[sdiy] Circle Machine prototype

[David Brown]

At 01:15 AM 8/16/2010, cheater cheater wrote:
>This is absolutely extremely cool. Can't wait to hear some sound demos.
>
>Several comments and questions:
>
>1. you mention interference between the lamps. Try adding small
>matte-painted black walls between the separate lamps. Even better if
>you are able to connect them all together and move them in and out -
>then you can change how separate the lamps are.
>
>2. different solution: take small plastic or sheet metal 'pipes' into
>which you insert the lamps. Make sure those pipes are at the top
>separated from the sensor by maybe 5 mm, and make sure the light
>source in the lamp is maybe 1-2 cm below the top of the pipe. Then you
>should get fairly good separation. Additionally you can move the lamps
>in and out to change the separation

All those are good suggestions.  I think the light is reflecting off 
the bottom vane.  I really don't need that vane as it just was a 
convenient place to mount the photo resistor.  I should be able to 
remove it and just let the photo resistor hang in free space.  It's 
light enough in mass that it won't be a problem.  The interference is 
pretty minimal but I won't mess with it until I decide if I'm going 
to put a photo transistor there.  I will have to remove the vane as 
there isn't physical clearance.

>3. you don't want too much separation, or the sensor will go dim
>between notes i guess :)

The sensor does go dim between notes.  You can see those as the 
vertical spikes on the last scope image.  The CdS sensor turns off 
faster than it turns on.  This was actually one of my concerns with 
the original design.  At slow speeds, there must have been a bunch of 
dead time if there was good separation between notes.  I don't see 
how you could get a good blending.  It was one of the reasons that I 
chose to use a stepper motor to step between lamps.  I figured I 
wouldn't have to experiment with the best light positioning to get a 
good transition.  Note that these spikes are only about 60 mS wide so 
they are not that objectionable.  A bit of lag eliminated them when I 
processed the output via analog.

>4. have you tried it with audio? I think that could sound like a
>really cool low-pass filter. But again, I'm not really sure what the
>settling time of such a sensor is :)

I'm not really sure what you mean with audio?  All I'm doing is 
generating an output of lamp brightess.  Certainly if I used LEDs I 
could modulate them at audio rates.

>5. How have you made the arm? did you weld it yourself?

The arm and the slip-disc are part of the Hammond vibrato 
scanner.  That's why I chose to use it as a base.  It's a great slip ring.

>6. regarding your hammond vibrato - how did you make the parts there?
>specifically the parts with small plates (i'm not 100% clear on how
>the scanner works)

The Hammond vibrato scanner is an interesting invention.  It is a 16 
capacitor switcher.  Around the ring where the lights are is the 
mounting for 1/2 of the 16 capacitors.  The arm formed the other half 
of the capacitor as it spun around.  It would capacitively couple to 
each fixed part and blend as it moved between them.  There's pictures 
of a scanner module and some theory on my scanner project page 
http://modularsynthesis.com/modules/DJB-scanner/scanner.htm  I didn't 
make any of this mechanism other than the brass parts to mount the 
slip-disc pickup.  That mounts on a cover which would hide all the lights!

>7. Do you think diodes could work just as well?

Diodes - you mean LEDs?  I thought of using those but I didn't want 
to build the circuitry for dimming them.  Incandescent worked fine 
with rheostats.

>8. How fast does it go?

I'm using half-steps so there are 200 per revolution.  I am clocking 
them at 4 mS each, so I can do 1.25 revolutions per second.  I can 
clock at 2 mS but there are some resonances so at specific 
frequencies I miss some steps.  I haven't tried between 2 and 4 mS 
since I decided this was too fast for a sequencer anyway.

I can also use full steps to double the speed.  The problem with full 
steps is they are 3.6 degrees and I can't get very close to the 22.5 
degrees required to read each lamp.  It works fine for continuous 
rotation.  I've written a separate program to do that as I 
experimented with this intially.  Note that at the half step 1.8 
degrees I alternate between 21.6 and 23.4 degrees which is reasonably 
close to the lamps.

>9. Can it go backwards? Have you tried a 'random mode'? :)

I have a direction pin so I can reverse the motor anytime.  That's 
another reason I chose to use a stepper motor and a 
ComputerVoltageSource as the controller.  I will most likely add a 
reverse, forward-back, and random mode.  The problem with random mode 
is keeping a constant rate for the sequence.  A 15 lamp rotation 
takes .71 seconds.  A single lamp rotation takes 0.048 seconds.  You 
would want to incorporate the rotational time into the per-step delay 
to hold a constant note rate.  This is another reason why I chose to 
sample the output as there is a lot of dead time with multiple lamp rotations.

>10. it's a bummer you can't get pitch perfect notes from it, but then
>again, maybe that makes it that much better for other uses :) That's
>not a reason to give up, much rather a reason to go on and extract
>that feature imo :) you've got something really cool on your hands :)

I can get perfect notes from it.  I just can't tune it statically and 
then rotate it.  I have to tune while rotating.  It's easy since I 
display the note on the LCD.  At the faster rates where it detunes, 
it's too fast to use for a sequencer as the notes just all mash 
together (to my ears).  Again, it is repeatable so you just have to 
tune it at the rotational rate.  I think a photo transistor would 
work much better.

Dave


>Cheers,
>D.
>
>On Mon, Aug 16, 2010 at 07:13, David Brown <davebr at earthlink.net> wrote:
> > I'm sure it did.  I wasn't trying to build a faithful recreation - just one
> > using lights and a spinner. It seemed like a stepper motor would be easier
> > than a slow variable speed.  Besides, it gave me a chance to play with a
> > stepper motor.  - Dave
> >
> > At 09:54 PM 8/15/2010, David Griffith wrote:
> >>
> >> On Sun, 15 Aug 2010, David Brown wrote:
> >>
> >>> I've been working on building a working model of Raymond Scott's circle
> >>> machine.  I thought it would be fun to build something with motors and
> >>> lamps.
> >>
> >> [snip]
> >>
> >> I was under the impression that the Circle Machine didn't use a stepper,
> >> but instead a more mundane variable-speed motor and didn't have a skip
> >> feature.
> >> --
> >> David Griffith
> >> dgriffi at cs.csubak.edu