Hello,
I would like to design a voltage bus monitor using AMC1100 or AMC1200. The only difference I seen on them are the isolation voltage, so I will go on with AMC1100, as it has enough for my application. The voltage to monitor in my application ranges from 0 to 40V. This bus power an inductive load that will be commutated , so negative transients are possible also. To solve this I put a 40V TVS in parallel to the load.
Now the relevant data for accuracy calculation: Gain 8 (1%, 56ppm/K), nonlinearity (0.1%, 2.4ppm/K), Vos 1.5mV, input range +-250mV, Rin 28 K (unknown tolerance and ppm/K), output noise (3.1VRMS), external resistors of 0.1% and 25ppm/K.
According to Figure 34 in AMC1100 datasheet, Gerrtot = Gerr + R2/Rin under the assumption that R2 in much smaller than Rin and R1. I've choose R1=28K, same as Rin. Even when maximum possible voltage will be 40V, the application will be normally between 0-12V or 0-24v. Taking this into consideration I've sized the resistors so that when a 40V is present in the bus, 320mV will be at the input of AMC1100. R1=226, reaching 250mV input when 32V is present in the bus.
With this data, Gerrtot = 1% + 0.0080 = 1.0080%, if I add 0.1% of each resistor and with a temperature drift of 20ºC (45ºC max.), 25ppm/K*20/10000=0.05%, I get another extra 0.1%+0.1%+0.05%+0.05%=0.3% error due to chosen resistors and 20*56ppm/K/10000=0.112% due to TCGerr.
So adding 1.0080%+0.3%+0.112%+0.1% (of nonlinearity) = 1.52% error, this is not taking into account the Vos, as it will introduce a variable error depending of the voltage measured and maybe it could be avoided with calibration. So when measuring 12V, I get an error of 182mV. I think this is very high error, are my numbers right? Any option to improve accuracy?
Also I would like to attach output directly to ADC as single ended, is it possible to do this with AMC1100 VOUTP or should I use an Op. amp at the output to feed the ADC?
Thank you.
