Elektor Voltage Current Calibrator Project

[KungFuJosh]

Of course. 😉

Test stability once a week or something and see if anything changes.

[Andreas]

Phew, then I would have to let it run for another 3 weeks without a problem....
The text was, however, before the oven, which constantly heats the voltage reference to approx. 55°C - should I still keep it running for so long?

Hello,

I would monitor the voltage output during ageing to see when it stabilizes.

my experiences with MAX6350 (in DIP8-Package).
It may need 50 - 100 days of operation until the ageing slope stabilizes to ~+10 ppm/year under constant conditions.
After some years the slope goes down to ~+5ppm/year.
(in my case they are in a ageing box which is warmer than environment and is heatet to 50 deg C every day during measurement for ~ 80 minutes).

On power outage the Epoxy package of the reference soaks humidity from environment.
After power outage I had a ~5ppm jump which stabilized after some days.

Even under constant temperature you get some seasonal changes due to air humidity.
In my case the humidity goes from ~40%rH in winter to ~70%rH in summer which affects the output.
(see normalized Output from 65kHrs)

with best regards

Andreas

[Martin72]

Hi,

For a week now, the reference has been in on-off mode, i.e. on for 2 hours and then off for 1 hour, through a timer, and this has been going on for a week now.
Before that, it was constantly on for 11 days.
On Sunday, I will create a way to measure the reference output and then leave the reference on constantly again, 24/7.
When the internal 5V reference voltage appears stable, the “critical” point comes, the calibration.
What makes this reference so special, and also justifies the rather high price, is that it is not only adjustable from 0-10V, but can also be switched from + to - and also provides a reference current of up to 40mA.
The calibration effort increases accordingly.
I would love to borrow a 7.5 or even 8.5 digit meter for it.
But it is to be feared that the rent will end up costing more than sending my reference in for calibration.
But I would rather do it myself.
The temptation is, of course, to do a first preliminary calibration with my SDM3065X.
The only thing that speaks against a “real” calibration with it is that it has an accuracy of 35 ppm and that the Siglent meters, although within the specifications, basically measure “less”, which somehow provokes the monk in me.
But maybe I'll do it anyway, just to see how stable the reference is afterwards.

@Josh: Since you already want to make changes to the original design, create a test point for the 5V reference, so that you and others don't have to tinker with it like I have to.

[Kleinstein]

There is not much lost in first doing a calibration to the own meter. With only the reference in a plastic case the accuracy is anyway not that high.

The reference and rest of the circuit tend to get more stable over time. Another point is the possibility of still finding an early fault. The chance for a failure is larger in the frist year or so. Another point could be that with use one may find points one would want to add / change (e.g. a fan or changes to the case), that might effect the calibration. So if an official calibration is not needed, why rush it.

[Martin72]

Hi,

With only the reference in a plastic case the accuracy is anyway not that high.

Would replacing the “oven” housing with a metal one, as well as housing the main board in an RF enclosure, help in this regard?

[Andreas]

@Josh: Since you already want to make changes to the original design, create a test point for the 5V reference, so that you and others don't have to tinker with it like I have to.

Hello,

take care that you do not have a output voltage shift by EMI when connecting any wire to the voltage reference.
Especially when you have a noisy power supply (maybe due to a switchmode supply / LED lamp nearby).
I already had up to 300 uV shift when suppling from RS232 interface from PC or several 10 uV with the reference battery supplied.

I would put at least a 100 nF (Foil or ceramics) capacitor in parallel to the 2.2 uF at the output of the reference to minimize shift due to the test point.
(of course a PI-Filter on every output would be better to avoid EMI shifts).

see output power stage here as example:
https://www.eevblog.com/forum/metrology/ad587lw-10v-precision-travel-standard/msg1449492/#msg1449492

with best regards

Andreas

[Martin72]

This week, I had the reference continue to run in on/off mode.
So it had 11 days of continuous operation and 14 days of on/off operation.
Now it is running continuously again, at least for another week.
That should be enough, because I can't measure the reference voltage when it's installed and have to rely on the duration of the aging.
A printed metal case for the oven is on its way, and I already have a metal case for the reference board itself.
I'll actually start by using my SDM3065X to calibrate the reference to see how it goes and what the result looks like afterwards.
And then we'll see.

[Martin72]

A printed metal case for the oven is on its way, and I already have a metal case for the reference board itself.

The printed case arrived today, looks good.
I will leave the reference running until the weekend, then it will have been running for 5 weeks.
And then I will install the new case parts.

[spo256al2]

Hello Martin,

How is run your Elektor calibrator, what about the stability?

Rdgs

[Martin72]

Hi,

I will probably be able to present new findings in April.
The reference and a multimeter have now been switched on permanently for weeks, I still have to modify the reference mechanically.
I didn't have the time for that, but soon I will, and then I can post news.

Martin