# Can 3-phase FOC control be done without current measurement?

Can a FOC control scheme be done using a rotary encoder to measure the position, speed and acceleration directly, or is measuring current necessary?

FOC is a motor control scheme as described in links below:
http://www.atmel.com/Images/doc32126.pdf
http://www.eetimes.com/document.asp?doc_id=1279321

Field Oriented Control is one of the methods used in variable frequency drives or variable speed drives to control the torque (and thus the speed) of three-phase AC electric motors by controlling the current. With FOC, the torque and the flux can be controlled independently. FOC provides faster dynamic response than is required for applications like washing machines. There is no torque ripple and smoother, accurate motor control can be achieved at low and high speeds using FOC.

• You need to regulate the current in order to achieve FOC. You need some kind of direct or indirect current feedback for the regulation. – Eugene Sh. Jan 13 '16 at 19:38
• @Tomsz, you might include the rest of us by editing your question and explaining what FOC is. – Transistor Jan 13 '16 at 20:13
• I don't think it's possible to answer the question without knowing what FOC is and how it works, but I'll edit in an explanation shortly. – Chumanista Jan 13 '16 at 20:15
• "There is no torque ripple" LMAO ... – JonRB Jan 13 '16 at 20:53
• yup. Between harmonics in the flux, cogging torque, PWM etc... "no torque ripple" is salesspeach – JonRB Jan 13 '16 at 21:00

You appear to have two questions.

Can 3-phase FOC control be done without current measurement?

No; The point of field orientated control is to ... control the field in the electrical machine. This relies on knowing what the field is

Can a FOC control scheme be done using a rotary encoder to measure the position, speed and acceleration directly, or is measuring current necessary?

Well, Field orientated control relies on knowing where the rotor is so that correctly phase aligned currents can be synthesised.

A rotary encoder (resolver, analogue hall effect sensing etc...) can be used to measure the instantaneous position & thus derive speed and acceleration. The existence of position sensing is to complement current sensing.

Sensorless control would remove the need for a rotary encoder, although the workable schemes end up using voltage sensing so "sensorless" is a bit misleading.

• So the best course of action would be to use shunt resistors (can the MOSFET resistance be used for this purpose?) to measure current AND the encoder? I was thinking AS5047P and DRV8302 for the driver + amplifier. Is this a good choice? – Chumanista Jan 13 '16 at 21:09
• such encoders are fine ( i have used equivalent from Allegro). FET's as current sensing... its doable and there are special FET's out there with additional pin to cut down on noise. Theoretically speaking... you could do it with one current sensor (in the DC+ between capacitor and inverter) & reconstruct each phases current by the fact you know what FET's are switching & where you are... It does remove capability to monitor and adapt. Due to this I settle for hall sensors in each 3 phases – JonRB Jan 13 '16 at 21:14
• You mean Hall sensors for current measurement? Could you link an example for ~15A 16V use? – Chumanista Jan 13 '16 at 21:15
• The one given by SunnyBoyNY LEM.com – Charles Cowie Jan 13 '16 at 21:40
• @TomaszTkacki Allegro do a nice range ( allegromicro.com/en/Products/Current-Sensor-ICs/… ) 5V supply, output is ratiometric to the rail so perfect for connecting directly to an ADC. Output biased at Vcc/2 for pos and neg current measurement, needed for FoC. Freq range from DC to 120kHz, more than enough for FoC – JonRB Jan 14 '16 at 21:08

I believe that you would need a secondary current control inner loop in the control scheme for any motor that has an outer speed or position control loop. The major general-purpose VFD products on the market have current based control as the standard offering and encoder feedback as an optional feature. Most research has been focused on improving performance with using inexpensive current based control methods rather than the more expensive encoder based methods. You could search the literature to see if anyone has proposed a scheme using an encoder without current sensing.

• Yes, yes, precisely. Good point about the literature search. – SunnyBoyNY Jan 13 '16 at 22:14

Proper observer structure can give an estimate of phase currents. But the observers would have to be open-loop, which makes them very sensitive to parameter variation (inductance, resistance, bus voltage, bEMF constant, ...)

But:

No serious FOC motor drive is implemented without any kind of current sensor. Phase currents can be reconstructed from the DC link current, which is an option.

There is always some torque ripple due to the switching nature of the motor drive and always somewhat unbalanced machine structure (e.g. phase inductances cannot be exactly the same). It could be 1% or 20%. High performance FOC or DB-DTFC (deadbeat direct torque and flux drives) can achieve the 1% but typically not less.

More on Current Sensing

The best results can be obtained with at least two phase current sensors and a resolver or encoder. A lot of research has been done on "sensorless" control, which actually does not remove the current sensors but the position sensor.

The purpose of phase current sensors is not the machine control but also the motor drive protection.

A very popular option is to use closed-loop LEM current sensors or three-phase current shunts. Both will give all three phase currents, which are needed to implement FOC.

• For my applications 1% is negligible, really. The quoted text was copied, thanks for the correction! – Chumanista Jan 13 '16 at 21:06