I have three questions about direct torque control:
1. what is the disadvantage of DTC?
2. Is DTC popular in industry drives? I only know ABB drives have DTC control.
3. Why we need sensorless control for the control of PMSM since DTC is sensorless in nature?
Waiting for your help.

Hi Muyang,
I don't know of any other company than ABB using DTC in their drives.

I think saying that DTC is sensorless in nature is a bit misleading, since it is only sensorless as a torque control, while most applications require speed (if not position) control.
The flux magnitude and angle must be estimated, which is common to sensorless FOC, too. How to achieve this is often left to the reader's immagination, especially in ABB's documentation.
There are also FOC techniques which are in principle similar to DTC, like Stator-FOC, where you estimate the total stator flux magnitude and direction, and regulate quadrature current for torque. There is also the so-called SVM-DTC, which is a torque and flux control by means of PWM.

IMO, the most evident disadvantage of DTC is that, unless you update control very very fast (MHz range), current ripple will be higher than with PWM, since in DTC an inverter active voltage vector can be applied at least for one update period, while in PWM it can be applied even for just one count.

PS: we already discussed about DTC vs. FOC in this group.

Thank you! Sandro. Your answer gives me a great help.
I noticed that some papers mentioned about the high current ripple which causes high torque ripple but others don't. At least the ABB document did not mention this. I agree with your opinion that the DTC is only sensorless for torque control but what I'm still confused is that the sensorless control can also control the torque. For now I can conclude the answers like this:
1, High current ripple or high torque ripple.
2. DTC is not pupolar in industrial drives.
3. The application decides which control should we use. Sensorless control is used for sensorless speed control of PMSM. While DTC is used for sensorless torque control of PMSM.

Muyang,

When ABB started using DTC, it had some advantages because the switching and fast control loop was simple enough to be executed in the FPGA's available at the time, which made the current control very good relative to the update rates possible with then-current 16-bit microprocessors. These days we have sufficient processing power in our drives that, as Sandro pointed out, either rotor or stator-flux FOC is capable of doing anything that DTC can do. Actually, if you optimize either technique for the application, the actual inverter switching patterns start to look nearly identical.

My approach has been to pick one, learn it thoroughly, and apply it.

I think DTC started with analog comparators, is it right?

I thought it was FPGA's, but I could be remembering wrong.

DTC is a replacement of motor+gear. Usually it has high torque and low speed (and many pole pairs). I don't think it has high current/torque ripple, because usually it has high inductance.
DTC can be control as any servo drive, if you have a high performance drive.

Muyang,
Direct torque control can be done with or without sensorless control - They aren't mutually exclusive. You don't "need" sensorless control of PM motors. As Neil stated, when optimized, the actual inverter switching patterns start to look nearly identical between FOC and torque control if you've got your control algorithms tuned.

Muyang
the reason DTC is not found in other drives out there besides ABB is because is a patented control method that has been released just recently

I think Eric's answer is more reasonable to explain why only ABB has DTC.

The original 1984 patent expired 10 yrs ago. There have been several patented improvements to DTC since then, mainly to reduce the torque ripple due to the original hysteretic control.