Home calibration/sanity check

[Blastcap]

So i got three meters one crappy (8€ from the home depot) a little less crappy (rebrand UNI-T, into Voltcraft VC175), and a not so crappy (Fluke 17B+).

It appears that the Fluke is a little off, but still just within spec (measures 3,497K  on a 3,5K resistance) and the UNI-T/Voltcraft seems to be more precise measuring 3,500K as hoped, the last meter is just way off, no use, so lets never speak about that one again.
The resistance is a 0,1% 6 digit decade resistance from this thread. Set to 3.500Ohm (+0,5Ohm contact resistance).

Well, now i found on the local flea-bay a Fluke 8050A for just 40€, if it works that would be a steal, so I'm considering grabbing it, and checking it against my decade resistance, and if its fine ill calibrate the 17B+ against it, its not very high res. but 4 1/2 digit (20K counts) should be enough to get the 17B+ (4K counts) better than before.

I have access to an Agilent 5 1/2 Digit meter, but i cannot bring my own meter to it, but i can build small test jigs and check them against the calibrated 5 1/2 and bring the jigs home and compare to my own meters. On the way i would probably loose a digit or so in precision from adding the tolerances and loss over time/distance/temperature.

I'm planning to build a battery powered 3V voltage "reference" with a 0,01% TI REF3230, check it against the 5 1/2 and bring it home without turning it off, that should allow me to calibrate the 17B+ with reasonable precision.

So is this doable or have i overlooked anything minor/major that screws up my master plan?

[MosherIV]

i found on the local flea-bay a Fluke 8050A for just 40€, if it works that would be a steal, so I'm considering grabbing it

I would go for it any way, you can never have too many meters 

It appears that the Fluke is a little off, but still just within spec

Then I would not touch the Fluke17B+, but that is me

I have access to an Agilent 5 1/2 Digit meter ...  i can build small test jigs and check them against the calibrated 5 1/2 and bring the jigs home and compare to my own meters. On the way i would probably loose a digit or so in precision from adding the tolerances and loss over time/distance/temperature.

The important thing to note is the temperature and humidity when checking your jig.
You may also want to leave the jig running for 1 or more hours before checking on the Agilent to allow warm up settling.

So is this doable or have i overlooked anything minor/major that screws up my master plan?

Normally, you would need a complete system which provides Volt, Amps and Ohms to do a full meter calibration.
They would normally calibrate to 2 or 3 reference points, one near the bottom of the range and one near the top and one in the middle to check full scale and linearity.
Again, leave your jig running for at least 1 hour before using it to calibrate and try to match the conditions with the Agilent.

Many, including myself, have gone for the Analogue Devices AD584 which provides 2.5V, 5V and 10V and you can get 7.5V by summing the 2.5V and 5V
I have built mine without trimming, ie just took the raw output, which you can then check it against the high precision Agilent.
I would recommend aging any Vref so it settles to a known voltage.
Personally I would only use it as a check and not use it to calibrate against.
I am also adding a 1.235V ref to my set up, with a LM385 ( to check my UT61E - in hind sight maybe I should have gone for a 2V ref instead )

[Blastcap]

I would go for it any way, you can never have too many meters 

I guess it would do nice on the bench.

Then I would not touch the Fluke17B+, but that is me

It just nags me knowing my everyday meter isn't that spot on as the other.

The important thing to note is the temperature and humidity when checking your jig.
You may also want to leave the jig running for 1 or more hours before checking on the Agilent to allow warm up settling.

That's why they would be battery powered, to keep them running 2-3 hours before checking against the Agilent and then keep it running until i gets home to check against my own gear. Housed in a housing with high thermal capacity and sealed as good as i can, to slow any change to the environment.

Many, including myself, have gone for the Analogue Devices AD584 which provides 2.5V, 5V and 10V and you can get 7.5V by summing the 2.5V and 5V
I have built mine without trimming, ie just took the raw output, which you can then check it against the high precision Agilent.
I would recommend aging any Vref so it settles to a known voltage.
Personally I would only use it as a check and not use it to calibrate against.

The Vref was chosen to 3V as the 17B takes a good 3V to calibrate the DC range witch translates over to the ohm range, so the usual AD584 wouldn't really work in this instance. But for a second Jig to check against afterwards very useful.

Edit:
I guess a AD780 would be better, as it can be trimmed.
... There more i think, the more i research... the deeper the rabbit hole goes. 

[MosherIV]

I guess a AD780 would be better, as it can be trimmed.

Trimming is only worth while IF you can guarantee that it does not drift.

The guys over in the metrology forum have lots of information about the right components to use and what circuits to use.

It is best to use the highest quality components you can afford. Get 100ppm/°C or even 50ppm/°C pots or better if you can. 0.1% or even 0.01% high precision resistors.
I would recommend using the lowest value trim pot eg 100\$\Omega\$ trimmer + 9.76K 0.1% precision resistor to get the lowest drift, both time and temp co.

The goal for any reference is 0 drift over time and minimal drift with temp (but ideally all the equipment should be kept at steady temp).

[Blastcap]

I looked in the metrology sub, but i think that is taking it too far, I'm looking for stability over hours not decades, (as cool as it would be) just trying to get a 100€ meter back on track (or close to) for less than 20€.

So getting a decent Vref, keep it running for 12 hours, and then trimming it to 3,000 +/- 0,0005V at work, take it home without disturbing the power supply, temp or any other things (too much) and calibrate my meter to it. Let it run till the next morning, take it to work and check that it its still around the set voltage 24 hours after turning it on.

And as Mark 9:31 said: " ...And when he is killed, after three days he will rise... "
Probably talking about the voltage of my homebrew Vref, and I'm fine with that, because i don't need a stable voltage anymore.

Edit:
Cal. Manual is here:
http://assets.fluke.com/manuals/15BPlus_cmeng0000.pdf
Pg. 16 and 17.

Pretty simple:
Apply 3V AC/DC, 3A, 30mA, 100uF or 0°c
Press the non-populated cal button.
...*beeeeep!*
Press hold.
Repeat.

[MosherIV]

So getting a decent Vref, keep it running for 12 hours, and then trimming it to 3,000 +/- 0,0005V at work, take it home without disturbing the power supply, temp or any other things (too much) and calibrate my meter to it. Let it run till the next morning, take it to work and check that it its still around the set voltage 24 hours after turning it on.

Temperature will be critical, you will be surprised what 3°C can do!

Also, if you have the time try to age the Vref device for a week or 2. Probably less critical for the ones you mentioned but some do have a settling period, eg the super low temp co LM399 has shockingly bad initial tolerance and takes around 3 months to settle (according to some reports in the metrology section). I have had my AD584 going continuously now for 3 months. It has drifted down by 2mV.
Edit : sorry the drift is only 200uV

I also have been offered to check my Vref at work 
I am trying to work out which is wrong, my 5.5 digit meter or my 6.5 digit meter 
My UT61E (4.5digit) agrees with the 5.5 digit upto its limits.
Getting a Fluke187 (4.5 digit) today - well hopefully - cannot wait.

[Blastcap]

Well, got my paws on that 8050A. Internally it looks great, and just 11 years past calibration, but till seems to be spot on. 
Now i just need to clean it up externally, and perhaps give it the LED mod.

[MosherIV]

Nice 

Keep us informed on how you get on with the Vref and calibration.

[Blastcap]

Well I've been thinking a little.

A 9V block into a 1117 Vreg to feed the Vref a stable 5-ish volt.
And a switch to tun the contraption on and off, and a button to check if the 5V is there.

Output is on binding posts, with ground in the center, 3V left and 5V-ish on the right.

The Vreg is roughly 20mm away from the Vref.





I'd hook it up to a small switchmode-powersupply from a 12v powerbrick, and do a few days burn in, before giving it a battery and put it in an enclosure.