Upgrading voltage reference

[Jester]

Hi,
I have an old  Data Precision 8200 dc calibration source that uses a LM299 voltage reference.  I’m wondering if there would be a worthwhile benefit to upgrading this reference to something like a LTZ1000?

There is the obvious complication of the change from 6.95 to 7.2V, and it would also require a daughter board. Or perhaps there is a pin compatible part better than the LM299?

Has anyone performed a reference upgrade on one of these older calibrators and what was the end result?

[TiN]

Why not ADR1399? Also the performance of the unit is not limited by even LM299, but rather rest of the amplifiers\dividers and DAC performance.

[Jester]

Why not ADR1399? Also the performance of the unit is not limited by even LM299, but rather rest of the amplifiers\dividers and DAC performance.

Well the ADR1399 seems like it would be an easier upgrade. Much less noise and long term drift is about 1/3.

I will need to open up my 8200 to see what the actual voltage is, as these seem to have a fairly wide range (6.8 to 7.1V) vs. 6.75 to 7.3V for the ADR1399, I would imagine that other than a full recalibration after some burn in time that the reminder of the circuit would accommodate the possible 200mV shift up?

I don’t recall what year my 8200 was manufactured but safe to say that LM299 is as aged about as much as a fine wine at this point. If I do this how long would you let it age before calibration?

This feels like a slippery slope, I see a 7.5 digit DMM in the not so distant future

[Andreas]

If I do this how long would you let it age before calibration?

Hello,

I have no results on ADR1399 but the ageing on LM399 has already a very large stray.
Most of the ageing is done after 200-400 days. Final stabilisation needs up to 1200 days on some devices.
Some devices still have large day to day variations (1-2 ppm) in final state.

So if you want to get really a improvement you should start e.g. with 10-20 samples and daily monitor the individual behaviour.
Then select after some kHrs those with the lowest drift and lowest day to day variations.

with best regards

Andreas

[srb1954]

I will need to open up my 8200 to see what the actual voltage is, as these seem to have a fairly wide range (6.8 to 7.1V) vs. 6.75 to 7.3V for the ADR1399, I would imagine that other than a full recalibration after some burn in time that the reminder of the circuit would accommodate the possible 200mV shift up?

I don't think there is sufficient range on the adjustment  pot to allow for the changed reference voltage. However. there are some select-on-test resistors in the voltage divider off the main reference so it should be possible to adjust one of those to accommodate the new reference voltage.

I don’t recall what year my 8200 was manufactured but safe to say that LM299 is as aged about as much as a fine wine at this point. If I do this how long would you let it age before calibration?

Your 8200 is probably at least 30 years old so the LM299 has done most of its aging by now. If you were to put a better reference chip in you would still need to let it age for at least 6 months before its aging rate was comparable or better than the older LM299.

[srb1954]

Why not ADR1399? Also the performance of the unit is not limited by even LM299, but rather rest of the amplifiers\dividers and DAC performance.

Rather than changing the reference it might be a better path towards reducing drift by upgrading some of the components in the DAC circuitry or output amplifier.

Easiest option would be to upgrade the AD517L op amp in the output amplifier and maybe the MSB stage of the DAC. The AD517L op amps used in these positions are pretty good but there are better devices around these days with lower offset drift and input bias currents leading to better zero stability for the whole unit.

It might be also worth while changing some of the critical divider resistors for improved temperature stability. My unit, which was built in 1991, already has Vishay resistors in the most critical positions but the PCB was originally designed for conventional axial resistors. If your unit is an older type upgrading to Vishay resistors would be an improvement although this process is made more difficult as many of the critical resistors are supplied as factory matched sets and you would need to replicate the matching criteria.

[Jester]

Why not ADR1399? Also the performance of the unit is not limited by even LM299, but rather rest of the amplifiers\dividers and DAC performance.

Rather than changing the reference it might be a better path towards reducing drift by upgrading some of the components in the DAC circuitry or output amplifier.

Easiest option would be to upgrade the AD517L op amp in the output amplifier and maybe the MSB stage of the DAC. The AD517L op amps used in these positions are pretty good but there are better devices around these days with lower offset drift and input bias currents leading to better zero stability for the whole unit.

It might be also worth while changing some of the critical divider resistors for improved temperature stability. My unit, which was built in 1991, already has Vishay resistors in the most critical positions but the PCB was originally designed for conventional axial resistors. If your unit is an older type upgrading to Vishay resistors would be an improvement although this process is made more difficult as many of the critical resistors are supplied as factory matched sets and you would need to replicate the matching criteria.

Did you upgrade your AD517L? If so what did you use for the upgrade?

[srb1954]

Why not ADR1399? Also the performance of the unit is not limited by even LM299, but rather rest of the amplifiers\dividers and DAC performance.

Rather than changing the reference it might be a better path towards reducing drift by upgrading some of the components in the DAC circuitry or output amplifier.

Easiest option would be to upgrade the AD517L op amp in the output amplifier and maybe the MSB stage of the DAC. The AD517L op amps used in these positions are pretty good but there are better devices around these days with lower offset drift and input bias currents leading to better zero stability for the whole unit.

It might be also worth while changing some of the critical divider resistors for improved temperature stability. My unit, which was built in 1991, already has Vishay resistors in the most critical positions but the PCB was originally designed for conventional axial resistors. If your unit is an older type upgrading to Vishay resistors would be an improvement although this process is made more difficult as many of the critical resistors are supplied as factory matched sets and you would need to replicate the matching criteria.

Did you upgrade your AD517L? If so what did you use for the upgrade?

No. I haven't progressed that option yet.

I am currently concentrating more on getting a precision voltage standard set up and maybe making a precision divider to allow characterisation of my DMMs and the 8200 before deciding whether to upgrade anything.  I haven't yet gone full volt-nut so I might not bother chasing up that last ppm of performance.