I am allowing the current through a DC motor to flow through a small resistor and measuring the voltage across it. I need to give this analog Voltage to an ADC of micro-controller to do some signal processing on it. My problem is that the Micro-controller runs on an Isolated supply and the voltage to be sensed is on the non Isolated side. I feel that using an analog isolator is not a good solution. One solution to it is using an External ADC on the nonisolated side and digitally isolating the ADC output, but this ADC cannot take negative voltages when motor runs in other direction. Please help.
While I understand that you are asking about how to use a current-sense resistor to sense current and then transfer the value over an isolation boundary, there are some alternatives to consider before you consider the decision finalized.
(Note that you didn't give any specifications or requirements such as bandwidth, packaging, or current range, so the specific part(s) mentioned may not be suitable, but there is a wide range of parts available that will probably work fine.)
Hall effect current sensors allow isolated sensing of current without the need for any electronics or series resistors on the "hot" side of the circuit. The output can be selected to be suitable for direct connection to your isolated microcontroller. For example, if you had a 3.3V microcontroller, and the current you needed to sense was less than +/- 12.5A, Allegro Microsystems' ACS711 will give you a linear output voltage between 0 and 3.3V, with 0A current centered at 1.65V.
To use this with your microcontroller, connect VIout to an ADC pin.
Of course, they make these sensors with different current sensitivities, capabilities, and packages. Digikey is your friend.
We'd like to help, but my answer is about an analog isolator. What's wrong with them? They're made for this.
The IL300 may be useful:
The IL300 has an excellent 0.01% servo linearity. If you want to power U1 from the motor's power supply, make sure it's properly decoupled.
(Vcc and ground left and right of the optocoupler are obviously different.)
(1) Any uC with an ADC of suitable performance. Read value. Send digital data via an optocoupler.
(2) You can buy "linear" opotocouplers that allow replicating a linear voltage across an isolation boundary
For $US2.85 you can get the LOC110 fron IXYS They claim:
The device contains 1 x LED and 2 x matched photodiodes. The pair of photodiodes are used to produce a "servo" such that the two photo diode currents are matched and the input voltage can then be deduced.
In each case, I_PD1 = I_PD2 and follow circuit from there.
As device is current driven via a series resistor and "thinks" in mA terms you will almost certainly have to amplify your motor sense_resistor voltage somewhat. Unlikely to be a problem in the overall order of things.
I faced the very same problem in 1978 for a 48V @ 1A DC motor with remote control and Current feedback over a custom telemetry design. ( now called SCADA) I had designed a 1MBps telemetry link and needed an analog current monitor about a 300m away using the digital telemetry channel from a Reactor power Bldg to the Control Bldg.
My Design Choice:
Instead of ADC , I used tach concept like car...
Now maybe your design to use a similar Tach circuit with Opto-Couplers instead of a telemetry on coax. Tach designs can be made simple for not depends on accuracy.