I recently repaired a Weston 4442 (3.5 digit dual-slope DVM) which had a failure of the integrating cap. It had decreased in value from 2.2 uf to 0.42 uf; a radial-lead green "box" containing an unknown dielectric) Not having a suitable polystyrene or Teflon cap in the junk box, I used a handy polyester cap (looks like an orange-drop but white epoxy). Any ideas as to how much error the (presumably) Mylar integrator cap will introduce? In other words, will such error be significant in a mere 3.5 digits? (I imagine that Weston would have used a less expensive cap if it would do the job). The absolute value isn't important due to the dual-slope technique, but time-variant storage effects are... I have an HP 4.5 digit DVM and a MAX6385 (2.5 volt) reference, but no "volt-nuts"-league meters or standards, other than a 596.5 ohm 0.01% resistor ;) thanks Charles

Charles Morris wrote: > I recently repaired a Weston 4442 (3.5 digit dual-slope DVM) which had > a failure of the integrating cap. It had decreased in value from 2.2 > uf to 0.42 uf; a radial-lead green "box" containing an unknown > dielectric) Not having a suitable polystyrene or Teflon cap in the > junk box, I used a handy polyester cap (looks like an orange-drop but > white epoxy). > > Any ideas as to how much error the (presumably) Mylar integrator cap > will introduce? In other words, will such error be significant in a > mere 3.5 digits? (I imagine that Weston would have used a less > expensive cap if it would do the job). The absolute value isn't > important due to the dual-slope technique, but time-variant storage > effects are... > > I have an HP 4.5 digit DVM and a MAX6385 (2.5 volt) reference, but no > "volt-nuts"-league meters or standards, other than a 596.5 ohm 0.01% > resistor ;) > > thanks > Charles > > > The effect will depend on the dual slope integration time and could be as high as 1% depending on the particular capacitor. Bruce

> Any ideas as to how much error the (presumably) Mylar integrator cap > will introduce? In other words, will such error be significant in a > mere 3.5 digits? Hello, several years ago I built a integrating ADC. 2.5V Reference as "zero voltage" and integrating +5V and 0V against the input signal. (0 .. 5V measured as -2.5 to 2.5 V). Due to capacitor soaking I had a gap in the middle of the 0 to 5V range which was about 1mV (0.02%) as non linear jump. I could reduce the non linearity by replacing the mylar capacitor with a polypropylene capacitor by a factor of around 10. So for a meter with 3,5 digits this should not play a role. With best regards Andreas