Alt-BEAM Archive

Message #04350



To: beam@sgiblab.sgi.com
From: Wouter Brok w.j.m.brok@stud.tue.nl
Date: Tue, 08 Jun 1999 13:40:39 +0200
Subject: [alt-beam] Re: Bicore wasting energy; Power-smart head


Hello Evaristo,

Thanks for you comments,

>I looked at your schematic and I'm wandering about the low-pass filter. In
>theory it will take R3*C3 to charge the capacitor up to 60% of its complete
>charge which is about the trigger level of the HC240 (@5V Vih=3.15 volts).
>If the duty-cycle time of 50% is equal to the RC time then you can detect
>the difference between more or less then 50%. This will only work if the
>50% duty-cycle time is equal to the RC time. But unfortunate the cycle time
>isn't constant but rather dynamic. I'm afraid that your schematic will only
>work under certain light conditions.

The original idea I had was that R3*C3 is much bigger than the slowest
period of oscillation of the suspended bicore, thus that for example the
frequency of the bicore is in the order of about 1 kHz and that the
timeconstant of the low-pass R3*C3 = 0.1 s or something like that.
This way you'll get a fairly nice DC-voltage at the input of U3, which
little ripple on top of it.

>Next. Suppose that the pulses coming from U1 are to short to charge C3. The
>output of U3 will remain high. As a result, U4 will be active all the time.
>The PWM output will show the inverted signal from U2. Now the pulses from
>U1 are becoming longer and C3 can be charged enough to trigger U3. U4 will
>be deactivated and PWM shows the signal from U2. When U1 is active, U2 is
>non-active (output low). PWM will be low. When U1 becomes inactive (low
>output) then the output of U3 will become high shortly after C3 has
>discharged itself enough. As a result U4 becomes active and PWM shows the
>inverted output of U2. Result: once the pulses coming from U1 pass through
>the low-pass filter, PWM will remain low permanent not effected by U2.

It doesn't matter if the pulses coming from U1 are short ... when switching
the power on it will take a while (1 s for example; i.e. 1000 pulses my
guess is) for C3 to be charged so that the input of U3 feels the average of
the output of U1. I think this is were we disagree or where you
misunderstood what I meant (sorry, I didn't put any values to the schematic).
What you describe is correct if the time-constant of the bicore is
approximately the same to the timeconstant of the low-pass before U3

How I saw it (a bit too ideal I think though):

R3*C3 is much slower than the bicore, as a result U3 will feel the average
of the oscillation (smaller than Vcc/2 if the dutycycle is smaller than 50%
and vice versa). In this case dutycycleU1 < 50% the output of U3 is high
and U4 will not be active: PWM is the output of U2 (which has a dutycycle
bigger than 50% because it's the inverted signal of U1). The dutycycle of
PWM will get bigger if the relative light-intensities on the photodiodes
are more different, thus if the thing is far from equilibrium-position. So
the further from this equilibrium, the bigger the duty-cycle, the more
power to the motor.
In the other case, if dutycycleU1 > 50% the output of U3 will be low, U4
will invert, thus the signal of U2 is inverted and PWM is thus equal to the
output of U1. Again, the further from equilibrium, the more power to the
motor.

However, the problem is when the input of U3 is in the linear region of the
inverter ... weird stuff will happen I guess and it won't work. I've been
thinking about using some feefback to speed up the process of changing
state in U3 ... don't know how yet.

>Now this could be handy because no PWM can function as a turn/stop output.
>You won't need the op-amp anymore. Unfortunate we still have the first
>problem. Still more brainstorming needed.

Still busy with that

>Alternative comparator (not tested yet):

>Look at the attached picture. This circuit can work as a voltage comparator
>using the trigger levels of the inverters. The "out" output will only be
>high when the input voltage floats around the trigger levels of the
>inverters. This schematics works due to the fact that no inverter input has
>the same threshold level. The output will probably oscillate a bit once the
>input voltage reaches the trigger point because the input voltage to the
>inverters are floating around the trigger points.
>
>Wilf has the photo diodes in series making a voltage divider. Comparing a
>voltage is easier then comparing the duty cycle of a dynamic pulse
>generator. Yeah a bicore is a simple but yet tough circuit.

That's why I wanted to average the dutycycle to obtain a voltage ....

I think your circuit will work .... but there is an XOR in it, which I
didn't want to use, although it doesn't matter, an extra chip is needed
anyway, so if it's a XOR or a comparator doesn't matter.

Thanks for have a critical look at the circuit; what do you think now I
explained a bit more about the low-pass.

Regards,

Wouter Brok.

------------------------------------------------------------------------

eGroups.com home: http://www.egroups.com/group/alt-beam
http://www.egroups.com - Simplifying group communications



Home