Voltage divider on the output of a voltage reference - Tests and considerations

[Kiriakos GR]

This is a quick presentation of my very own voltage divider construction which now serves at dividing 10V voltage reference voltage down to 5 Volts.
At the moment my most high resolution multimeter this is BRYMEN BM869 at 500.000 counts at 5V range 0.02% accuracy.
In order to use this rich resolution that is equal to old times benchtop 6 ½ meters, I was forced to create my very own voltage divider.
Project success did not come easy, I was have to spent significant time to trim the circuit so it voltage output to be adjusted at wanted detail.

Resistors selection this is Vishay Film 0.1% 10K.
Resistor divider consumption this is 0,5 Milliamperes.
Voltage output this was adjusted (circuit trim) to match impedance of 10 Mohm multimeters.

I had to take apart one BNC to dual banana plug (cheap version) so to obtain two male banana 4mm pins, which was soldered over the PCB, underneath banana plugs they Plug-in directly to voltage reference female bananas.

Therefore I did create a Plug-In resistors divider, easy to use, easy to remove.

My learning curve by experimenting on this project it did loaded me with new experiences.
Several hours of data logging made it clear to me amount of noise floor of BM869 in microvolts (ADC noise), and that I can not go any further exploring my voltage reference stability with out using a higher in class multimeter.
My small victory this is enjoying four zeros stability and the fifth last digit this is simply an indicator of ADC noise.

I am using third party Japanese software as data logger, six digits resolution and graph chart with 100 microvolt stepping.
 
   
BM869 this is a feature rich and accurate multimeter which seems as equal or better performer compared to Fluke 8842A, but BM869 this is very sensitive to temperature change, even by measuring a true stable reference in the range of 23C down to 17C there is about 14 counts drift to lower value.
I am now thinking to create a thermally isolated (storage box for data logging) for it, with active heating and thermostat so this staying at 23C day and night.
The major advantage of Fluke 8842A this is great stability as soon it gets warm to 22 Celsius due it own ACV transformer.

[Kiriakos GR]

This article it would no be complete with out the addition of clues regarding how a higher in precision and in resolution multimeter this does measure such a voltage divider.

New factor in this game this is the comparison between impedance of 10M Ohm VS 10G Ohm that is what quality benchtop multimeter they can also use.

First picture this is at 10M Ohm, this is comparable to handheld multimeter impedance.
 

My Fluke 8846A this is superior than any other tested multimeter regarding technology in use, calibration this is verified, and it does have much less internal noise when this is compared to ancient 8842A.
 
Second picture this is at 10G Ohm impedance setting, as you can see 10G Ohm this really influencing Voltage divider circuit and voltage output this is now higher by +0.0025 %.

What worth's remembering for my tests this is a single significant detail, there is no Voltage divider this be compatible for both multimeter impedance's at 10M & 10G.
My circuit this is trimmed for 10M and I very happy regarding it performance.       
8846A belongs to high class of benchtop DMM and it is normal this to be able to measure in finer detail because of it higher sensitivity and it lower ADC noise.

Standard deviation of my reference voltage source this is down to 4uV, and therefore neither digital filter at 8846A this has much of work to do.     

[Kiriakos GR]

This is hart of summer update and I am speechless.
Back in May I did use my 8846A taking measurements’ at much lower room temperature (24~26C) and I did repeat this test now in August that indoor temperature this is at 30~31C that is troublesome for living people.

And in my surprise this voltage divider due the quality of resistors in use, it seems to not obey at the temperature coefficient law.
While my voltage reference at 10V this is affected by temperature coefficient and delivers as low as 9.99998 at 31C, the voltage divider does not demonstrate any change at all.

In both tests with Impedance at 10M and 10G Ohm, voltage output this is identical to my test this made back at May.

I can only conclude that while voltage divider this demonstrates small change in output this related to impedance value, other than that, this is a rock solid solution for trimming down voltage (when DC current considerations are well calculated so the resistors to operate with out be overloaded).

I am very happy, my First DIY voltage divider for metrology testing, this is now a proven success.
Naturally and the Vishay film metal resistors they worth some credit … Lol   

[Kiriakos GR]

This is hart of winter update and I am also speechless.

How about evaluating my voltage reference among with my divider at 14~15 Celsius as Max indoor. 

Spesific LT1021 reference chip this is my property and use since 8-2010 at the 8-2020 this will have it ninth birthdays.
In the past eight years I do nothing other than confirming over and over ppm drift due environmental temperature change of my LT1021, and this is 100% according Linear-Technology brand specs.

In my eyes the specific LT1021 chip this worth today 100 times higher of it retail value as new/unused.
This is now fully aged, fully stable, fully predictable ppm/C drift, with a measured Standard Deviation that is second best in my lab.
The most lowest Standard Deviation this were measured at totally analogue portable calibrator, but voltage output repeatability this can not be warranted as there is a ten turns potentiometer controlling the output.
From the other side they are the digitally controlled portable calibrators, the one which I tested has significantly higher Standard Deviation.
In conclusion the autonomous controlled LT1021 chip this is a proven victory in the regard of low Standard Deviation among with fully predictable ppm/Celsius drift.

Speaking of voltage divider resistors, these are Vishay metal film at 0.1%, they are also eight years old, and aged too.
It is interesting that these Vishay resistors seems to not effecting at all the delivered precision due their own not that extremely tight tolerance.   
Entire project similes to an cocktail that any Volt-Nut will drink and fully enjoy.