On Jul 2, 2009, at 6:02 AM, Jones Beene wrote:
Excuse me for jumping in late on this thread, not having followed it
closely, but this may be worth a mention from the peanut gallery
(unless it
has already been covered)...
WRT the current squared hypothesis - there should an obvious way to
falsify,
or to add a level confirmation to this. Unfortunately, if there is
more than
one thing going on, like the heat hypothesis, then the following
may not
tell you much.
That would be to measure the RPM at DC for your baseline and at
various
levels of current and the same voltage. Is the rotational response
linear or
exponential to the current? Even with friction and other losses, it
should
be exponential, no? ... and alternatively, or in addition to that,
compare
against the same setup at 50% duty, square wave, but the same
voltage and
twice the current. In the case of twice the current, over half the
time
interval, the expected proportionality would be 4/2 or double.
Correct?
The implication of that is that very low duty, but very high
current (cap
discharge?) might even make the thing useful... (Unless I am missing
something which is likely) i.e. 1% duty with 100x current pulse
gives an
enticing relative gain ....
The fact the effect is due entirely to hysteresis limits the
effective rpm range across which the motor responds with a force
(torque) proportional to i^2. There has to be a balance of current
to load to optimize the motor efficiency. For a DC test, an
appropriate test would simultaneously increase the load in order to
sustain a constant rpm, and thus maintain the magnetization timing.
There was an inherent assumption on my part, in making the "torque
proportional to i^2" assertion, that the motor was operating in an
efficient range, and as well not saturating.
A sufficient time is required to overcome the magnetization
hysteresis in order to have a sufficient M to produce the i L x B
torque. Similarly, the M field must last long enough in the material
with out the supporting H that it rotates into position such that the
current i passes through it. The combined effect is a kind of wave
of magnetization to both sides of the contact points. Optimization
places the current right in or near the appropriate peak of that wave.
If pulsed DC is used, and the pulse of current is too fast, and the
time between pulses too long, then the initial magnetization will
occur, and possibly even saturate the material, but by the time the
magnetized material rotates into the contact point location, there is
no current with which to generate the i L x B force, thus the motor
will have no torque at all.
I would note that, under the thermal scenario, the heat (energy)
applied is an i^2 R effect, where R is the resistance. However, to
maintain a constant torque for a given current, the same degree of
expansion has to be maintained at every rpm. Therefore the power
requirements must increase with angular velocity. The energy to
support, via thermal expansion, the extreme speeds at which some of
the motors now operate should take an extreme amount of power. As
the designs improve magnetically, you can see the power required
drops, the current required drops, and the zero load to angular
velocity and the initial acceleration both increase dramatically.
In any case, I maintain that a Marinov ball bearing motor made
entirely of non-magnetic material will quickly resolve the thermal vs
magnetic explanations. A complex FEA dynamic model would be required
to optimize the design, or to verify the theory quantitatively, i.e.
perfectly match theory to performance.
Best regards,
Horace Heffner
http://www.mtaonline.net/~hheffner/
