perm filename ORIN1.VL[1,VDS] blob
sn#079507 filedate 1975-02-26 generic text, type T, neo UTF8
00100 COMMENT ⊗ VALID 00002 PAGES
00200 RECORD PAGE DESCRIPTION
00300 00001 00001
00400 00002 00002 \\M1BDR25\M2NGR40\M3NGR25\M4NGR20\F2\CSTANFORD UNIVERSITY
00500 00003 ENDMK
00600 ⊗;
00100 \\M1BDR25;\M2SIGN57;\M3NGR25;\M4NGR20;\F2\CVICARM
00200 \F3\C154 EAST DANA STREET
00300 \CMOUNTAIN VIEW, CA. 94041
00400 \F4\←L\-R\/'7;\+R\→.\→S Telephone:
00500 \←S\→.415-965-0557
00600 \F1\CDecember 18,1973
00700
00800
00900
01000 Mr. David Orin
01100 Department of Electrical Engineering
01200 Ohio State University
01300 Columbus, Ohio
01400
01500
01600 Dear David:
01700
01800 \J Enclosed you will find some of the drawings you requested. I
01900 do not yet have any detailed drawing showing all the internal parts.
02000 Nor do I yet have a really good assembly drawing showing how you
02100 maintain the arm, and just how everything goes together and how it
02200 all works. These I am working on, and expect to be able to send you
02300 copies of these drawings shortly. In the meantime, I hope a brief
02400 explaination will suffice.
02500
02600 The arm is powered by permanent magnet type d.c. electric
02700 motors. A two stage spur gear train connects the motor shaft to each
02800 of the six rotary joints. The gear ratios are in the range of 30 to
02900 50/1, depending on the joint. A brake on each of 5 joints allows the
03000 joint to be locked in a fixed position without applying motor
03100 current. Joint position feedback is from potentiometer elements
03200 mounted on each joint, directly attached to the joint axis of
03300 rotation. Velocity feedback is done electronically by measuring motor
03400 back EMF, or by optional tachometers mounted on each joint. The full
03500 scale drawings show just what the overall size of the arm is, and an
03600 idea of the useful working space can be obtained by assuming that
03700 each joint has a minimum of 300 degrees of rotational freedom.
03800
03900 A power supply is included which contains the motor
04000 power amplifiers and the manual control electronics. The manual
04100 controller allows the arm to be positioned remotely by selecting a
04200 joint and then moving the joint at a slow speed. Only one joint at a
04300 time can be selected, as the manual controller is not intended to be
04400 a general purpose type of controller. As all joints of the arm are
04500 "free", meaning that they can be moved around by hand merely by
04600 pushing on them, the arm can be manually positioned just by grasping
04700 it and leading it to a desired position.
04800
04900 The electronics package is designed such that a dac signal
05000 from the computer can set the torque on each of the joints. We
05100 normally use an 8 bit dac for each joint, but maybe you can get by
05200 with less. The a-d interface is not included. We use a 12 bit
05300 interface here, with lots and lots of channels. You need a minimum
05400 of seven channels for the arm joints and the simple hand, plus, if
05500 you want to read motor back EMF, or the tachometer output, another 7
05600 channels are necessary. Thus, a minimum of a 16 way multiplexer is
05700 required. You can get by with less than 12 bits but resolution will
05800 go down, unless you read in multiple samples. The a-d range can be
05900 whatever you have, as the potentiometer power supply is not included.
06000 This is normally a precision supply of typically 10 vdc at 100ma. The
06100 required precision can be lessened by reading the actual supply
06200 voltage with a spare a-d channel, or else using this supply as some
06300 sort of reference.
06400
06500 With a PDP-8, I think you are going to have some
06600 trouble using the computer both for the servo controller and the
06700 trajectory programmer. I have sold an arm to the Stanford Research
06800 Institute (SRI- an independent corporation not affiliated with
06900 Stanford Univ. anymore, but still very close). They are working on a
07000 hardware servo system which will take a lot of the workload off of
07100 the computer. This way, the computer can do planning and trajectory
07200 calculation and the hardware will handle much of the real time
07300 servoing. The fellows at SRI have promised to let me use all the
07400 designs they develop, and I hope to keep people like you posted on
07500 their progress as it is very closely related to your wants and those
07600 of many other people. Enclosed is Richard Paul's thesis which
07700 describes the computer servo and planning programs we use- they are
07800 basically the same for any arm, only the constants change.
07900
08000 The basic philosophy of the arm is to have a
08100 relatively low cost device which is easily moved about and can
08200 perform simple human type tasks. Thus, the arm can be clamped to an
08300 office desk top, or similar surface. The electronics package is a
08400 small box which sits on the floor under the desk, or on the table
08500 near the arm. A single 3-m flat cable runs to the computer i-o
08600 interface.
08700
08800 As for the nitty-gritty. I originally wanted to sell this
08900 arm at cost, for about $4000, but it now looks like the break even
09000 price is somewhere around $5000, especially since everything is
09100 inflating like crazy these days. Thus, I must price the arm at $5000
09200 for the arm, with a simple hand, power supply, amplifiers, driver
09300 electronics, and a simple manual controller. Delivery is about 16
09400 weeks.
09500
09600 I hope this answers some of your questions. Let me know if I
09700 can be of help in any other way. By the way, it looks like I will be
09800 visiting in Cincinatti around the middle of January. I may have time
09900 to stop by- possibly with a completed arm.\.
10400 \←L\→S\←R\-L\/'2;\+L\→L
10500
10600 Yours sincerely,
10700
10800
10900
11000 Victor Scheinman
11100 President
11200
11300 \←S\→L
11400 VDS:pdp10