Oh, now this has become curiouser and curiouser.
"Variable torque" does NOT mean you want the drive to be in torque control mode. All that means is that your LOAD demands less torque from the motor as speed decreases, ie a centrifugal machine like a pump or fan. Vector control, whether torque or velocity, should be irrelevant, and in fact if it is a centrifugal machine, that is the SIMPLEST of control capabilities of a VFD. But why would you be so concerned for low speed operation at all if it is a centrifugal machine? Generally speaking at some point in the operating curve of a centrifugal machine, usually somewhere above 30% speed, it ceases to provide any useful work at all! So either you have a misunderstanding of what variable torque means (or I do!), or you are meaning something different and CALLING it variable torque.
Torque Vector Control in the VFD is for when Torque must be maintained at the EXPENSE of speed accuracy. A classic example is a winder application, something like a toilet paper winder or wire drawing machine. As the take up roll changes diameter, the surface speed at which is pulls the product through would change if the shaft speed of the roll remained constant, increasing the likelihood of breaking or stretching the product. So the shaft speed must be varied to keep the surface speed of the product relatively constant. But the real culprit in risking breaking or stretching the load is the tension applied to the product regardless of speed; that must remain much more closely controlled. So Vector Torque Control will maintain a more precise CONSTANT torque on the shaft as a form of tension control, and the actual speed accuracy becomes secondary to that torque control. If that's what you meant, that is not VARIABLE torque, that is CONSTANT torque at variable speed.
So if that is the nature of your application, and you have properly set up the drive yet continue to have low speed instability, now you might be experiencing the differences in quality that exists out there with regard to vector control algorithms and the feedback mechanisms inside of the VFD, especially when it comes to "sensorless" vector control; not all are created equal and as a gross general rule, the less expensive the drive, the more compromises were made in accomplishing the finer points of that task. For 99% of applications out there, what you are experiencing is uncommon, so the ability to keep it from happening suffers from the laws of diminishing returns for a VFD mfr, meaning if they chase it too religiously, they add cost faster than they increase market share, which makes them lose even more market share.
But all that said, it also might be that if indeed this is a VT application on a centrifugal machine, and you have been chasing accuracy trying to use a torque vector control loop, then I might be inclined to side with Mike Kilroy on this: you have been chasing your tail here when a simpler form of control, such as V/Hz would serve you better.
"You measure the size of the accomplishment by the obstacles you had to overcome to reach your goals" -- Booker T. Washington
