Building a calibration Device

[snx]

I have multiple old devices, that run on the rather ancient ADC MP2734

This ADC has the following parameter:

- 14bit
- +-0.5 LSB Differential Linearity
- +-1ppm FSR/°C Differential Nonlinearty
- 0.005% p-p Noise
- +-3ppm Offest Tempco FSR/°C
- 250 Ohm Impedance
- Relative Accuracy 0.006%
- Absolute Accuracy 0.006%
- Noise 83uV rms
- Full scale 0-10V

Before this ADC, there are multiple gain stages, the lowest is 4, so about 2.5V input, will produce a full scale of 10V at the ADC. There are multiple gain settings by software which are where the majors are [4,10,20,40,100,200,500,1000]. At a Gain of 1000, 10mV at the input, will create a full scale range.

For this, i would like to build a calibration device, that can produce at least three calibration points on the scale, for example [500mV. 1500mV, 2000mV] on the gain of 4. I do not expect to need gains higher than 100, so i only need to calibrate stages [4,10,20,40,100].

Important for the calibration, it that a repetitive, exactly matching pulse can be produced. It does not matter if the pulsed voltage is near a defined number, it can be any number, but it should be always precise. For example, it does not matter if its possible to set 1.00001V, because 1.02345V will also do, but this 1.02345V should always be exactly the same number

I guess, the best way would to use Reference+DAC+Opamp and a MCU to set the values.

But since there are alot of different opamps, dacs, and references, i would like to ask, if someone could give me some tips how to choose the parts

[nctnico]

You might be better off by buying or renting a calibrator.

[mawyatt]

You can find a number of DIY reference designs over on Metrology. We did a few ourselves using various voltage references like the LT1236, ADR01A, ADR4550, LM399, and LTZ1000. The last two images are for DMM calibration, directly plug into DMM and include precision DC, AC and Resistance.

Best,

[mawyatt]

Also made a couple Precision Resistances fixtures, including zero ohm shunt, for DMMs as shown.

Lots of options for DIY if you are inclined.

Best,

[snx]

Sorry if i've not mentioned it / wrote unclear.

I know that there are references, but the problem is, that i need pulsed signals, which means they have to turn on for a few us, and then turn off again, this cycle has to be repeated multiple times, how to identify how much the ADC will give "off" results, to find out the tolerance.

So i need to switch off and on the signal by some ways.

[Kleinstein]

For such signal switching one could use CMOS switches (e.g. like 74HC4051, DG408 or similar) and than maybe a simple resistor divider to get different reference levels needed. Chances are one would measure the actual voltages with a DMM anyway.   

[mawyatt]

So you want a pulsed output of precise amplitude in the "ON" state and ground or an open in the "OFF" state? You can use analog switches for this, or use discrete MOS devices. The DMM calibrator shown switches from 5.000VDC (reference voltage) to ground at precise 50% duty cycle, which yields an accurate 2.500VAC rms waveform for DMM calibration. This uses discrete N & PMOS devices (low Ron) for final switching, and driven by standard CMOS logic.

Best,