55°C, and take care not to create temperature shocks during soldering, and no low temperature storage of the unheated chip (below 10°C), then probably each of your assembled references will have a drift of less than - 1ppm/year, after maybe an initial drop of -1 .. -2ppm within a few weeks after assembly.

At least my experience is different here, LTZs that were not pre-aged took much longer to stabilize and I thought your green dotted LTZs stabilized much faster too?

-branadic-

[Andreas]

Maybe an answer to my question will help you with your own problem

I want to experiment with less noble voltage references which don't use ovens.

Hello,

I fear the ageing mechanisms of a buried zener in metal can case is very different from other types of references and/or in different packages.
So I would open a new thread with the exact type of reference (and package) in headline to get suitable answers.

with best regards

Andreas

[Dr. Frank]

Hello branadic,
so what's your quantitative experience?
All my 7 LTZ1000 stabilised within a few weeks, i.e. after this initial downward drift of several ppm, all went into the usual downward drift. After the first year, it was mostly -1ppm.
In the following years, drift decreased on some references, on others it stayed at these -1ppm/year, or took longer to decrease in rate. I would not call that behavior "stabilization".
These 2 other 'green' references were LTZ1000As, and show only up and downward drift  of +/- 0.5ppm over 2 years , or so... thats correct.
Anyhow, on a scale of -1ppm/yr. for these untreated LTZ1000,  other effects are much more disturbing than to have even more initially stable references available.
In other words, for volt-nuts with maybe unstable environmental conditions, it would be of more importance to cancel those effects.
Not to forget, that even a burn-in on the 7000 reference chips does not guarantee a low drift... my own, quite old 7000 drifts constantly at -1ppm/yr.
Frank

[NaxFM]

Hello,

I fear the ageing mechanisms of a buried zener in metal can case is very different from other types of references and/or in different packages.
So I would open a new thread with the exact type of reference (and package) in headline to get suitable answers.

with best regards

Andreas

Yeah, you are probably right.
I just opened a new topic. Thank you for the advice!

[branadic]

Hello Frank,

so what's your quantitative experience?

I can't seem to find the data at the moment, but from what I remember the reference drifted upwards in the first two weeks and settled there. I currently don't have the reference under observation, due to limited measurement capabilities - you can't have enough 8.5-digit meters or nanovoltmeters or scanner/multiplexer or all of it - but I will get back to it.

These 2 other 'green' references were LTZ1000As, and show only up and downward drift of +/- 0.5ppm over 2 years , or so... thats correct.

So the results are inconclusive? If they "show only up and downward drift of +/- 0.5ppm over 2 years" that's different to "after this initial downward drift of several ppm, all went into the usual downward drift. After the first year, it was mostly -1ppm. In the following years, drift decreased on some references, on others it stayed at these -1ppm/year, or took longer to decrease in rate.", at least to my understanding. 

The LTZ in W/F7000 are specified to have a stability of -0.8 ± 0.7 ppm (1 year, ± 1° C, constantly energized) [1] but group annual mean drift of -0.06 ppm (45°C) with 0.25 ppm standard deviation have been reported in [2]. They use LTZs that were pre-aged. So why you recomment not to pre-age voltage references?

-branadic-

[1] The 7000 Volt Maintenance System - Users Manual
[2] Setting new standards for DC Voltage Maintenance System - A Solid State DC Reference System, John R. Pickering