Temperature controlled resistive divider? Why not?

[Moon Winx]

What are your thoughts on building a temperature controlled resistive 10:1 & 100:1 divider? Something not unlike a Fluke 752, but instead of having to calibrate the ratio before each use, the ratio would be relatively stable due to constant temperature and would only need aligning to correct for resistor drift. I know Guildline has recently started selling something similar and it includes self-aligning with an internal null meter, but I'm surprised no one seems to make a cheaper/more simple solution. Or maybe they do and I'm just unaware.

[Edwin G. Pettis]

It is a bit more complicated than just controlling temperature for stability even for PWW resistors, TCR is only one parameter, aging is another and is affected by various factors such as temperature, operating parameters such as how much power is applied and for how long, how often.  Drift is considered an uncontrolled parameter as such mainly because it depends on both internal and external factors that are not under control by the manufacturer.  It also depends on the type of resistor (and no hermetic sealing does not eliminate all factors, it helps minimize external factors and can slow down internal factors to some degree), is it PWW or film/foil, ratings, type of construction, its intended purpose and has it gone through a PMO (post manufacturing operation, otherwise known as thermal shock and any other operations deemed necessary).

There are many grades of resistors, from commercial jelly bean one cent per to the primary standard types such as the SR-104 which cost accordingly as to specifications.  There are no free lunches, you aren't going to get a transfer grade resistor out of a $2 or $4 resistor or for that matter even $20 or $30, those are not intended to be such and no matter what one does to them, will never achieve the same specs as a transfer standard resistor which is designed for the purpose.  You aren't going to make an SR-104 out of a sow's ear no matter what you do, if it could be done, it would have been done already.  Somebody is going to argue that somehow someone is going to do it, well have at it but don't bet the farm on it, remember such performance must be verified by a proper calibrated resistance system not some jury rigged mess on a table top, sorry, that is just the way it must be.  You can't verify parameters if everything is moving around at the same time.  In fact, someone else capable of verifying said performance should do it independently, again on a certified calibrated system of sufficient accuracy; no DVMs need apply here.

Could you possibly build something like a Fluke 752, yes with the proper parts but don't expect to get actual 752 performance without getting the correct parts, otherwise you are going to fall short of the target.  There are many parameters to be determined for such a project, some of which many people are not aware of.  If you want to play with PPMs, you're going to have to do it the right way or you're really just wasting time and effort.

[Echo88]

The magic lies in the fact that short term instability errors in the 752A are cancelled by the user before usage, thermal stabilization will not transform very good resistors into Quantum Hall Standards. As far as i concluded from reading the datasheet and patent of the Guildline 7520 they have improved the 752A-design and have done everything to ensure it is the best resistor-based-divider one can build, including of course heater compartments for the resistors.

I think one can copy the 752A-design for a few hundred dollars with new good resistors and add thermal regulation. But those resistors might not be as good as the preaged and selected ones in the 752A; you might not reach the 752A-long-term-specs. Dr Frank copied the 752A-design i think, that might be of interest to you.

[Moon Winx]

Sorry if I was unclear, I was asking why some manufacturer hasn't produced one (aside from Guildline just very recently). I'm not trying to build one personally and I'm not stating that I think one could be built cheaply. The Guildline box is over $25k, which isn't too bad (compared to MI's 8000 binary-divider system), but anyone who does primary DC volts transfer knows that the 752 is a time sink due to it's constant need for pre-measurement calibration. A temperature-controlled version would not need self-cal everyday therefore saving time and would eliminate a source of error.

[Edwin G. Pettis]

A temperature controlled version would take 'a little off the top' of the total drift, how much would still vary with its use.  To some degree, if a 752 has been taken care of, it should have a fairly minimal drift to compensate for after it has settled down.  Of course you've still got the switches and connections to bother with.  But yes it does take some time if you're going for best possible measurement.  A temperature controlled unit might be able to go for longer periods with smaller drift but it is hard to say just how much of the drift you are seeing is just temperature effect.

I do understand your complaint though about some calibrations being a time sink.

I am rather glad that I can repair my 242D and associated transfer standards which saves substantial cost.  The mica card resistors, particularly lower values, have a tendency to drift with age, sometimes quite a bit.  It also appears that IET has opted for less quality in some resistors they are using, no doubt because they can't make the mica card resistors as well as ESI or GR did.   

[MisterDiodes]

There is also the issue on a 752a of what voltage input level is fed into the divider - even with a temperature controlled unit you probably would still need to re-cal often if you're handling 500V ~ 1100VDC on the inputs, because the 752a is designed from day one to handle that power (Maybe you need to test HV sources?).  Now you might need a cooler AND a heater for the box if you're trying for a thermal controlled design, and it would have to be able to handle whatever heat you're pumping into the box.

If you were never going over 10~30VDC inputs to the divider a simpler and smaller thermal/ controlled design might work.

[lukier]

This Guildline device is quite interesting.

According to the patent (EP3236271A2) it uses electronic potentiometers for the automated adjustments. I wonder which digital potentiometers are good for metrology applications.

Anyone has any idea what "new optical null detection" might be? The patent doesn't go into detail on the null meter section.

[rhb]

This Guildline device is quite interesting.

According to the patent (EP3236271A2) it uses electronic potentiometers for the automated adjustments. I wonder which digital potentiometers are good for metrology applications.

Anyone has any idea what "new optical null detection" might be? The patent doesn't go into detail on the null meter section.

Optical  beats between laser diodes?  A bit of a stretch, but it might well be that there is  a sweet spot where very small changes in current result in frequency shifts sufficient to be useful.

http://www.measurement.sk/2002/S3/Bartl.pdf

Then again, optical null detection via a galvanometer is an ancient method.  Done with a quartz fiber very sensitive.  Probably several ways to do something similar  with a MEMS device.