calibration circuit suggestions

Erik Friesen

I am in need of a circuit that can output a calibration value of 0.1 mv, and 40mv. I suppose I could buy a fluke 715 www.myflukestore.com/calibrators/loop-705_707_707ex_715.php but I don't feel like have two or three of these $700 items floating around. Right now I have been using a voltage divider (pot) with a 1.5v alkaline cell, and I use a separate multimeter to view the output. I want to make a few calibrators which only have two or three preset outputs.

Does anyone know of a inexpensive item that would work, or a simple circuit that could be easily produced? I have considered using a precision voltage reference, divided with precision resistors, but I am not sure that would be accurate enough, as accuracy needs to be within 0.05 mv, and the resolution on the instrument to be calibrated has a .025mv resolution.

Phil Pilgrim (PhiPi)

My inclination here would be to think more in terms of a current reference than a voltage reference. For example, by connecting a precision 0.1 ohm current sense resistor across the device inputs, right at the device itself and driving it with a 1 ma source, the device will see 0.1 mV. By regulating the current instead of the voltage, you avoid problems like line loss and noise pickup.

-Phil

Erik Friesen

That is a new thought to me, but the problem is still similar, because how can you get a source tight enough, coupled with the right shunt to produce exactly 40.00mv? I would be very curious how fluke is doing it, and I am doubtful that they have $100 chips doing the work.

Leon

Use something like this:

www.national.com/ds/LM/LM4050.pdf

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Leon Heller
Amateur radio callsign: G1HSM

T Chap

Also look at AD628, programmable output precision dif amp.

www.analog.com/en/amplifiers-and-comparators/current-sense-amplifiers/ad628/products/product.html

Produce whatever output you want.

Erik Friesen

@leon, how do you suggest to divide the voltage from the output? That particular item has a minimum of 2.048 output, as near as I can tell.

@Todd Chapman, with a min gain of .1, and an offset of +-1.5mv, that would equate of .15mv that would need to be calibrated somehow, unless I am not understanding your usage idea.

Leon

Just use a couple of precision resistors. They also have a circuit in the data sheet that gives a precise current, you could use that with a suitable resistor.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Leon Heller
Amateur radio callsign: G1HSM

Erik Friesen

Ok, at 0.1% output tolerance that would be acceptable for the shunt, but that would put a fairly high precision requirement for the resistors, unless I use a pot to calibrate them. According to my math, a .01% resistor would still allow a variation of 0.2 mv at my output, assuming I would use the 2.048 output. Correct me if I am wrong here.

Setting up a 0.1% or greater current source doesn't seem any easier to me.

Tracy Allen

One solution, maybe even easiest and most versatile, would be a precision 16 bit DAC. Something like the Linear Tech LT1655 in an SOIC8 package with SPI interface to your Stamp or Prop to set the voltage to whatever you want. It has a built-in reference, about 0.3% or better initial accuracy, but you can overdrive the reference input with an external precision reference, for example, the LTC6655 with 0.025% accuracy and <5ppm drift. The DAC also has an output buffer to drive a small load. The step size is on the order of 50 microvolts, so the PCB layout has to be good. There are quite a few capable DACs on the market.

There is no getting around the fact that you either need a calibrator for your calibrator, or you need to start with components of high precision. Maybe you can find a lab at a school or local industry that would give you half an hour to adjust your trimmers alongside a trusted 6 digit or better multimeter.

I bought an Extech CMM-15 process calibrator/multimeter several years ago, and while not as pricey as the Fluke 715, it has been an indispensable tool because of the frequent of calibration and linearity testing I need to do in various ranges. These things are not without shortcomings. In the range of 40mV or 0.15 mV, I would find much too much noise on the output. (Noise is due to the switching power supply that generates the +/- 15 volts inside the meter) I need to do millivolt level testing quite often, so I built external circuits using precision voltage dividers, filtering, and CAZ op amps (LTC10551), to bring down 0 to 10 volt from the Extech to 0 to 100 mV or 0 to 10 mV or 0 to 100 microAmp for device under test.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Tracy Allen
www.emesystems.com

Erik Friesen

Here is what I am going to do -

One of these search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=811-1054-ND

Plus one of these search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=GU2021S26-ND

I am going to set up a voltage divider based on a 1.5v battery, and I'll tie the output to the meter, and to the outgoing leads. The murata engineer seemed to think his panelmeter was up to the task, so I am going to give it a try.

kwinn

An excellent idea. I was going to suggest you use a 4 1/2 digit meter, 3 resistors in series for your 3 ranges and a constant current source with a trim pot to get the voltage dead on.