
2E9/(FREF/64).
There are three parameters that are affected by
the choice of FREF:
1)Maximum masking time, Tmax, that can be
calculatedas:
Tmax = 2E6/(FREF/128) = 2E13/FREF
2)Minimum time resolution of the maskcounter,
that is 1 bit or:
Tres = 128/FREF
3)Truncationerror, Et, coming from the appros-
simation caused by the division
will typically generate non integer numbers,
whose decimals will be skipped.This error is
again one bit (with the period of the frequency
used by the down counter) or:
Et = 128/FREF
As a result of the above the maximum error of the
masking time can be up to:
Emax = Tres + Et = 256/FREF
Please note that the truncation error is not fixed,
but depends from the period count and can also
be null if the count between two zero crossings
can be exactly divided by 4.
As an example, we can consider the case of an
8pole, three phases motor rotating at 5400 rpm.
Let consider FREF = 8MHz.
8 poles
→
4 electrical cycles for each mechani-
cal revolution
and
3 phases
→
6 commutationsfor eachelectrical
cycle
therefore:
Commutation Frequency =
1min/60seconds* 24comm/rev= 2160Hz.
This means that the commutation period is about
463 microseconds. Considering the above ex-
pression we will have:
Tmax = 2E13/8E6= 1.024ms
This means that the masking time will be propor-
tional starting from a commutation period lower
than 4Tmax = 4.096ms that means a speed
higher than 610 rpm. Additionallywe will have:
Tres = 128/FREF = 16
μ
s
Emax= 256/FREF= 32
μ
s
With a commutation period of 463 microseconds,
we should have a masking time of 463/4 = 116
μ
s
so that we obtain:
Accuracy = 32/116 = 27.6%
This is the maximum error. Considering the real
situation and mainly the real truncation error we
will have in this particular situation an Emax=20
microseconds so that the accuracy is about
17.3%
by 4, that
5400
rev/min
*
APPLICATION INFORMATION
A typical application configuration of the L6237
driving a three phase brushless sensorless DC
motoris shownin Fig.6.
The spindle motor typically is a 2.5” Rigid Disk
Driver having 1.3
- 0.1mH per phase, star con-
nected.
This kind of load requires a suitable compensa-
tion of the linear control loop thatcan be achieved
by an RC network of 10K and 10nF, connected to
the ”COMP” pin.
Changing the motor characteristics, the RC net-
work could be modified for the best performances
of the system.
This is a suggestion about how to choose the
value of the RC compensationnetwork of the cur-
rent loop: the following figure shows the entire
control
system
of
the
The error amplifier is a transconductance ampli-
fier. It is used in open loop configuration inside
the main control loop and its gain and frequency
current
regulator.
Figure 2:
TypicalNormalized R
DS (on)
vs.
JunctionTemperature.
Figure 3:
TypicalTransient Thermal Impedance
vs. Time or Pulse Width
SINGLEPULSE
L6237
7/11