Lowest drift, lowest noise voltage reference (ADR1000AHZ)

[CalibrationGuy]

@Soulman: We've powered up a reference sitting on our shelf for months from Doug at voltagestandard.com (no affiliation) and it was less than 5ppm off. So 5ppm is quite significant and in a constant humidity environment, certainly not ideal behavior.

TomG.

[dietert1]

Can you tell how much it drifted until reaching a stable condition? Which chip? Hermetic enclosure?

[CalibrationGuy]

Yes, the time scale is typical for a humidity related process and it can happen inside the hermetic cavity.
The observed behaviour isn't that unusual. I have seen it with LTFLU references and on a LTZ1000, too. If you always turn on a precision voltmeter for a short period, you won't even notice, as the reference will remain close to the "off" condition.

Regards, Dieter

Dieter, this would imply that during the manufacturing process there was humidity present, a problem with the manufacturing process, very concerning.

TomG.

[dietert1]

Why would somebody want to run Fluke 732 references on a UPS backup unit? Don't those references have built-in batteries that keep them hot for several hours or days?

[CalibrationGuy]

Why would somebody want to run Fluke 732 references on a UPS backup unit? Don't those references have built-in batteries that keep them hot for several hours or days?

The short answer is yes. But, for someone who has several 732a units with extremely low drift, you don't want a battery failure in the middle of the night to interrupt that very carefully maintained drift history. These lead acid cells go unexpectedly. Also, unfortunately, power here has been known to glitch. Murphy's Law says that for me, these 2 events will come together to spite me.

TomG.

[Andreas]

I have not seen anybody else reporting this effect, but probably nobody else has done this test.
If this is a common characteristic of all ADR1000, this would inhibit a usage as a reference for DMMs.

Hello Frank,

I also can confirm the effect on both of my ADR1000A references.
Its only 0.3 - 0.5 ppm and restoring within 1-2 weeks but also only for a few days of (accidently) "off time".

https://www.eevblog.com/forum/metrology/lowest-drift-lowest-noise-voltage-reference/msg5188620/#msg5188620

I fear that they have some problems with the die attach.
And I hope that they do not use the same gunk now for all precision references.

with best regards

Andreas

[CalibrationGuy]

I have not seen anybody else reporting this effect, but probably nobody else has done this test.
If this is a common characteristic of all ADR1000, this would inhibit a usage as a reference for DMMs.

Hello Frank,

I also can confirm the effect on both of my ADR1000A references.
Its only 0.3 - 0.5 ppm and restoring within 1-2 weeks but also only for a few days of (accidently) "off time".

https://www.eevblog.com/forum/metrology/lowest-drift-lowest-noise-voltage-reference/msg5188620/#msg5188620

I fear that they have some problems with the die attach.
And I hope that they do not use the same gunk now for all precision references.

with best regards

Andreas

I really hope that this is not true. I wonder if they are using different processes to manufacture the original LTZ1000A. Hope they are not.

TomG.

[dietert1]

Why would somebody want to run Fluke 732 references on a UPS backup unit? Don't those references have built-in batteries that keep them hot for several hours or days?

The short answer is yes. But, for someone who has several 732a units with extremely low drift, you don't want a battery failure in the middle of the night to interrupt that very carefully maintained drift history. These lead acid cells go unexpectedly. Also, unfortunately, power here has been known to glitch. Murphy's Law says that for me, these 2 events will come together to spite me.

TomG.

You mean there is no preventive maintainance on your equipment? Although you believe in Murphy's law?

[Dr. Frank]

@Soulman, DieterT and CalibrationGuy:
Now I have understood your concerns about humidity.
Anyhow, I use hermetic resistors for the crucial 20k collector and 100 Ohm zener current resistors, the other 69k8 collector Thin Film resistor is uncritical, and for the oven control ratio divider, I use this TDP1603, but exactly in the zero gradient @ 54°C of this ADR #3. Therefore, I guess that an influence of humidity on that order of magnitude is very unlikely. 
I confirm the observation of Andreas, i.e. short interruption of a few, up to 24 hours might not shift the voltage that much, and it returns quite fast to its former value.

All the LTZ1000 /A based references do not show any relaxation effect at all, i.e. they are within 1ppm of their default output value after 1min, and within < 0.2ppm after 1h latest.
The 3458A can be switched off for weeks, and will always return to its calibrated value after 2h, (using ACAL DCV), within < 0.3ppm. So no sign of reference hysteresis or relaxation at all.

For exchange of references between volt-nuts, like the CAL-Clubs here in Germany, as well in U.S. by xdevs, we all need references, which can be shipped "cold", and stay repeatable during the whole journey. That seems to be the case for the LTZ based references.

At least, my non-battery, LTZ #3 (non-A version) and #7 (A version) can be used for environmental temperatures above ~ 13°C. #7 seems to be more robust at even lower temperatures.
I can only wildly speculate, that this has something to do with the different glueing material. I have not done any systematic analysis yet.
Both their repeatability after a ring comparison (e.g. with ab-precision) is < 0.2ppm under these conditions, which would then be their Transfer Uncertainty.

My ADR #3 is not suited for volt-nuts exchange, obviously.

I'd like to ask you guys, to do similar interruption tests with your ADRs, even if this might hurt in first place.

Frank

[CalibrationGuy]

You mean there is no preventive maintainance on your equipment? Although you believe in Murphy's law?

If you knew me personally, you would not say that even in jest. The 732a units have their batteries replaced every 18-24 months max. I don't know if you have any manufacturing experience, or dealing with QA groups in the field, but I assure you defects occur all the time despite best efforts. The battery industry is nowhere near as stringent as the Aerospace industry that I deal with. As such, batteries do fail, even new ones, and if you happen to be unlucky enough to have a battery fail during even a brief interruption of mains power, then you are out of luck. Also, 2 comments about that sentence, 1) It is an extremely unlikely event, and 2) I would never risk it even if it were as unlikely as winning the lottery because we've spent tens of thousands of dollars on calibrations and certifications to get very low uncertainties.

@Dr. Frank, I would gladly do this test as an experiment, but alas, I've not been able to obtain this device for testing. If there is anyone in the US who has this reference that they've been observing for drift and want to have us put it in our lab for long term testing against our 732a/b/c bank, let me know.

TomG.

[dietert1]

For me your contributions are welcome yet worthless as long as you painfully avoid any evidence to backup your opinions. Others in this forum present numbers, measurement logs, images and videos. Or they refer to third party contributions.
Enjoy!

[mawyatt]

You mean there is no preventive maintainance on your equipment? Although you believe in Murphy's law?

If you knew me personally, you would not say that even in jest. The 732a units have their batteries replaced every 18-24 months max. I don't know if you have any manufacturing experience, or dealing with QA groups in the field, but I assure you defects occur all the time despite best efforts. The battery industry is nowhere near as stringent as the Aerospace industry that I deal with. As such, batteries do fail, even new ones, and if you happen to be unlucky enough to have a battery fail during even a brief interruption of mains power, then you are out of luck. Also, 2 comments about that sentence, 1) It is an extremely unlikely event, and 2) I would never risk it even if it were as unlikely as winning the lottery because we've spent tens of thousands of dollars on calibrations and certifications to get very low uncertainties.

@Dr. Frank, I would gladly do this test as an experiment, but alas, I've not been able to obtain this device for testing. If there is anyone in the US who has this reference that they've been observing for drift and want to have us put it in our lab for long term testing against our 732a/b/c bank, let me know.

TomG.

We can personally confirm that the Aerospace Industry here in US was quite stringent, at least where we spent 20 years at Honeywell, they paid attention to every detail however minute!!

Best,

[CalibrationGuy]

I don't know if you were in Honeywell's industrial division, but they made the best controllers. I have customers who have UDC3000s in service for over 30 years.

TomG.

[Andreas]

Hello,

I now passed 19 kHrs:

nothing new so far:

the 10V/7V divider on ADR#01 drift has stabilized but has still a slight upward drift.

The 7V outputs of ADR#01 and ADR#02 trend to further settling.

Detailed explanation of the diagrams see post of 14 kHrs
https://www.eevblog.com/forum/metrology/lowest-drift-lowest-noise-voltage-reference/msg5138967/#msg5138967

with best regards

Andreas

[aronake]

Now officially in stock at Analog Devices. Need to buy 100 though, and only 52 in stock. But at least things are moving.

[Andreas]

Hello,

I now passed 20 kHrs:

the 10V/7V divider on ADR#01 drift has stabilized.

The 7V outputs of ADR#01 and ADR#02 are now in the target range of ~1ppm/year.

Detailed explanation of the diagrams see post of 14 kHrs
https://www.eevblog.com/forum/metrology/lowest-drift-lowest-noise-voltage-reference/msg5138967/#msg5138967

with best regards

Andreas

[branadic]

Thanks Andreas, I'll keep my updates of 1100 days of continiuously operation hopefully for MM2024. This month 3458A will be cal'ed.

-branadic-

[Andreas]

Hello,

I now passed 21 kHrs:

nothing new, continuous stabilizing.

Detailed explanation of the diagrams see post of 14 kHrs
https://www.eevblog.com/forum/metrology/lowest-drift-lowest-noise-voltage-reference/msg5138967/#msg5138967

From MM2024 I have the following indications for my LTZ#4 and LTZ#9
(LTZ#4 drift has to be added to the diagram since it is reference, LTZ#9 diagram was added as plausibility check)

LTZ#4: Drift -1.74 ppm/3.2 years or -0.55 ppm/year
LTZ#9: Drift -3.66 ppm/3.2 years or -1.16 ppm/year

with best regards

Andreas

[ITArchitect]

Interesting how ADR 1 & 2 drift is so similar.  Here's hoping the ADR1001's are more like the LTZ1000's in stability.   

[Andreas]

Hello,

I now passed 22 kHrs:

nothing new, continuous stabilizing.

Detailed explanation of the diagrams see post of 14 kHrs
https://www.eevblog.com/forum/metrology/lowest-drift-lowest-noise-voltage-reference/msg5138967/#msg5138967

with best regards

Andreas

[branadic]

I'm slowly approaching 1200 days of observation, here is an update.

-branadic-

[Kleinstein]

The drift for the last 200 hours looks good. Its in a range where it is hard to tell which of the reference / meters is actually drifting.

Somthing happend to the ADR1000 at around 1080 days - maybe a power down / move ?

[branadic]

At around day 1080 I've added the raw zener output to the reference, which caused a sudden jump and which was already shown at MM2024.

-branadic-

[Andreas]

Hello,

I now passed 23 kHrs:

looks like after a plateau phase the devices begin to drift again (slowly).

Detailed explanation of the diagrams see post of 14 kHrs
https://www.eevblog.com/forum/metrology/lowest-drift-lowest-noise-voltage-reference/msg5138967/#msg5138967

with best regards

Andreas

[branadic]

Could you plz. plot drift (with the drift rate itself removed for sure) vs. ambient temperature? Is it just me or is there a small correlation?

-branadic-