Author Topic: real-world voltage reference?  (Read 23490 times)

0 Members and 1 Guest are viewing this topic.

Offline ralphd

  • Frequent Contributor
  • **
  • Posts: 442
  • Country: ca
    • Nerd Ralph
real-world voltage reference?
« on: July 25, 2015, 11:58:20 AM »
Is anyone aware of a non-semiconductor voltage reference source with ~1% or better accuracy?
For humidity, a saturated NaCl paste is a decent 75% RH reference @25C.
For temperature an ice bath is a decent 0C reference.
One idea I consider is making a simple lead-acid cell.  The output varies not only with temperature but with the sulfuric acid concentration.  The temperature I can easily measure, but without a hygrometer I can't measure the sulfuric acid concentration.  And even if I could, I doubt I could get 1% accuracy when temperature and acid concentration are factored.

Any ideas?  I'm going to order some TL431A's (+- 0.8%), but while I wait for them to arrive I'm wondering if there's something I might already have around the house I could use.
Unthinking respect for authority is the greatest enemy of truth. Einstein
 

Offline DanielS

  • Frequent Contributor
  • **
  • Posts: 798
Re: real-world voltage reference?
« Reply #1 on: July 25, 2015, 12:36:22 PM »
If you apply a reasonably constant forward current to some diodes and keep them in a controlled temperature environment, you get a relatively constant forward voltage drop.

Of course, forward voltage being so wildly different from one device to the next, you would need to calibrate your circuit on a per-diode basis. Not ideal but still far more accurate and convenient than a lead acid battery.
 

Online Ian.M

  • Super Contributor
  • ***
  • Posts: 5750
Re: real-world voltage reference?
« Reply #2 on: July 25, 2015, 12:37:20 PM »
Heathkit used to reckon a fresh zinc carbon cell was 1.55V.

See http://www.ohio.edu/people/postr/bapix/vtvms.htm and search the page for calibration.for

Its probably not 1%, but is certainly better than anything else 'domestic'.
 

Offline radioFlash

  • Regular Contributor
  • *
  • Posts: 150
  • Country: us
Re: real-world voltage reference?
« Reply #3 on: July 25, 2015, 12:48:03 PM »
Are you familiar with a Weston Cell?


Sent from my iPhone using Tapatalk
 

Offline ralphd

  • Frequent Contributor
  • **
  • Posts: 442
  • Country: ca
    • Nerd Ralph
Re: real-world voltage reference?
« Reply #4 on: July 25, 2015, 01:20:21 PM »
Are you familiar with a Weston Cell?
I wasn't.  After reading about it, I won't be trying it since I don't have a local source for the cadmium and mercury.
Unthinking respect for authority is the greatest enemy of truth. Einstein
 

Offline FrankT

  • Regular Contributor
  • *
  • Posts: 132
  • Country: au
Re: real-world voltage reference?
« Reply #5 on: July 25, 2015, 01:24:24 PM »
On a slight tangent, I needed to calibrate/verify a sound level meter.  Anyone know of a way of doing that without an expensive source?
 

Offline ralphd

  • Frequent Contributor
  • **
  • Posts: 442
  • Country: ca
    • Nerd Ralph
Re: real-world voltage reference?
« Reply #6 on: July 25, 2015, 01:32:17 PM »
Heathkit used to reckon a fresh zinc carbon cell was 1.55V.

See http://www.ohio.edu/people/postr/bapix/vtvms.htm and search the page for calibration.for

Its probably not 1%, but is certainly better than anything else 'domestic'.
No zinc carbons around, but I checked a pack of Wal-mart AA alkalines.  My meter read 1.636 for 4 of them, and the fifth read 1.635 at 23C.  That would suggest my meter is a touch high...
Unthinking respect for authority is the greatest enemy of truth. Einstein
 

Offline ralphd

  • Frequent Contributor
  • **
  • Posts: 442
  • Country: ca
    • Nerd Ralph
Re: real-world voltage reference?
« Reply #7 on: July 25, 2015, 03:22:16 PM »
If you apply a reasonably constant forward current to some diodes and keep them in a controlled temperature environment, you get a relatively constant forward voltage drop.

Of course, forward voltage being so wildly different from one device to the next, you would need to calibrate your circuit on a per-diode basis. Not ideal but still far more accurate and convenient than a lead acid battery.
I don't have a known calibration source, so that doesn't really help.
When I get the tl431s, I'll test a dozen or so, and the one closest to the mean should give me a .02% accuracy reference.  With that I can calibrate my 4-digit multimeter.  I'll also have a bunch of tl431's to use on various projects (for only 2c ea on Aliexpress)
Unthinking respect for authority is the greatest enemy of truth. Einstein
 

Online Ian.M

  • Super Contributor
  • ***
  • Posts: 5750
Re: real-world voltage reference?
« Reply #8 on: July 25, 2015, 04:07:59 PM »
1.6V is about right for a fresh alkaline. 
 

Offline Gyro

  • Super Contributor
  • ***
  • Posts: 2542
  • Country: gb
Re: real-world voltage reference?
« Reply #9 on: July 25, 2015, 08:32:24 PM »
You might be better looking at button cells...

A Silver Oxide (watch) battery has a stable 1.55V output for most of its discharge life when very lightly loaded (say, a few uA)

The older Mercury button cells had a stable 1.35V (in fact I think they were spec'd to a few more decimal places) but are unobtainable now.

To be honest you'd be better going for one of the tighter initial tolerance semiconductor reference. Some of them are not that expensive and are spec'd tighter than your TL431s for initial un-trimmed voltage. By the time you've bought the battery and bought/fabricated some kind of holder, you won't be that far out on cost. Not susceptible to accidental shorts like a battery (which can cause short to long term voltage change).

Chris

"Victor Meldrew, the Crimson Avenger!"
 

Offline fcb

  • Super Contributor
  • ***
  • Posts: 1197
  • Country: gb
Re: real-world voltage reference?
« Reply #10 on: July 25, 2015, 08:40:24 PM »
On a slight tangent, I needed to calibrate/verify a sound level meter.  Anyone know of a way of doing that without an expensive source?

Not really - if you come up with something please let us know.
 

Online Ian.M

  • Super Contributor
  • ***
  • Posts: 5750
Re: real-world voltage reference?
« Reply #11 on: July 25, 2015, 09:24:20 PM »
When I get the tl431s, I'll test a dozen or so, and the one closest to the mean should give me a .02% accuracy reference.  With that I can calibrate my 4-digit multimeter.  I'll also have a bunch of tl431's to use on various projects (for only 2c ea on Aliexpress)
TL431A is only good to +/-1% if its from a reputable manufacturer. How confident are you that the parts from Aliexpress will be genuine Texas Instruments TL431A and not reconditioned used parts or Chinese copies?  It is likely that all your parts will be from the same batch so they are likely to all have more or less the same deviation from the nominal reference voltage.  Averaging doesn't improve the accuracy unless you have a truly randomly selected sample set.
 

Offline Whales

  • Frequent Contributor
  • **
  • Posts: 512
  • Country: au
    • Halestrom
Re: real-world voltage reference?
« Reply #12 on: July 25, 2015, 10:06:31 PM »
On a slight tangent, I needed to calibrate/verify a sound level meter.  Anyone know of a way of doing that without an expensive source?

Not really - if you come up with something please let us know.

Is it possible do something clever using two identical mics at different (exactly known) distances and compare the levels from them?  I'm stabbing in the dark at whether or not the inverse cube root of sound amplitude falloff will allow you to find a unique solution to the source's loudness.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 9331
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: real-world voltage reference?
« Reply #13 on: July 25, 2015, 10:20:35 PM »
Assuming anechoic conditions, inverse square or cube gets you a hint on the size of an object, but the level is all ratios to the source intensity.

Mind that almost all easy "references" (like an electrochemical cell, or hunks of metal alloy in France) are temperature sensitive, so you implicitly have to include an extremely tight temperature control while you are performing this measurement.

Tim
Seven Transistor Labs, LLC
Electronic Design, from Concept to Layout.
Need engineering assistance? Drop me a message!
 

Online Alex Eisenhut

  • Super Contributor
  • ***
  • Posts: 1651
  • Country: ca
  • If you can buy it for 4$ on eBay, why design it?
Re: real-world voltage reference?
« Reply #14 on: July 25, 2015, 10:50:46 PM »
Is anyone aware of a non-semiconductor voltage reference source with ~1% or better accuracy?

https://en.wikipedia.org/wiki/Voltage-regulator_tube

Not sure what "Real world" is supposed to mean. Semiconductors are real?

 

Offline ralphd

  • Frequent Contributor
  • **
  • Posts: 442
  • Country: ca
    • Nerd Ralph
Re: real-world voltage reference?
« Reply #15 on: July 26, 2015, 01:23:02 AM »
You might be better looking at button cells...

A Silver Oxide (watch) battery has a stable 1.55V output for most of its discharge life when very lightly loaded (say, a few uA)

The older Mercury button cells had a stable 1.35V (in fact I think they were spec'd to a few more decimal places) but are unobtainable now.

To be honest you'd be better going for one of the tighter initial tolerance semiconductor reference. Some of them are not that expensive and are spec'd tighter than your TL431s for initial un-trimmed voltage. By the time you've bought the battery and bought/fabricated some kind of holder, you won't be that far out on cost. Not susceptible to accidental shorts like a battery (which can cause short to long term voltage change).

Tried a LR44 battery I have lying around and got 1.138V, so I guess it's not silver oxide.
I tried a new CR2032 and got 3.331.  Varta says 3.2V open circuit (though one of their graphs looks more like 3.25).
http://www.varta-microbattery.com/applications/mb_data/documents/sales_literature_varta/HANDBOOK_Primary_Lithium_Cells_en.pdf

I'm not getting the TL431s just to make a voltage reference; I want them so I don't have to expand my collection of Zeners.  I've also seen circuits for using them as amplifiers and oscillators.
Unthinking respect for authority is the greatest enemy of truth. Einstein
 

Offline ralphd

  • Frequent Contributor
  • **
  • Posts: 442
  • Country: ca
    • Nerd Ralph
Re: real-world voltage reference?
« Reply #16 on: July 26, 2015, 01:54:33 AM »
When I get the tl431s, I'll test a dozen or so, and the one closest to the mean should give me a .02% accuracy reference.  With that I can calibrate my 4-digit multimeter.  I'll also have a bunch of tl431's to use on various projects (for only 2c ea on Aliexpress)
TL431A is only good to +/-1% if its from a reputable manufacturer. How confident are you that the parts from Aliexpress will be genuine Texas Instruments TL431A and not reconditioned used parts or Chinese copies?  It is likely that all your parts will be from the same batch so they are likely to all have more or less the same deviation from the nominal reference voltage.  Averaging doesn't improve the accuracy unless you have a truly randomly selected sample set.

I'm confident what I'll be getting are new, genuine Wing Shing TL431As.
http://www.aliexpress.com/item/Free-shipping-NEW-100PCS-TL431-TL431A-TO-92-Silicon-Transistors-triode-transistor/32328977920.html

Based on beta testing of transistors (like 3904s), I've found wide variations (200-300 beta) within the same batch.  I just measured Vf for a few 1N4006's off the same tape: .508, .510, .524, .513.  That's .516V +- 1.5%.
So I doubt all the batch of TL431As will have more or less the same deviation.  Care to wager?

Unthinking respect for authority is the greatest enemy of truth. Einstein
 

Offline ralphd

  • Frequent Contributor
  • **
  • Posts: 442
  • Country: ca
    • Nerd Ralph
Re: real-world voltage reference?
« Reply #17 on: July 26, 2015, 02:00:20 AM »
1.6V is about right for a fresh alkaline.

Anyone with a 0.1% calibrated meter tried measuring them?  I did a couple of google searches and couldn't find anything.

Now, I think I did see some thorough voltage testing done on Sanyo Eneloops.  I have some AA HR6 Eneloops I could top up, let the charge settle for a day, then compare my voltage readings with published ones...
Unthinking respect for authority is the greatest enemy of truth. Einstein
 

Offline DanielS

  • Frequent Contributor
  • **
  • Posts: 798
Re: real-world voltage reference?
« Reply #18 on: July 26, 2015, 02:22:22 AM »
I'm not getting the TL431s just to make a voltage reference; I want them so I don't have to expand my collection of Zeners.  I've also seen circuits for using them as amplifiers and oscillators.
The TL431 is basically a bandgap reference and an opamp in simple three pins packages. It is a really handy little chip with countless potential uses.

If you already have a wide assortment of zeners though, you could simply use some of those for the time being. Give constant current and temperature, they won't drift much.
 

Offline Kalvin

  • Super Contributor
  • ***
  • Posts: 1473
  • Country: fi
  • Embedded SW/HW.
Re: real-world voltage reference?
« Reply #19 on: July 26, 2015, 03:09:00 AM »
Thermocouple in 0C ice water and some precision amplification?
« Last Edit: July 26, 2015, 03:12:47 AM by Kalvin »
 

Offline saturation

  • Super Contributor
  • ***
  • Posts: 4768
  • Country: us
  • Doveryai, no proveryai
    • NIST
Re: real-world voltage reference?
« Reply #20 on: July 26, 2015, 03:12:46 AM »
To get a reference voltage from natural items requires making some form of a battery.  To get a predictable output from a battery requires reagent purity then the calculated reaction is more predictable.  Alas, any commercial grade ingredients have undefined concentrations or impurities, so its hard to say if you'll get within 1% or so of any defined voltage. 



For about 1V, a lemon, penny or orange battery.

https://en.wikipedia.org/wiki/Lemon_battery

https://en.wikipedia.org/wiki/Penny_battery

Best Wishes,

 Saturation
 

Offline Gyro

  • Super Contributor
  • ***
  • Posts: 2542
  • Country: gb
Re: real-world voltage reference?
« Reply #21 on: July 26, 2015, 03:18:07 AM »
Quote
Tried a LR44 battery I have lying around and got 1.138V, so I guess it's not silver oxide.

No, the LR44 is a standard Alkaline button cell (and yours is a fairly flat one). You need something starting 'SR'

Quote
Anyone with a 0.1% calibrated meter tried measuring them?  I did a couple of google searches and couldn't find anything.

Really not going to do you any good - the discharge curve is too steep. The variation between batteries from the same pack will be more than 0.1% and strongly temperature dependent too. You need to ditch the idea of an Alkaline as any sort of viable reference.

Just go and compare the discharge voltage curves for the SR44 versus LR44 in the datasheets on the web and you'll see what I mean.

Chris

"Victor Meldrew, the Crimson Avenger!"
 

Offline saturation

  • Super Contributor
  • ***
  • Posts: 4768
  • Country: us
  • Doveryai, no proveryai
    • NIST
Re: real-world voltage reference?
« Reply #22 on: July 26, 2015, 03:18:38 AM »
This is a good idea.  You can convert the thermocouples output voltage at 100C instead, but I don't know if it will be within 1%.



Thermocouple in 0C ice water and some precision amplification?
Best Wishes,

 Saturation
 

Offline Gyro

  • Super Contributor
  • ***
  • Posts: 2542
  • Country: gb
Re: real-world voltage reference?
« Reply #23 on: July 26, 2015, 03:26:01 AM »
Quote
Thermocouple in 0C ice water and some precision amplification?

A thermocouple needs two junctions - the voltage is proportional to the temperature difference between those two junctions. That's why thermocouple thermometers include cold junction compensation, usually implemented with a semiconductor temperature sensor. Keeping 2 junctions at two precisely known temperatures is only likely to compound your problems in establishing a stable reference. Possible but not very practical.
Chris

"Victor Meldrew, the Crimson Avenger!"
 

Offline Kalvin

  • Super Contributor
  • ***
  • Posts: 1473
  • Country: fi
  • Embedded SW/HW.
Re: real-world voltage reference?
« Reply #24 on: July 26, 2015, 03:49:14 AM »
Measuring the junction voltage of the two different alloys at 0C ice water and amplify that. Cannot be done? If the environmental temperature causes problems, just seal the op amp and drop it into ice water as well :)

Ps. Honestly, I have no idea how accurate this would be.
« Last Edit: July 26, 2015, 03:51:30 AM by Kalvin »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf