Re: [CAD_CAM_EDM_DRO] To convert Quadrature Encoder signals to Tachogenerator signals

Bot Great wrote:

> I assume that you have two square waves (in
>quadrature) coming from your encoder and you want an
>analog signal proportional to the frequency of the
>square waves which is proportional to the rotation
>speed of the encoder shaft. What you are looking for
>is a frequency to voltage converter. Analog Devices
>has an IC, the AD650, which will do what you want.
>

I believe the OP needed a bipolar output, as he was going to
be using this with a 4-quadrant axis servo. Also, to get the
best performance, it needs to sense all transitions of the
encoder, otherwise the axis could drift up to two encoder
counts before movement is sensed. So, it would take a
quadrature to step/direction converter (pretty well-known
circuit), the F-to-V converter, and a linear inverter to flip
the polarity of the velocity signal when the direction is
negative.

The whole problem with this approach is that to produce
a smooth velocity signal at constant speed, there MUST,
mathematically, be delay in the system. The servo drive
was constructed to not expect ANY delay in the tachometer
signal, as a DC tachometer is totally free of delay, except
that which might be caused by a filter to remove brush noise.
So, the servo drive will, absolutely, require re-tuning after
an encoder to velocity converter is installed.

The major machine tool builders use extremely high resolution
encoders, often 100,000 counts/inch and above, to combat this
problem. This greatly reduces the interval between encoder
pulses even at low velocity, reducing the need for smoothing
delay.

Just to give some data points, I have a true velocity servo on
my mill, with DC tachometers feeding a traditional 3-loop
control system. (Position, velocity and torque) I tested it to
see where the motion broke down into stepwise movement, and
it came out about .01 IPM. You have no sense that it is
actually moving except for the screen position display. This
worked out to about 3 encoder counts/second. The tach output
was calculated to be 7 micro-volts (no way I could measure such
a small signal). But, if I took a screwdriver and shorted out the
tachometer, the axis started to vibrate. So, the differential amps
in the servo amplifier were accurately measuring the techometer
feedback, and keeping the loop closed.

Jon