perm filename RHODE2.VL[1,VDS] blob sn#157534 filedate 1975-04-29 generic text, type C, neo UTF8
COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	\\M1BDR25\M2SIGN57\M3NGR25\M4NGR20\F2\CVICARM
C00011 ENDMK
C⊗;
\\M1BDR25;\M2SIGN57;\M3NGR25;\M4NGR20;\F2\CVICARM
\F3\C154 EAST DANA STREET
\CMOUNTAIN VIEW, CA. 94041
\F4\←L\-R\/'7;\+R\→.\→S   Telephone:
\←S\→.415-965-0557
\F1\CApril 29,1975



Professor John Birk
Department of Electrical Engineering
University of Rhode Island
Kingston, R.I. 02881


Dear John:

\J
	My apologies for the delay in this reply.  I have been very busy
recently, and have been lax at replying to your letter.  I hope the following
information will suffice for a while.

	You asked several questions, I will try to answer some of them now.

Joint travel-  600 degs. vs.  330 degs.   Most everyone is  using 330
degs.  now although the  original arms at  Stanford had  600 degs. of
travel.  To resolve 600 degs., I add another wiper to the pots.  This
then requires an extra A/D channel. When  one of the wipers is in the
dead  band at the  ends of the  pot element, you switch  to the other
wiper. This  requires some  computer bookkeeping.   In addition,  you
have a  startup problem, as you  dont know which stop  you are near. 
Because of the extra travel,wire failure rate is potentially greater
too.  I've thought of  designing a proper 600 deg. pot,  but have not
gotten  to that yet.   More likely, I will  eventually get to intalling
encoders before fancy pots.  Encoders will resolve  continuously over
any rotation. 

	Yes  you can  damage  the arm  by bumping  into  things.   My
suggestion  is to  put some  styrofoam around  while testing  it out.
Most likely  to fail in  this mode  are the  fingers or hand  slides.
Both are easily replaced.   The arm uses permanent magnet D.C. motors
with  Alnico  5  magnets.  These  magnets  will  become   permanently
demagnetized should  you ever  exceed the  maximum allowable  current
rating.  Remagnetizing  requires dissassembly and is a hassle.  Thus
your power supply  and amplifiers must  have a  current limit
with protection from turnon and turnoff transients, etc. 

	Brakes are used to hold the  arm in position when not moving.
This  allows the motors to be turned  off thus reducing motor heating.
At Stanford we  use a software  servo.  Thus  the computer turns  the
brakes off only  when moving the arm.  At  all other times the motors
are off and the brakes  on.  With a  hardware position servo you  can
probably turn the  brakes on when  the corresponding joint  is within
the error tolerance region.  

	Tachometers  are  used  to  provide  damping  inside  of  the
backlash on the joints.  Yes, you can differentiate the  pots signal,
but its good to have extra motor  damping.  You can do this two ways-
with  tachometers, and  by using  back EMF  to indicate  motor speed.
Both systems  work.   In  fact, I  can deliver  arms  either with  or
without tachs.   The no tach  version costs $75 less per  joint.  For
your wrist thats $225  less as the hand  doesnt have provision for  a
tach. 

	Power amplifiers are not supplied in the  price I quoted you.
You must provide them,  or else I can provide them.  At the present time I
am trying  to come out  with a  power amplifier  with hardware  servo
electronics  all in  one unit.   Price  is $1200  for the  four wrist
joints.  You then interface this to your computer thru an A/D and D/A
interface  ( plus  some  brake  bits).    I can  also  send  you  the
schematics of what we have at Stanford, so you can build it yourself.
The Stanford system is a software  servo based system, and the  power
amplifers acccept current commands from the DACS rather than position
commands.  

	Speeds are based on power supply voltage, and are independent
of  duty cycle.   but  they  are for  unloaded joints.   Basically,  a
motor's speed is a function of the voltage supplied, the back EMF and
the load.   The power supply is nominally about  +-30vdc.  With more,
you can go faster, but power dissipation goes up too.
My power supply keeps track
of motor heating  with  a thermistor  attached  to a  resistor model  of  the
motors.  This cuts out  the supply in the event of motor overheating.
At Stanford we just use software to limit duty cycle. 

	The wrist  unit  is provided  with  8 spare internal wires  for  your
instrumentation.   Extra wires can  be added, but  if they  are bulky
coax, they probably will have to be run outside the arm structure. 

	I have  set a delivery of 24 weeks ARO  because I am always in
the process of improving the arm.  At the present time I  am hesatant
to start on  a new arm order  until I get a little  feedback from the
three systems  I am currently finishing up.  Very possibly, I can get
something to you  sooner.  But from  past experience, I can  only say
that  I am striving  for the best  system I  know of, and  this takes
time.  

	Pricing.  The complete arm with power supply runs around $20,000. 
It costs just as much to make a wrist joint as a shoulder joint.
I am trying to keep the costs as low as possible consistent with quality and
performance.  In addition the volume of these arms is not yet to the point where
I can justify production runs. 

	I will be comming to the East coast in about two weeks.  I'll plan on
calling you and stopping by for some furthur conversation. 


	I hope this information is adequete.\.



\←L\→S\←R\-L\/'2;\+L\→L

Yours sincerely,



Victor Scheinman
    President

\←S\→L
VDS:pdp10