Using an N-Channel mosfet to control current in a regulated power supply

[SharpEars]

Is the idea in the subject line sound? For example, let's say I want to accurately limit my power supply's current output to a range of 0-500 mA. If I use an N-Channel enhancement mosfet in its saturation region and vary Vgs with for example a DAC between the threshold voltage Vth and Vdd - X, where X represents whatever voltage will cause the current to be limited to 500 mA, I should be able to accurately control current for a load, assuming that Vds (i.e., the power supply output) is relatively constant.

Is this sound?

[Kleinstein]

Using just the MOSFET curve to do current limiting gives a crude, not very accurate limiting of the current. Close to the maximum set current the voltage will already drop by something like 0.5 - 1 V. The current output for a given gate source voltage depends on temperature. So the current will depend on temperature.

In case of a short the full voltage will be at the MOSFET and thus there will be quite some heat produced. No all MOSFETs can handle that, especially at higher voltages like 20 V. An MOSFET large enough to handle the heat can have quite a steep curve.  So small changes in the gate voltage can change the current quite a lot. It will thus be difficult to get an accurate setting.

[Kalvin]

Adding an op amp would give better current regulation. Here are some ideas: http://www.ti.com/lit/an/sboa046/sboa046.pdf See section "Current Sources and Sinks and Current Mirrors Using an Amplifier and a Series Pass Element" on page 7 onwards.

[SharpEars]

Adding an op amp would give better current regulation. Here are some ideas: http://www.ti.com/lit/an/sboa046/sboa046.pdf See section "Current Sources and Sinks and Current Mirrors Using an Amplifier and a Series Pass Element" on page 7 onwards.

Wow, thanks for that link. It's a treasure trove of ideas for current limiting...