Simple but stable 5v voltage reference for ADC calibration?

[max_torque]

I have a device that has a load of ADC inputs (16 of) and i want to use ROM tables to calibrate each input channel with a reasonable level of accuracy.   These inputs are 0-5Vdc and 100KOhm impedance.

Is there a better way that a 9v battery, some sort of stable (ish) voltage reference, a couple of 10 turn pots and a unity gain opamp to generate a reference output?   Output needs to be stable over a few min, but not have long term accuracy or stability as i will be using a 6.5 digit bench meter as a "master". 


ADC's are 12bit only, but might be oversampled in a few cases up to around 14b

What have others done to achieve this sort of reference?

[KJDS]

The lazy way is something like this, other suppliers are available and I have no idea of the quality of that or other sellers of these, but the base product is decent enough.

http://www.ebay.co.uk/itm/High-Precision-4-Channel-2-5V-7-5V-5V-10V-Voltage-Reference-Module-AD584LH-/251617395537?pt=UK_BOI_Electrical_Components_Supplies_ET&hash=item3a9590c351

[ManateeMafia]

How about the PentaRef?

http://www.voltagestandard.com/PentaRef.html

I have one of these and it is a good product. It is not cheap, the web site lists it at $78.00. It depends on if you prefer a pre-built solution.

[Flenser]

How is this for simple?

"A" grade version of theLM4040 costs £1.34 or £1.46 in single quantities depending on the package.
http://uk.farnell.com/webapp/wcs/stores/servlet/Search?catalogId=15001&langId=44&storeId=10151&sort=P_PRICE&st=lm4040&pageSize=25&showResults=true&pf=110003511,110033312,111439590,111439591

Adafruit sell this breakout which has both the 2.048 and 4.096 versions of the chip
http://www.adafruit.com/product/2200

[dannyf]

ADC's are 12bit only, but might be oversampled in a few cases up to around 14b

If a pot is involved in generating a reference, there is no need for 12-bit resolution, or calibration.

A TL431 (or a few if you insist on getting to 5v) would be more than sufficient.

[max_torque]

TBH, i don't care about the absolute accuracy of the voltage source, as i will be using my 6.5 digiit bench meter as the calibration reference.  As such, i just want a voltage source, that is short term stable and adjustable to a resolution of around 0.5mV output.

It needs to have a reasonably low output impedance (say <1kohm (ish) so it can drive the ADC input circuitry on various different kinds of test pcbs etc

If i used two decent quality  pots say one 10Kohm single turn, and a 500ohm 10 turn one, that would look to work.  In effect, the actual pot setting to achieve the outpout voltage is un-important, but just needs to be noise free (can easily have a low pass filter on it as i have no response time limitations for the circuit etc) and stable over say 5min (after a warmup period) etc

[paulie]

In reality these "precision" voltage references are little more than a couple zeners back to back. In any case the simplest circuit to provide low impedance might be an LM4040 with low cost op amp offset adjusted with fixed resistors and a pot. A coarse adjust pot, even 10 turn, might hinder stability. I built something similar many years ago and every time recalibration at the lab showed almost no drift.

[f5r5e5d]

bandgap refs are most common, a few other techs available depending on V, stability, noise requirements

plenty of manu selection guides, app notes

[Andreas]

The design should follow your needs.
So how much temperature change do you have?
for a 0.01% stability over 10 deg C temp change the tempco of the reference should be below 10 ppm/deg.

So any trimpot at the output of a voltage (with 300-500 ppm/deg) will be insufficient.

So I would choose either a precision bandgap (LT1019, Ref02 w. 10ppm/K)
or a zener based reference (MAX6350, LT1027, AD586 w. 1-5ppm/K).
And use the trimming scheme within the datasheet which is much less sensitive to the tempco of the trimpot.

With best regards

Andreas