ADR5045 for voltage divider as shunt regulator

Hi Andy,

I'm currently working on this, I'll get back to you as soon as possible.

Regards,

Jay

Output is to an ADC (stm32) and NOT DAC for measurning voltage. I'm sorry for my error.

Andy

Hello Andy,

in principle your setup should work, though I calculate slightly different numbers than you do. With Vin=16V, I get 797µA through the divider and 3,7V at the ADC input (which is most probably above the supply of your STM, so the voltage should start clippling.)

The ADC input on your µC most probably behaves like a switched capacitor due to its sample and hold circuit. Make sure, that the sampling time of your ADC is long enough for the given source impedance (4k7 // 9k1). If Vinput varies slowly I would add a small capacitor to GND at the ADC input.

The benefit of this subtraction-circuit is not too high: if you would omit the subtraction and adjust the voltage divider, it would cost you (less than) one bit of your ADCs resolution. Subtraction of offsets could easily be done numerically in your µController after conversion.

Hi

Thank you very much. Very helpful input. How do you calculate the benefit of this subtraction-Circuit? May you give to me a hint?

I will add a small capacitor and also adjust the voltage divider. i have calculated with an impedance of 50k for the stm32. so i will get slightly lower voltage maybe?

Hello,

without analog subtraction and a properly dimensioned divider, you would map the range 0..16V to your ADC. I assume you ADC has 10bit, so the voltage resolution refered to Vinput would be 16V/2^10=15,6mV.

With subtraction and a proper divider you map a range 11V to the input of your ADC (from Vinput=5V to Vinput=16V). The voltage resolution referred to Vinput is 10,7mV. The improvement in resolution beween both scenarios is a factor 1,5, i.e. less than a factor of 2 (so less than one bit of resolution). The disadvantage is the additional error introduced by the subtraction (though this may be almost negligible for the ADR5045).

The subtraction would help more in a case, where the subtracted voltage is large and the remaining voltage range is small. You have such situations often e.g. in high-side current measurements. But then I would normaly use a diffamp with adequate input voltage range to execute the subtraction, not a voltage reference in series.

"i have calculated with an impedance of 50k for the stm32" Typically the DC impedance for an ADC input will be much higher. I do not have the STM datasheet in front of me, but maybe you calculated the 50k based on the biggest possible leakage current at the pin? Well, the typical leakage current should be much smaller than the biggest possible value, which occurs only at extreme temperature, voltage and process corner. Typically  the DC input current is much less important than the pulsed input current which occurs, when the sample and hold stage is switched to the input. The source impedance msut be low enough to charge the S&H-capacitor in time. An external capacitor of some nF will provide the charge for this pulsed current and eliminate this problem.

Thank you very much for this clear exlenation. You helped me alot. I will certainly change the voltage divider resistor

Andy