real-world voltage reference?

[Kalvin]

Ikea #502.405.02
1.5V Alkaline LR6 AA-AM3-MN1500
Expiry Date: 2019-01-13
Made In China
18862
Unused, batteries measured in same order as in package, measured twice, rounded to four digits
1.594, 1.596, 1.594, 1.596, 1.598, 1.595, 1.585, 1.586, 1.587, 1.586

Ikea #802.405.05
1.5V Alkaline LR03 AAA-AM4-MN2400
Expiry Date: 2018-12-11
Made In China
18862
Unused, batteries measured in same order as in package, measured twice, rounded to four digits
1.591, 1.592, 1.590, 1.592, 1.590, 1.592, 1.589, 1.592, 1.592, 1.591

Room temperature: +24 Celcius
Multimeter: UNI-T UT61E

Comparing the UNI-T UT61E to the UT33C in 2000mV range and Vichy VC99:
UT61E: 1.586V
UT33C: 1.583V
VC99:   1.586V

[ralphd]

Picked up a 3-pack of Sunbeam CR2032s at a local dollar store today.  Made in China, best before 03/2018.  Measured at 22C:
3.357, 3.353 & 3.342V
Subtracting 0.4% (my guess for my meter offset until I have a good reference) gives:
3.344, 3.340, & 3.329V

[Macbeth]

Consider sticking a button cell in your navel, under your armpits, under your tongue or up your A$$, and use it always at rest and at the same time of day. Natural oven controlled cell !

You are a fecking genius! The holy grail - an aged and selected LTZ1000 shoved up the arse for proper tempco. The voltnuts will surely go for it? Of course it will need doctors certificates for traceability instead of the usual NIST paper trail, but I think its a candidate 

[ralphd]

Ikea #802.405.05
1.5V Alkaline LR03 AAA-AM4-MN2400
Expiry Date: 2018-12-11
Made In China
18862
Unused, batteries measured in same order as in package, measured twice, rounded to four digits
1.591, 1.592, 1.590, 1.592, 1.590, 1.592, 1.589, 1.592, 1.592, 1.591

I was at Ikea a couple weeks ago and had considered picking up a pack.  I'm guessing they're either a lot older than the Wal-Mart AA's I have, or the chemistry is slightly different since the voltages are about 2% lower even after making a correction for the .35-.45% offset of my meter.

[IanB]

Picked up a 3-pack of Sunbeam CR2032s at a local dollar store today.  Made in China, best before 03/2018.

As did I (two-pack though). Same date. 3.347 V, 3.377 V.

The lack of consistency implies poor quality control. Oh well, who can complain at 50¢ each?

[ralphd]

Picked up a 3-pack of Sunbeam CR2032s at a local dollar store today.  Made in China, best before 03/2018.

As did I (two-pack though). Same date. 3.347 V, 3.377 V.

The lack of consistency implies poor quality control. Oh well, who can complain at 50¢ each?

No typo there?  3.377 on the 2nd one?
40-50c ea is OK for something you can grab while you're out picking up milk, but I order most of my CR2032s online.  I've found the ones that come in the trays are not worth the trouble (many arrive dead), but the ones in the 5-packs on a cardboard strip are good, and can be found for <$2 for 10 (2 packs of 5).

I find the cheap Chinese CR2032s don't put out much more than 10mA, which can be a problem for little sensor nodes.

[deadlylover]

I checked a few old unused batteries lying around @ 17C, quite chilly in Sydney today...
Using an R6581 8.5 digit meter, accurate to about 7ppm, slightly overkill I think. =P

Varta 9V alkaline, expiry 12-2020: 9.7649V

Sony AAA heavy duty, unknown date/expiry : 1.5975V

Titanium Innovations CR123A, purchased mid-2011 : 3.2750V, 3.2779V

Probably not very useful data for anyone, but I couldn't resist joining in on the fun.

[Gyro]

Hmmm, It looks as if CR2032s from the same pack have surprisingly large variations. Must be either lousy quality control or sensitive to very small variations in the recipe.

The Alkalines seem much more consistent in the same pack, but still fairly large variations between kitchens (see what I did there?  ).

Still quite surprised about the Lithiums though, maybe initial voltage stability is inversely proportional to energy density? Older chemistry may be more robust against slight variations in the mix.

Come on folks, there must be some electro-mechanical types out there with SR44s - I'm trying to make a point here! 

[ralphd]

Come on folks, there must be some electro-mechanical types out there with SR44s - I'm trying to make a point here! 

Well, to make you happy, I went out and found a Duracel 301/SR43 silver oxide mercury-free battery.  $3 (more than I paid for 100 TL431As) only gets you one of them at the local Wal-Mart!
Anyway, my meter fluctuates between 1.597 and 1.598, so call it 1.5975.  Temperature: 21C.
That puts my reading 2% over yours.  I'm pretty confident my meter is not 1% high (more like 0.3-0.5%), so already with a sample of 2 silver-oxide batteries are not good enough for a 1% reference.

[Gyro]

Doh! That's not what I was expecting 

Mine is probably a year or so old now (waiting for one of the others to run out). Maybe they're only stable when loaded 

Thanks for your dedication to the cause of household metrology anyway 

[lewis]

Please, can everyone stop taking about using batteries as voltage references! No standard household battery is going to be accurate enough,  stable enough or precise enough to calibrate a DMM.

Ralph dude - if you want to calibrate a DMM you're just going to have to bite the bullet and spend the few Canadian Pesos on a decent enough semiconductor reference. Or buy a new meter. You mentioned in a previous post that hobbyists won's spend $10 on a reference. But you've got a several hundred dollar Rigol (on your website). A few bucks on a ref would more than offset the time you've spent trying various different batteries.

[ralphd]

Please, can everyone stop taking about using batteries as voltage references! No standard household battery is going to be accurate enough,  stable enough or precise enough to calibrate a DMM.

Ralph dude - if you want to calibrate a DMM you're just going to have to bite the bullet and spend the few Canadian Pesos on a decent enough semiconductor reference. Or buy a new meter. You mentioned in a previous post that hobbyists won's spend $10 on a reference. But you've got a several hundred dollar Rigol (on your website). A few bucks on a ref would more than offset the time you've spent trying various different batteries.

I ordered 100 TL431As, and a couple REF5050s, but it will be a few weeks before they arrive from China.  In the mean time I was looking for something around the house I could use as a rough reference.

Yes, I treated myself to a DS1054Z after >30yrs of tinkering with electronics, and I could only do that after divorcing the first wife. :-)
For those with a $10 meter and a couple of $2 Arduino nanos, it would still be useful to have an easy way to tell if their meter is modestly accurate.

[retrolefty]

Yes, I treated myself to a DS1054Z after >30yrs of tinkering with electronics, and I could only do that after divorcing the first wife. :-)

 Oh you mean the 'starter wife' 

[ralphd]

Yes, I treated myself to a DS1054Z after >30yrs of tinkering with electronics, and I could only do that after divorcing the first wife. :-)

 Oh you mean the 'starter wife' 

I don't know, she was pretty hard to get started. :-(

[ralphd]

Just watched Dave's latest video, and the Duracell AAA started off at 1.602V.

[Nerull]

My point was to see if there is a common household item like a battery that can be used as a voltage reference.
It would be useful to hobbyists that aren't going to spend $10 on a precision reference.
Anyway, I'll probably end up getting a max6126 just to satisfy my curiosity.  And maybe I'll find a common household item that others can use for a 1% reference...

Your meters out of the box specs are orders of magnitude more accurate than any of the sources you're looking at, it is almost impossible that you will adjust the meter to be more accurate using them, and almost certain that you will make it worse.

Pretty extensive experience by a lot of people shows that multimeters really don't drift all that much, at least not at the number of digits you're looking at, and even free meters from hobby freight are spot on for their level of precision out of the box. Meter calibration for hobby use is mostly a waste of time, and if you do want to calibrate something, you need something more accurate to check against, not less.

I've taken my never-recaled 87V and checked against freshly calibrated meters at work, no significant difference. When I got my 87V I wanted to see how bad the cheapass no-name walmart meter I had been using before compared - also no difference (though it had fewer digits).

[Lightages]

If the concern is to get a voltage reference that is cheap:
http://www.ebay.com/itm/AD584KH-4-Channel-2-5v-7-5v-5v-10v-High-Precision-Voltage-Reference-Module-/181112567159?hash=item2a2b267577
or even cheaper:
http://www.ebay.com/itm/AD584-4-Channel-2-5v-7-5v-5v-10v-High-Precision-Voltage-Reference-Module-/261868849256?hash=item3cf8998868

[lapm]

Personally i would not use batteries as reference voltage source. They vary way too much by age and temperature... Heck even those 30$ Chinese cheap reference kits are better..

[ralphd]

If the concern is to get a voltage reference that is cheap:
http://www.ebay.com/itm/AD584KH-4-Channel-2-5v-7-5v-5v-10v-High-Precision-Voltage-Reference-Module-/181112567159?hash=item2a2b267577
or even cheaper:
http://www.ebay.com/itm/AD584-4-Channel-2-5v-7-5v-5v-10v-High-Precision-Voltage-Reference-Module-/261868849256?hash=item3cf8998868

I already mentioned I ordered a bunch of Wing Shing TL431As, and a couple of TI REF5050s.
http://www.aliexpress.com/item/REF5050IDGKR-R50F-TI-VSSOP-8-absolutely-new-original-quality-assurance-FREE-Shipping/32327854830.html

[ralphd]

The  Wing Shing TL431As arrived this week.  All labeled:
WS
TL431A
155SD

I tested 11 of them at 24C, 9.4mA:
2.510, 2.508, 2.511, 2.511, 2.513, 2.517, 2.509, 2.516, 2.514, 2.509, 2.515V
mean = 2.5121, SD: .00308, <0.2% variation

The 2.512 measured vs 2.495 spec voltage suggests my meter is high by 0.68%.

p.s. I noticed an Aliexpress seller (but not the one I bought from) that had a photo of a box of (10,000?) Wing Shing TL431As, and it had a 0.3% label on the box.  I suspect the TL431As I received are the same.

[jwm_]

2.512v is perfectly within spec for the tl431a. All your measurements are consistent with your meter being spot on spec.

You _cannot_ average voltages like that and assume it ends up at the canonical value. There is no expectation that the voltages are normally distributed and it is in fact unlikely for many types of trimming. A hundred 2.520v tl431a's are still legit and in spec. And we're you to repeat the test  on a cold day, they all might be 2.470v and you would conclude the opposite about your meter.

It is extremely unlikely your meter is out of spec, almost unheard of. And if improving it is a matter of just trimming it, it would be sold with better specs, because your meter has a tempco and aging parts too.

But the main issue is assuming averaging gives more accuracy, when in this case, and the battery case, it does not.

[ralphd]

For anyone that might be inclined to believe unsupported assertions that the mathematical laws of statistical sampling don't apply to semiconductors, read this:
http://www.maximintegrated.com/en/app-notes/index.mvp/id/4489

[ralphd]

2.512v is perfectly within spec for the tl431a. All your measurements are consistent with your meter being spot on spec.

Did you bother to calculate 2.495 + .3%?  It's 2.5025.

The .05% REF5050s I ordered should arrive within the next few days, and I'll bet they'll show my meter is at least 0.5% high.
In the mean time, I suggest you start searching for a palatable hat.

[ralphd]

And we're you to repeat the test  on a cold day, they all might be 2.470v and you would conclude the opposite about your meter.

Just to prove you're full of shit, I stuck my test circuit in the freezer for a few minutes, then measured the reference voltage while it was still in the freezer: 2.511V.
After warming back up to 23C I'm reading 2.515.

[jwm_]

2.512v is perfectly within spec for the tl431a. All your measurements are consistent with your meter being spot on spec.

Did you bother to calculate 2.495 + .3%?  It's 2.5025.

The .05% REF5050s I ordered should arrive within the next few days, and I'll bet they'll show my meter is at least 0.5% high.
In the mean time, I suggest you start searching for a palatable hat.

I was going by the datasheet.

https://www.fairchildsemi.com/datasheets/TL/TL431.pdf

tl431a
min voltage  2.470
max voltage 2.520
with a 17mv drift over full temperature range.

I was just using the cold day as a way to say that there are systematic reasons they may all appear off spec in the same direction. 17mv difference over temperature range (again, from the data sheet) is enough to make a meter seem high or low, especially if the meter is at a different temperature.

I am not sure what you are arguing, your data is consistent with an in spec meter.