|
Actually, LSE's rpm data is analog from
the get-go. One way to turn one-three pulses per rev to an rpm output is to
have a multiplier and counter. This can be done with a phase-lock-loop
multiplier set to the proper division rate, i.e. if you want a readout every
second, a one-pulse-per rev would have to be multiplied by 60 before being input
to a counter. Then you are faced with the choice of closed-loop bandwidth
that will have a fast enough response time to keep up with motor acceleration,
but slow enough to filter the crankshaft angular-rate variations. You'd be
amazed at how much the crank accelerates and decelerates during one revolution!
I know; I've built several. I guess you might call that a digital approach.
I flew with a fellow that was trying a PLL tach I made on his 140,
that had a one second readout rate. At first he thought there was a problem with
it as it sometimes jumped aroud tens of rpm, until I pointed out it was smooth
until we had some turbulence. If you see stable rpm readout in
your digital display in turbulence, your tach's smoothing constant or
time-base rate is quite large. A u-processor-based system will count
time between pulses or the number of pulses in a fixed-length gate, then
calculate the rpm. That's still a multiplier. Any time you have a harmonic
relation between the pulse rate into a counter and the gate rate, you can have
gate opening and closing errors where pulses are picked up or
dropped.
A lot of tachs start with a triggered
pulse, at crank rate, to give a fixed width and amplitude pulse, then filter
that through a simple R-C smoothing net for display on an analog meter of some
sort, either DVM or D'Arsonval movement. LSE's rpm is 100 uV / rev / min, so if
you do not have a good closed signal system from the LSE to the meter, common
mode noise injected into the wiring can cause errors. The 40-70 rpm error you
mention can be from 4mV to 7mV common mode error in the ground return; that's
0.004V to 0.007V. LSE's MAP is 10mV / inchHg. Again, 6mV, 0.006V, on the ground
can give 0.6" MAP error + or -! Proper wiring and grounding practices, as was
much discussed on LML several weeks ago, is critical with low-level signals
of any kind. Whether digital or analog, each approach has its own type of
errors.
|