Error budget for any reference designs

[mendip_discovery]

As the has mentioned it I thought it might be worth exploring this subject. I have two LTZ1000 voltage references here and as an experiment I would like to know how to go about deriving a specification or at least error budget for them.

Mains voltage swing will be one. Though I am not sure how you get to 20 µV from a 5% swing in mains voltage.

[TimFox]

That is still called "line regulation", but that is a misnomer since the circuit rejects a change in the line, rather than regulates the line.
It can be defied in terms of
-Fractional change in output voltage divided by fractional change in input voltage (e.g., ppm/%)
-Fractional change in output voltage divided by absolute change in input voltage (e.g., ppm/V)
-Absolute change in output voltage divided by absolute change in input voltage (e.g., mV/V)
-etc.
For a valid specification, this measurement must be done with all other parameters, including load current and junction temperature, held constant during the change.

[alm]

I've never seen an LTZ1000 design that is powered from unregulated mains voltage. So it's not just the regulation of the LTZ, but also that of the regulator powering the LTZ that compensates for mains voltage fluctuations.

[TimFox]

To get the final line-regulation result for a rectifier-gross regulator-reference regulator design, one needs to first compute the effect of the AC line on the rectifier, then the effect of the rectified voltage on the "gross regulator", and finally the effect of the gross regulator output on the reference regulator output, all while keeping the other operating parameters constant.

[mendip_discovery]

So much like with mos test gear voltage swing has a negligible effect on the reference.

I guess the only other main contributors are ambient temperature and connection errors.

[TimFox]

Also, the load current on a reference regulator has a small effect, but measurable.

[Andreas]

Hello,

depending on the reference, you should also consider influence of humidity.
Especially with references in plastic package (I have measured up to 0.5PPM/%rH) or SMD references which are soldered on epoxy PCB directly.

with best regards

Andreas

[dietert1]

Most of the time people do not use a mains replacement circuit with variable, yet controlled mains voltage. Also they don't use a controlled humidity chamber to really pin down the effect. Instead they observe their precision devices under "natural" ambient variation and it may become difficult to separate various effects.
A mains replacement circuit i made two years ago used a DDS generator, an audio power amplfier and a mains transformer to get 220 V AC at 50 Hz. Easy to make, except one needs to watch out for safety. Another version is an isolation transformer we have with +/- 10 % settings. With unstable mains voltage the analysis will be more work though.
For humidity i used bottles with used drypacks. When stable after some hours each drypack would provide a different relative humidity. That was used to compare humidity sensors.

Regards, Dieter