Could anyone suggest low-cost and reliable alternatives for resolver? This position sensor is used in an automotive PMSM motor.

Back EMF sensing might be feasible. It will not work for low or zero speed though.

Thank you Anirban. I ever considered the sensorless control, but worrying about its impact on automotive safety. This motor is a driving force in a pure electric/hybrid vehicle.

resolvers can be had for same or lower cost than digital encoders.....

But for auto propulsion system, 3 simple 10 cent hall sensor transistors should work just fine. My 100% electric car uses resolvers but halls would have been ok I believe.

Do you mean hall sensor for brushless DC motor? Mine is pmsm. I need a position sensor to generate sine waveform.

no you don't but that is another discussion.

perhaps I should explain what I mean. A "pmsm" is same animal as a "brushless dc;" they are both simply ac permanent magnet synchronous motor and operate the same way, so both can be commutated with halls only or some other scheme. Either can be wound to have sine or more trapazoidal BEMF, but the fact remains you can commutate either design with 6 step or 1,000,000 step method. Now, if you are stuck with a given DRIVE that accepts only resolver feedback, you must use only resolver feedback - or make a converter from your device to a resolver like signal. Ditto for halls or encoder or other scheme. But your question had no reference to a drive and its limitations, so I answered your question. You CAN use halls only at 10 cents each x 3 to run YOUR motor.

If your drive limits you to sine feedback only, you make your own sine/cos feedback as I have with 2, again almost free, analog halls separated by 1/4 your pole pitch.

Mike, thank you for this detail comment. To make it clear, the drive limits the motor commute not as 6-step way, but much more fine. It does not limit the position sensor feedback as sin/cos format, but should be fine enough to support commution and smooth torque.

There are plenty of pmsm motors running smooth enough to satisfy your comment "but should be fine enough to support commution and smooth torque." But if your drive limits you from using halls only, so be it. I just wanted to point out that your motor design does not limit you to just resolver, which is how I read your initial post. BTW, what is the minimum sin/cos resolution per pole your drive can accept? Does your drive have any built in sin/cos division (multiplication) capability? The homemade sin/cos 1 sine/pole unit we made was used in a Kollmorgen servodrive with built in upto 4096x multiplication; hence, we had upto 4096PPR and had very very smooth performance on these linear motors from 0mm/sec to very very fast.

I think Mike is suggesting here to use the speed of the motor along with Hall sensor crossing of the rotor poles to estimate the rotor position for sinusoidal commutation. It should work pretty reliably at high speeds and can be made ideally efficient if you write a good algorithm.

I have used Hall sensors with high-speed timers and hardware capture inputs on processors to obtain a position estimate between Hall transitions. It works well even at relatively low speeds but I might add that this depends on how good your velocity term and algorithm are. If there is substantial velocity term "jitter" then you will have position estimation jitter as a result. Overall, however, I found the Hall sensor approach very reliable, cost effective and not subject to the tracking problems at higher speeds inherent to resolver-to-digital conversion processes.

Brad / Anirban - forget sensorless or interpolating from digital halls for traction applications operating near zero speed. I agree with Mike that two linear halls 90 electrical degrees (or the equivalant) apart with some relatively simple processing could be a good low cost solution. However if you are using a FOC algorithm you do need quite low distortion feedback otherwise you will generate all kinds of horrible current harmonics that will give you torque ripple and NVH issues. On some applications it possible to directly sense off the main rotor magnets which is clearly the cheapest option.

if one is buying off the shelf stuff (motors/drives/feedback) then I think one needs to use stuff MADE to work together.

if one is into making their own stuff then the homemade sine-analog-hall or digital hall ONLY approach is worth pursuing.

Keep in mind that for a car electric system there is NO velocity loop but simply a torque loop with throttle pedal being command for torque, so no need for real fine velocity loop high BW control.

KB Electronics has developed a drive for 230 VAC PMSM motors that works without any external feedback. What is the motor rating and what is the speed range?