Well, the way the data is presented looks like this is static but there’s a lot going on here. Klippel uses a statistical model of the driver under power (using voltage and current sensing, microphone and laser) to derive the root causes of distortion and present you sort of composite graphs.
For instance, the inductance shown is a lumped parameter and not able to be split out by frequency (a kind of limitation of the system). Things like shorting rings have a frequency dependent effect and so, if you go by this curve alone, you can make some incorrect assumptions about how much metal to use and where (based on the intended frequency range of the driver)
During testing, this driver saw in excess of 110V and 30 amps of current. We use a split gap XBL type design for our coil/top plates which gives you some of the best qualities of both underhung and overhung magnet structures, low coil mass, good inductance behavior (large shorting ring location at the rest position).
Woofer motors, particularly at very high current, modulate the flux in the motor some and the BL curve will dynamically skew to either side as the coil is essentially adding and subtracting from the permanent field. We are doing some deep undercuts to push the steel into saturation (where it’s hysteresis curve is flatter) and using multiple thick faraday rings (which counter this effect through an induced current/magnetic field). Part of this is a measurement artifact of pushing the driver so hard to 1.5" peak to peak) to resolve the curve shape out at the extremes but we are seeing some modulation of the upper magnetic gap.
Asymmetry here does effect 2nd harmonic distortion but we’re only varying a little and overall the BL is very flat and extended (and is +/- 9.6 or ~12mm depending on it you’re looking at the 82% or 70% limit that defines “Xmax”) and you can see in the “distortion analysis” window that the distortion from BL isn’t the primary contributor until the end of the test where they are pushing things to the ~1.5" peak-to-peak physical limit of the unit. Basically, at typical current levels, it’s flatter.
There are a few other techniques to improve this a bit more like bucking magnets, changes to the pole extension geometry and additional return paths, pre-charged magnets in place of some of the steel, more shorting rings. I may do some future R&D on some of these things but assembling a motor with pre-charged magnets is dangerous and costly and we have to make decisions on how much to spend and where to put our money.
Anyway, I was happy with where we ended up on this and it isn’t as simple as a little longer pole or something improving things further.
Transducers are very complex and we’re doing a lot to simulate, measure, iterate and optimize each design.
