The Korf Blog

The inside story: our research,
development and opinions

10 January 2023
Turntable Drives. Part V, Direct Drive Torque Ripple
In our previous posts, we have talked about the turntable belt drive and explored its function as a lowpass filter, rejecting the motor's torque ripple. Then, we went through the key problems associated with the belt drive—and some of the solutions.

Then we shifted our attention to the idler drive, and went through its typical troubles.

We talked about direct drive, how it works, what is good about it and what isn't. Today, we will dip our toes into the murky pond of direct drive motor imperfections and errors.

We ended our previous post with a somewhat shocking chart of a direct drive turntable's momentary speed wandering far and wide. Why does this happen? Simplifying things a bit, there are two major influences on momentary speed:
Torque ripple

Remember, one of the key functions of a belt drive is filtering out the torque ripple. And the idler drive allows the heavy flywheel of a platter to filter our the relatively high frequency torque ripple of a fast rotating motor.

With a DD, we are left face to face with the low frequency torque variation of a slowly rotating powerful motor.

The electronics needs to somehow know when to switch the motor coils, and to be sure that the platter is rotating at the prescribed 33 1/3 RPM.

For that, DD motors use one or many feedback loops. And the whole setup isn't nearly as simple or benign as it seems.
Direct Drive Torque Ripple
People often think of electric motors in terms of converting voltage into rotation. What these motors really do is convert current into torque. As no electric motor is a perfectly stable load (its impedance is always in flux), no motor is ever drawing constant current. A variation in current inevitably means a variation in torque. It is called torque ripple.
    All electric motors suffer from torque ripple: cored, coreless, brushed, brushless, synchronous, asynchronous, etc
    Like taxes, torque ripple is unavoidable, but it can be minimised. Simplifying things rather a lot now, we can point out three key sources of torque ripple:

    • Unsuitable drive current
    • Cogging
    • Coils and magnets imperfections
    Drive Waveform
    If we spin the platter by hand, and look at the waveform that the direct drive turntable motor generates—we will see a sine wave (or something very much like it).

    This is the motor's way of telling us what it needs to be driven with. It's easy to oblige, isn't it?

    Waveform of one phase of a simple DD motor driven by hand
    Back in 1970s, generating a clean synchronised sine current was very hard. Most manufacturers used a simple trapezoidal ("on/off") commutation instead.

    This is a bit like ordering ice cream and being served herring.

    Voltage waveform of one phase of a 1970s DD motor controller
    Some manufacturers tried to beautify the ugly trapeze a bit, with varying success. The others (Hitachi, for example) designed the winding to minimise torque ripple with the simple commutation available.

    Now, of course, the situation is completely different. A typical $1 microcontroller has all the power and storage of a 1970s university computer. Using a technique called DDS, a microcontroller can synthesize a perfect waveform to drive those motor coils. There really is no excuse not to.
    In the marketing literature, cogging and torque ripple are deliberately conflated. If you believe the glossy brochures, avoiding cogging automatically gets rid of all the torque ripple. We have already seen that it isn't so—but what really is cogging and how does it contribute to torque ripple?
    Cogging happens when moving motor parts get magnetically attracted to the stationary ones.

    In our example, magnets A attract stator slots B. As you rotate the stator, there will be a moment when the magnet will "jump" from one stator slot to another.

    Coreless motors have no magnetic stator slots, and usually experience no cogging at all. Excellent! But remember, our goal is reduce torque ripple, not get rid of the cogging. How are the two really related?

    Cogging feels bad, but is largely benign. It only is a major factor when the motor is not energised. Once the current is flowing through the coils, magnetic attraction to stator slots becomes a lesser part of the forces acting inside the motor. Knowing it is there, one can adjust the driving current to compensate for this unwanted force.

    For example, stepper motors are cogging personified. If you have ever held one, you know what I mean. But have a look at the surface of the FDM 3D print. Cheap stepper motors position the nozzle that dispenses molten plastic—and yet the print's lines are quite smooth. There's nothing in the print to suggest the heavy resistance you need to overcome rotating the motor through the steps by hand.
    Cogging is only a problem in "freewheeling" motors, and the turntable motors never idle. Yes, if the driving current does not reflect the motor's internal design, then cogging will contribute to the torque ripple. But the closer the driving waveform to the motor "demand", the less cogging matters.

    In direct drive turntable motors, cogging is largely a marketing straw man. It is extremely easy to slay, but the real improvement from its elimination is somewhat marginal.
    First, a dirty little secret: permanent magnets are not precise at all.

    Usually, the more powerful for a given size they are, the less precise and uniform their magnetic field is. This is one of the reasons why, for speaker drives, field coils > alnico > ferrite > rare earth (with exceptions, of course).
    "Magnetic unloading" of a turntable bearing is a terrible idea largely because of permanent magnet's poor field uniformity. It adds random modulation to the load experienced by the bearing.
    If the magnet irregularity wasn't bad enough, the coil tolerance adds to it. Check out any coil supplier and see what tolerance is specified for large-ish inductances. ±10% is considered good, ±20% typical. Yes, it's possible to hand trim coils to 1% accuracy or better, but this will add rather a lot to our motor cost.

    Variations in magnets and coils cause repeating changes in driving current as the motor rotates.

    How Do We Solve Torque Ripple?
    There is a common theme to all three sources of torque ripple. To extinguish them, we need to feed the motor with a beautiful waveform that accounts for our knowledge of this exact motor's quirks.

    We need a method that would constantly adjust the driving current to compensate for the imperfections. As these things usually go, there's a right (but complicated) way of doing this, and a wrong (yet easy to implement) one.

    The wrong way is a feedback loop.

    Please subscribe to receive blog updates in your inbox!
    comments powered by HyperComments