Has anyone experimented with a gimbal bldc motor for a direct drive turntable?
My belt drive turntable DIY experiment is working well. I am finding different belt materials have huge impact on speed stability/ wow and flutter. (Silicone belts are terrible for this btw. Standard butyl is ok, but needs a second pulley opposite the motor to reduce lateral stress on the platter. My next experiment there will be with surgical silk for a custom belt.)
But I would like to compare it to a direct drive platter to see if there is further improvement. I ordered an 11pole pair hollow shaft gimbal motor. I have several FOC/sinusoidal /vector controllers to play with to see how well I can spin ot at 33.33 with a weighted platter.
My belt drive turntable DIY experiment is working well. I am finding different belt materials have huge impact on speed stability/ wow and flutter. (Silicone belts are terrible for this btw. Standard butyl is ok, but needs a second pulley opposite the motor to reduce lateral stress on the platter. My next experiment there will be with surgical silk for a custom belt.)
But I would like to compare it to a direct drive platter to see if there is further improvement. I ordered an 11pole pair hollow shaft gimbal motor. I have several FOC/sinusoidal /vector controllers to play with to see how well I can spin ot at 33.33 with a weighted platter.
Gimbal motors rely on external sensors, and are not necessarily designed for low cogging, so a quality tacho might be needed for getting a good speed control loop. Are the bearings quiet enough for a turntable? Turntables use a central bushing in heavy oil for a reason - its acoustically very quiet.
I suspect a heavy platter will help with performace, but spin up may be sluggish.
I suspect a heavy platter will help with performace, but spin up may be sluggish.
I have tried this and with relative sucess on a reel to reel deck. Use an FOC bldc motor driver with a motor with most poles available and a somewhat heavy platter just to be absolutely sure but I suspect it wint be an issue. What might become a problem is the bearings in the gumbal motor. While these motors are built with weird torques and loads in mind from unusual directions of unusual magnitude, axually the bearings can be noisy. Perhaps replacing them with angular contact ball bearings or self aligning double row ball bearings would be a great solution for you aswell. On my reel to reel I have replaced both front and rear bearings for self aligning ones and also changed the shaft so this became a direct drive capstan motor.
I bought this 22-pole gimbal motor: https://www.ebay.com/itm/133942052013
It is made for heavy camera gimbal / motion tracking. Nice motor.
I am using an Arduino-compatible Nucleo64 and the SimpleFOCShield controller. I have a 4lb glass platter and a subplatter from a Rega. Preliminary spins set at 33.33rpm are pretty good in OpenLoop sinusoidal/space vector mode but still not good enough speed stability. I have a rotary encoder that I will attach and try closed loop next. The motor is dead quite and I dont feel any cogging when it is weighted down with the platter (I do feel cogging when unloaded tho).
My goal is to beat my belt driven table setup, where I have speed stability within 0.2% and wow& flutter down to 0.1%. I found there that belt material / streachiness, amount of tension, pulley diameter, etc all have huge effects on speed stability and it was driving me a bit insane.
It is made for heavy camera gimbal / motion tracking. Nice motor.
I am using an Arduino-compatible Nucleo64 and the SimpleFOCShield controller. I have a 4lb glass platter and a subplatter from a Rega. Preliminary spins set at 33.33rpm are pretty good in OpenLoop sinusoidal/space vector mode but still not good enough speed stability. I have a rotary encoder that I will attach and try closed loop next. The motor is dead quite and I dont feel any cogging when it is weighted down with the platter (I do feel cogging when unloaded tho).
My goal is to beat my belt driven table setup, where I have speed stability within 0.2% and wow& flutter down to 0.1%. I found there that belt material / streachiness, amount of tension, pulley diameter, etc all have huge effects on speed stability and it was driving me a bit insane.
Yeah simplefoc arduino is not gonna cut it man. You will have to oay up and get Odrive for this one. The limitation is indeed your controller and adding a encoder is not ginna help.
In constant velocity mode it is beneficial to have hall sensor for rotor pole position tracking so that it can best apply the sinusoidal currents. Its much better than the encoder way, and as a bonus you can make a PLL feedback aswell which is going to truly get you an accurate speed regulation. Or just add mass to the platter. That is a much easier approach (add until you get to 70% of the axial load of the bearings.)
However if you want to keep it lightweight, you cant avoid Odrive with a better quality motor or least one with higher pole count.
On a sidenote please take electrical noise into account as the cheap FOC controllers tend to use relatively low frequency switching frequencies to create the PWM sinewave. Some drivers can and will go up to 100+kHz and those are desireable...from what I heard Odrive stuff goes right up to 500kHz.
In constant velocity mode it is beneficial to have hall sensor for rotor pole position tracking so that it can best apply the sinusoidal currents. Its much better than the encoder way, and as a bonus you can make a PLL feedback aswell which is going to truly get you an accurate speed regulation. Or just add mass to the platter. That is a much easier approach (add until you get to 70% of the axial load of the bearings.)
However if you want to keep it lightweight, you cant avoid Odrive with a better quality motor or least one with higher pole count.
On a sidenote please take electrical noise into account as the cheap FOC controllers tend to use relatively low frequency switching frequencies to create the PWM sinewave. Some drivers can and will go up to 100+kHz and those are desireable...from what I heard Odrive stuff goes right up to 500kHz.
I also have a Solo Mini controller. Cant get it to work with any of my motors reliably. I have found SimpleFOC to be much better.
Neat project! Will you compare it to a vintage DD motor? My faves are the Technics SL-5/SL-6/SL-Q5/SL-Q6 "pancake" motors and the JVC (Victor) JL-B37R FG servo motors, both available for under $50 with drive electronics.
Cheers, Mike
I dont have either of those motors or tables to compare to. I used to have an old JVC QL-A5, but the tonearm broke and I retired it. I should have saved the motor.
I might look into picking up one of those vintage motors.
I might look into picking up one of those vintage motors.
Technics motor PCB on EBay just need the platter/magnet to go with it Technics SL-QD33 platter
PCB runs on 12VDC, simple 2-wire hookup; no other inputs needed to run at 33 rpm.
PCB runs on 12VDC, simple 2-wire hookup; no other inputs needed to run at 33 rpm.
Do these BLDC motors need trapezoidal waveforms? The ones I have used do, but sort of work on sine waves…
Brian
Brian
Gimbal motors while can operate in trapezoidal mode are designed for precision positioning using sinusoidal controlling. Its far better than trapezoidal controll as it doesnt exert a constant torque or power onto the driven rotor in its entire electrical 360° cycle. It rather behaves as pulsed torque. To state the obvious this is inadequate.
They are mostly just driven like a synchronous motor. Ideally you would put hall senser to magnetically figure out where the rotor magnets positions are so phase current could be applied reasonably
Did I stutter? No encoder, hall effect sensors.
An encoder magnetic or optical can only tell the position or velocity of the rotor. However the controller has still no idea where the magnets, the poles are located relative to the slats , the coils so it cannot deliver phase current properly.
To achieve rotation where cogging wouldnt happen you need positional knowledge of the poles not the rotor.
The driver cannot correct for cogging based on positional feedback of the rotor.
Once you know how often you "commutate" or how many times your phase current passed over the 0 crossing point you also have a relative and accurate velocity measurement. Instead of rpms however you get rpm in the frequency of commutation divided by the number of poles.
(24 poles commutate at the rate of 4.5Hz : 24/3=8
4.5/8=0.5625 rps=33.75rpm.)
Unless you are looking for precision angular position you dont need no encoder.
An encoder magnetic or optical can only tell the position or velocity of the rotor. However the controller has still no idea where the magnets, the poles are located relative to the slats , the coils so it cannot deliver phase current properly.
To achieve rotation where cogging wouldnt happen you need positional knowledge of the poles not the rotor.
The driver cannot correct for cogging based on positional feedback of the rotor.
Once you know how often you "commutate" or how many times your phase current passed over the 0 crossing point you also have a relative and accurate velocity measurement. Instead of rpms however you get rpm in the frequency of commutation divided by the number of poles.
(24 poles commutate at the rate of 4.5Hz : 24/3=8
4.5/8=0.5625 rps=33.75rpm.)
Unless you are looking for precision angular position you dont need no encoder.
