AD584KH-based DC Voltage reference review and test



Friend of mine wanted to have ability to calibrate his portable 3½-digit and 4½-digit DMMs, without spending too much money. So reference units, such as Pentaref are out of question, due to their cost. Also using ovenized reference circuits, such as LM399 or LTZ1000 would be major overkill for testing portable 4½-digit DMM anyway.

Choice was made for quad-output Analog Devices AD584K-based module from (Taobao). But my friend did not trust values writted down on reference, so he asked my help to check references output and calibrate it if required. My Keithley’s Model 2001 capability should be fit well enough to perform this task, so I took the challenge.


According to deciphering label sticker on package, our unit was calibrated using 8½-digit DMM HP 3458A, somewhen in 2015, with ambient temperature +26°C with next data:

Voltage output mode Labeled reading Ambient temperature Value off
2.500 VDC 2.50101 VDC +26 °C +404 ppm
5.000 VDC 5.00158 VDC +26 °C +316 ppm
7.500 VDC 7.50408 VDC +26 °C +544 ppm
10.00 VDC 10.00404 VDC +26 °C +404 ppm
  • Module input voltage : from +14 to +30 VDC
  • Input current : 10 mADC
  • Output current, max : 5 mADC sink or source

Calibration temperature setpoint at +26 °C is unusual for test gear, as most of instrumentation is calibrated for +23°C.


It was shipped in usual letter package, with good enough packing and multiple wraps of foam.

Nothing fancy at all, not even kelvin-sense for VREF output. Manufacturer even cheaped out with output film capacitor, as PCB footprint remained unpopulated.

Dual-layer PCB have very simple construction


Before we go deep into testing, let’s take a good look on used voltage reference chip datasheet, AD584KH

Here’s most important specs from datasheet:

OUTPUT VOLTAGE TOLERANCE, Maximum Error for Nominal Value Unit
10.000 V ±10 mV
7.500 V ±8 mV
5.000 V ±6 mV
2.500 V ±2.5 mV
OUTPUT VOLTAGE CHANGE, Maximum Deviation from 25°C Value Unit
5V and above outputs ±5 ppm/K
2.5V output ±10 ppm/K
NOISE (0.1 Hz to 10 Hz) Value Unit
Typical 50 µV peak-peak
30 mADC
Absolute maximum input Voltage VIN to Ground Value Unit
40 VDC

Performance test


Declared on the bag : 7.50408 V , actual measured 7.504066 V
Declared on the bag : 2.50101 VDC , actual measured 2.501040
Declared on the bag : 5.00158 VDC , actual measured 5.001555
Declared on the bag : 10.00404 VDC, actual measured 10.004057

To do test I’ll use next equipment:

Test item Instrument Related spec Calibration date
Power supply A Keithley 2400 SMU 20VDC 20mA 2015
Power supply B Pair of NiMH 9V cells 18VDC N/A
Output voltage accuracy Keithley 2001 DMM 20VDC range Feb 2014
Output voltage accuracy Keithley 2002 DMM 20VDC range 2007
Output voltage accuracy Keithley 182-M voltmeter 3 and 30V ranges 2015
Output voltage noise Keithley 182-M voltmeter 3mV range N/A
Relative stability EDC MV106 DC Standard 30ppm tempco Feb 2014

Tweaks and calibration

Would be great to shield reference module more from careless use, as well as trim output voltages to be more accurate.


This reference is fine to test 4½-digit DMM points but nothing more than this.

Author: Ilya Tsemenko
Created: Sept. 30, 2015, 4:15 p.m.
Modified: Nov. 12, 2023, 6:52 a.m.