Current sense

If something blocks an electric motor from turning ("stalled"), the current rapidly increases far above normal levels -- then the motor driver needs to turn off the power before the motor, or the motor driver MOSFET, or both are destroyed.

Systems that use a 4-20 mA current loop also need to measure current.

There are 3(?) basic techniques: for measuring current:
 * resistive shunt
 * low-side current shunt
 * high-side current shunt
 * magnetic field sense
 * MOSFET voltage
 * the "non-dissipative overcurrent protection", a kind of current mirror used in the L6208N ...
 * ... (any others I missed?)

resistive shunt
With the resistive shunt technique, all the charge flowing through the the load also flows through a resistor (the shunt resistor) in series with the load.

"What's All This Shunt Stuff, Anyhow?" by Bob Pease 2002 describes one way to custom-build a high-current shunt.

"What’s All This Wattmeter Stuff, Anyhow?" by Bob Pease 2002 describes how to use the previous shunt in a circuit to be spliced in the middle of an extension cord to measure true RMS watts flowing from a power outlet to a device under test. Even though a copper shunt has a pretty bad resistance variation with temperature (tempco), other parts in this circuit compensate for it.

low-side current shunt
+-+      |                 |       |  +            (load) (power)             | | -            (shunt R)       |                 | +-+      |      GND

Low-side is (electrically) the simplest.

For smaller motors, the current is usually measured by running the current through a shunt resistor to ground, and measuring the voltage across the resistor.

high-side current shunt
+-+      |                 |       |  +            (shunt R)    (power)              | | -            (load) |                |       +-+       |      GND

In situations where low-side sensing is difficult ( automobile electronics bonded to the "GND" car frame; other systems where it is inconvenient to put a resistor on the "lo" power wire), we turn to high-side sensing.

Maxim application note 746: "High-Side Current-Sense Measurement: Circuits and Principles"; Newark: high side current sense; Digikey: high side current sense; Linear: current sense circuit collection (why doesn't this include the Linear LTC6103 ?); Texas Instruments: "Current Sensor", Silicon Labs' current sensors. A few op amps can handle common-mode voltage well outside its power supply -- such as the TI INA117, which when powered by +/-15 V, can handle a common-mode voltage of +/-200 V. This is useful for high-side current sense and also 4-20 mA current loops.

magnetic field sense
For large motors, the current is measured by running the power wires through a magnetic field sensor -- either
 * directly measuring the magnetic field with a Hall effect sensor, for example, the Allegro ACS712 or other [Allegro Hall-effect current sensors]), which can measure DC and AC current, or
 * 30 Amp Current Sensor AC&DC A commercial sensor based on the Allegro ACS712.
 * directly measuring the magnetic field with a magneto-resistive effect sensor, for example, the Zetex magneto-resistive current sensor, which can measure DC and AC current, or
 * indirectly measuring the magnetic field with a "one-loop current transformer" (which can only measure AC current). AC current sensor
 * indirectly measuring the magnetic field with a "Rogowski Coil" (which can only measure AC current) and an integrator circuit. Wikipedia: Rogowski Coil.

Because magnetic field sensing is inherently non-contact, it works just as well high-side as low-side. ( "Closed-Loop Magnetic Current Sensor". )

Douglas W. Jones "Current Limiting for Stepping Motors" mostly discusses resistive shunts, but it has some information on magnetic field current sensors in the "Other Current Sensing Technologies" section.

MOSFET voltage
pseudo 'mirror' current sensing a MOSFET -- sampling the voltage across a MOSFET while it is turned on. That voltage is linear with current but varies with temperature. If the purpose of measuring current is to turn off the MOSFET before it overheats, the variation with temperature doesn't matter. (A true current mirror isn't useful for motors, right?)

This uses the turned-on resistance of the MOSFET as if it were a shunt resistor.

current estimation
Rather than directly measuring the actual current, many people measure something else that (they hope) is close enough to being proportional to the current.

lossless average inductor current sensing
The "lossless average inductor current sensing" technique:

(switched)+-+ (power)          |     | |    Rs                      |     | (L1 + DCR) +--- Isense+ |    |                      |     C                      |     | +-+--- Isense- |                   (load) |                    GND

The current through an inductor with some known internal parasitic resistance DCR can be sensed with a series resistor (Rs) and capacitor network connected in parallel with the inductor-resistor network. The designer picks the sense components Rs and C such that Rs*C -- the time constant of the RC network -- is close enough to L/DCR -- the time constant of the inductor resistor network. Then the voltage across C is proportional to the current through L. (IR3508Z data sheet ) current through L == voltage across C * (1/DCR).