What's new
What's new

Inexpensive Millivolt (around 20mV?) Reference Sought

dgfoster

Diamond
Joined
Jun 14, 2008
Location
Bellingham, WA
Hey, electronics gurus, I am working with thermocouples in my cast iron foundry and would like to come up with a reference millivolt source so that I could calibrate and confirm calibration ongoing of a device that would read the thermocouple output to four significant figures. The voltages in question would range between 0.000 and 20.00 mV with the area of greatest interest being around 15.00 mV.

I have been trying to locate a reference voltage source in the 10 to 200 mV range that I could use to check my voltage measuring system to ensure that the gain in my system is correctly adjusted. So far, I have found precision voltage generators in the 200 to 500 dollar range. But, I suspect there is something out there much cheaper.

Any help?

Denis
 
200 dollars is pocket change for anything "precision" and "calibration"

0.05% claimed accuracy thermocouple calibrators (Chinese ones course at this price) seem to start at around 150 usd price.

Or you could fabricobble your own for 20 usd based on voltage reference and resistive voltage divider but in that case you get either fixed steps or another meter to indicate the output. Better buy something ready made if you don't have reasonable accurate DVM at use.

10v precision voltage reference and bunch of 0.05% resistor would give you x number of fixed voltage steps.
ie:

10v reference+
100kohm
10r
10r
10r
10r
10r
100r
ground

would give you steps from 10 to 15mV in 1mV steps.
 
And old HP 3478A DVM's have resolution down to 1uV and these can be had from ebay for 50 to 150usd per piece.
Super crude but workable hack would be to use one to measure the output of 10k fixed resistor and 100ohm 10-turn potentiometer. Use Nimh or nicad battery for reference voltage.
 
Better put another bigger resistor in series with the multiturn pot. 100k pot alone would have the adjustment range withing first 1/10th of turn.

that helps...the 2025 has 2.5 V[SUB]o[/SUB] you coud use a 1:100 divider first. (Votage Divider Formula: Vout = Vs*R2/(R1+R2)) a 99k fixed with a 1 kohm 10 turn pot will give you 2.5 mv per turn.


dee
;-D
 
that helps...the 2025 has 2.5 V[SUB]o[/SUB] you coud use a 1:100 divider first. (Votage Divider Formula: Vout = Vs*R2/(R1+R2)) a 99k fixed with a 1 kohm 10 turn pot will give you 2.5 mv per turn.


dee
;-D

The thing is, I do not want variable voltage but rather one fixed and reliably accurate voltage to use as a standard much as one would use a 1" standard to set/check a micrometer.

Denis
 
Probably the cheapest option would be to add offset to a stable precision op-amp.

Here is one way although this only goes to +/- 3.5 mV.

OPAMP_AS_MILLIVOLT_REFERENCE_1 - Power_Supply_Circuit - Circuit Diagram - SeekIC.com

Standard methods of introducing an offset at the inputs should get you to your 20 mV range. As long as you are willing to check the output with a good meter at each use an op-amp based circuit should work just fine.

Here is a good reference on designing gain and offset.

http://www.ti.com/lit/an/sloa097/sloa097.pdf

OR

Use the method in the first link with a second op-amp stage set for a gain of 10. That would give you a -35 mV to + 35 mV range and require no voltage reference.
 
The thing is, I do not want variable voltage but rather one fixed and reliably accurate voltage to use as a standard much as one would use a 1" standard to set/check a micrometer.

Denis

Fixed voltage problem was solved already earlier.
REF102CP 0.025% voltage reference and two 0.01% precision resistors from digikey are about 30 dollars total and will give you better than 0.05% accuracy even if you don't have any measurement equipment.
 
Not exactly what you're looking for, but Scientific American's Amateur Scientist had plans for a triple point cell. That would let you calibrate the entire system.

Tackling the Triple Point - Scientific American

Cool but absolutely irrelevant in this case :)
Besides, melting point of crushed ice is more than enough accurate for anything homebrew. 0.01c accuracy easily and within 0.003c with some care.
But also very much useless for high temperature thermocouple calibrations.
 
I built a kiln temperature probe for my ceramic kiln and used a thermocouple to drive a meter directly. I had no way of calibrating it accurately so I used the melting point of various metals. One was lead I remember and one was copper. I also used zinc since I had some sacrificial anodes. Not exactly precise but it was close enough for me to set the gain and range on the meter.
 
The thing is, I do not want variable voltage but rather one fixed and reliably accurate voltage to use as a standard much as one would use a 1" standard to set/check a micrometer.

Denis

omit the pot use a fixed divider. The part referenced is cheap and very stable. Mattj suggested REF 102CPt with 10V out, which has through hole legs. any of them can give you a fixed reference with dividers. If you need low impedance just put in a buffer. something like http://www.analog.com/media/en/technical-documentation/data-sheets/AMP03.pdf

dee
;-D
 
I built a kiln temperature probe for my ceramic kiln and used a thermocouple to drive a meter directly. I had no way of calibrating it accurately so I used the melting point of various metals. One was lead I remember and one was copper. I also used zinc since I had some sacrificial anodes. Not exactly precise but it was close enough for me to set the gain and range on the meter.

Melting/freezing point of metals is actually the most accurate temperature reference and basis of todays thermometry.
Electric copper wire would be probably enough pure to provide reasonably good accuracy at 1084c and 99.995% pure aluminium(660c) is also pretty cheap.
Primary standard grade 99.9999% pure metals on the other hand are NOT cheap.
 
There are all kinds of methods to get there in terms of hardware, for instance a precision voltage source with a precision voltage divider, but none of them will give you an accurate voltage without a traceable calibration. Anything you buy will probably cost you $300-500 just to have calibrated. Any decent calibration is going to be traceable to a NIST standard. Accuracy isn't cheap.
 
I would recommend you look at a Reference Cell. I have a couple I use to check my volt meters from time to time. They have a voltage of around 1volt and the voltage is marked on the unit out to at least 4 decimal places. I don't see why you couldn't use a voltage divider to get the voltage down to 15-20mV. I would think you would want to use percision resistors of a very high resistance as the cells produce very little current. A quick google search comes back with a "Weston Cell". Not sure if this is exactly what I have but it is along the same line.
 
There are all kinds of methods to get there in terms of hardware said:
Pretty good likelihood if you order 0.01% resistors and 0.02% voltage ref that your bodge is going to be accurate withing 0.05% even if you don't calibrate or check it.

And 50 to 200 usd would be probably more realistic range (even for one-offisie customer) for basic calibration price if needed accuracy is only four digits.
HP/Agilent 34401A 6½ digit DVM calibration is about 150 usd and it has shit ton of measurements and ranges compared to fixed-voltage thermocouple simulator.
 
I would recommend you look at a Reference Cell. I have a couple I use to check my volt meters from time to time. They have a voltage of around 1volt and the voltage is marked on the unit out to at least 4 decimal places. I don't see why you couldn't use a voltage divider to get the voltage down to 15-20mV. I would think you would want to use percision resistors of a very high resistance as the cells produce very little current. A quick google search comes back with a "Weston Cell". Not sure if this is exactly what I have but it is along the same line.

Reference cells were valid technology something like 40 or 50 years ago. IC voltage references are the way to go.

OK, seem like I get carried away since half of replies on the thread are mine... ½ EE who has been working in a (temperature) calibration lab for last 10 years
 
Do it correctly and buy a certified and maintainable reference.

These types of things when drift can cause much damage in many ways and if discovered you made one and that it has no way to certify you are toast.

Less than $1000.00 would be easy dudget.

Sent from my SAMSUNG-SGH-I337Z using Tapatalk
 








 
Back
Top