TESTING AND CALIBRATING SVR BOARDS - MEASURING TEMPERATURE COEFFICIENTS
(be sure to hit refresh to see our latest changes!)
In the past, we have made measurements of SVR board characteristics using a Thermolyne heating oven with a thick walled chamber. However, temperature changes take some time to settle and the Thermolyne oven has no cooling features.
July, 2013 The system now uses a double loop temperature control. The "inside loop" is faster and forces the cover to follow a temperature measured by a thermistor embedded in the aluminum cover, the thermoelectric heat pump controlled by the ILX 5910B shown in the picture. The second "outside loop" is slower (longer time constant) and uses an RTD extending into the air in the interior of the chamber and read by an Agilent 34401A in the 4 wire ohms mode. A LabVIEW control program sets the set point temperature of the ILX controller to achieve a desired heat soaked air temperature in the interior of the chamber.
June, 2013 Autocycling to pre-test a SVR board (before compensation to become a SVR-T) over about 7+ hours PDF.
January 31, 2012, Present configuration: Back to our old ILX Lightwave 5910B, Gain 100, AD590 temperature sensor. Fluke 845AR High Impedance Null meter, + to SVR under test, - to Fluke 732B transfer standard, guard to 732B guard, 732B chassis ground to Agilent E3610A chassis ground (SVR power supply +15 V). Fluke 845AR set to 100 uV scale (-10 ppm to +10 ppm), output gain set to 1V for 100uV, output connected to hp 3456A (1 V scale). Agilent 34410A connected to 100 ohm RTD in aluminum box in the stack. Labview program takes data (temperature and difference voltage plotted versus time and logged to a text file for XY plotting in Excel), typically one data point per minute. A CPU cooler is coupled to the Peltier device via Arctic 5 heat sink compound (very mean stuff for eyes and sinus, might replace it later with a traditional heatsink compound) and the 12 V fan is running at 6 V on an economy import power supply (HY1803D).
Case Study of SVR board S/N 697
Some years back, we used an old ILX 5910B temperature controller to build a 1980's style micro environmental chamber with heating and cooling capability. These ILX instruments were designed to control thermoelectric heat pumps (Peltier devices) in the 1980s and apparently sold through the '90s (this one appears to have been new in 1993 or 94). (I had one on my bench around 1989.)
This test chamber will serve both for calibration as well as for measuring the SVR board temperature coefficient for about +/- 1 c about the desired calibration temperature.
The silver cylinder (back wall of the lower section of the aluminum box) is a RTD which is monitored in data acquisition by an Agilent 34410A. The yellow line in the cover is a small piece of plastic insulation holding a 10k thermistor in a tiny hole in the cover (for the 5910B feedback temperature). The industrial surplus heatpump (possibly a Marlow or Melcor device, about 4.5 to 5 cm square filled with resin) sits between the CPU cooler heat sink and the top of the Hammond box. The ILX 5910B has a modest 16 W power output (4 V, 4 A), enough power to control the temperature of the small Hammond box with the relatively small SVR heat load (at 15 V, a SVR board draws about 2 mA, no load).
Hammond Manufacturing Die Cast Water Tight Aluminum Box No. 1590Z110 (do not use the rubber gasket) 75 x 80 x 56 mm
Arctic 5 thermal compound was used to couple the ceramic heatpump faces to the the milled copper surface of the CPU cooler as well as to the lightly sanded (emory cloth) cover of the Hammond box. The 12 V CPU cooler fan was run on a 6 V battery. There is also some thermal compound in the thermistor hole.
We ran the micro-environmental chamber at a constant temperature for some hours, followed by approximately one degree positive and negative changes. Other small changes (0.1 or 0.2 C) were made during the run. A box was placed over the stack as a further wind break. The small temperature changes overnight were actual changes that appeared to follow the building's hot air heating system operation (outdoor temps overnight were near 0 F).
In time, here is a graph of voltage difference (SVR to our Fluke 732B transfer standard measured with a hp 3456A 6.5 digit DMM on the 100 mV scale) PDF. The exponential curves on the RTD temperature data (Agilent 34410A) show the delay in "pulling" the temperature of the lower section of the Hammond box to match the controlled temperature of the aluminum box cover.
The data was also plotted in Excel PDF, XLSX, TXT. This board was found to have a measured temperature coefficient of about +1.8 ppm / c (the SVR spec is 5 ppm / c as per the AD587LQ specification sheet). The present value of our recently normalized Fluke 732B transfer standard is about 10.000 001 V, so "zero" on the graph is our 10V calibration point. 10.000 011 is plus 1 ppm (10 uV / 1 ppm - for a 10 V set point) and 9.999 991 is negative 1 ppm for our desired 10 V reference.
The rounded curves are not thermal-voltage hysteresis at the AD587 monolythic reference chip, but rather caused by the relatively large thermal delay between the cover held at a desired set-point temperature and the lower body section including the SVR board under test in the lower section of the aluminum box.
Later testing with a Marlow SE5000 Peltier device temperature controller The small supply on top runs the fan on the CPU cooler that we use on the "Air" side of the temperature controlled stack. The Agilent E3610A powers the SVR board during testing.
COPYRIGHT © 2012 JOSEPH M. GELLER, All rights reserved.