0% found this document useful (0 votes)
20 views21 pages

Error Budgeting in Machine Design

The document outlines a deterministic design approach for machines using error budget methods, emphasizing the importance of error apportionment and budgeting in the design process. It describes the steps for assessing performance requirements and developing concepts, including the use of spreadsheets for error analysis. The document also highlights the significance of early-stage design and creativity in machine development.
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
20 views21 pages

Error Budgeting in Machine Design

The document outlines a deterministic design approach for machines using error budget methods, emphasizing the importance of error apportionment and budgeting in the design process. It describes the steps for assessing performance requirements and developing concepts, including the use of spreadsheets for error analysis. The document also highlights the significance of early-stage design and creativity in machine development.
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

©

 Alexander  Slocum,  last  update  2016.03.27  

Deterministic  Design  of  Machines  using  Error  Budget  Methods    


 
In  the  process  of  designing  a  new  machine,  the  designer  typically  starts  with  ideas,  
sketches,  quick  calculations  (at  least  first  order  guestimates)  of  feasibility,  and  then  gets  
ready  to  start  investigating  ideas  in  more  detail.    However,  before  investing  a  lot  of  time  in  
solid  models  and  finite  element  analysis,  it  can  be  helpful  to  more  deeply  consider  in  a  
deterministic  way  the  performance  of  different  concepts  either  wholly  or  partially  (e.g.,  a  
particular  axis).    The  technique  of  first  order  error  apportionment  among  axes  and  
components  followed  by  more  detailed  error  budgeting  can  be  helpful  to  investigate  design  
ideas  and  options.  
 
First  step:    Error  apportionment  
 
The  designer  needs  to  have  an  idea  of  the  number  of  axes  needed  to  accomplish  the  task,  
and  the  overall  accuracy  (or  repeatability  )  of  the  machine.    Assuming  no  particular  axis  
should  “work  harder”  than  another  to  deliver  performance,  the  total  allowable  error  can  be  
apportioned  amongst  axes  and  components  within  axes.    This  can  be  done  with  the  
spreadsheet    Axis_error_apportionment_estimator.xls  
 

 
 
We  now  have  an  idea  of  the  dimensional  performance  requirements  (e.g.,  accuracy  or  
repeatability)  for  the  primary  elements  in  the  system.    When  very  high  performance  is  
required  where  it  seems  like  we  will  have  to  work  very  hard  to  meet  the  requirements,  The  
next  step  is  to  assign  coordinate  systems  and  then  to  estimate  the  error  motions  in  the  
elements  and  stiffnesses  of  connections  (e.g.,  bolted  joints,  bearing  elements)  between  
coordinate  systems  and  the  compliance  of  structural  elements.  
 
Second  step:    Concept  Development  with  the  aid  of  an  Error  Budget  
 
 

  1  
©  Alexander  Slocum,  last  update  2016.03.27  

An  error  budget  can  be  used  to  first  assess  just  geometric  errors,  and  the  designer  can  
assume  if  geometric  errors  can  be  managed  according  to  limits  allocated  in  the  error  
apportionment  phase,  then  the  load-­‐induced  errors  could  also  be  dealt  with  by  appropriate  
component  sizing.    The  designer  can  then  proceed  using  simple  calculations  to  size  
elements  to  solid  model  the  concept,  and  then  use  FEA  to  predict  deflections  and  make  sue  
they  are  within  the  limits  set  by  the  error  apportionment  phase.    This  requires  an  iterative  
process  and  some  hope  that  when  the  entire  machine  is  assembled  all  will  be  well.      
 
A  more  deterministic  approach  is  to  do  a  second  round  or  error  budgeting  where  applied  
loads  and  the  stiffness  of  joints  and  bearings  and  compliance  of  structural  elements  are  
added  to  the  model.    This  discrete  element  model  effectively  considers  the  net  effect  of  
geometric  errors  and  load  induced  deflections  to  predict  overall  machine  performance.    
Being  spreadsheet  based,  the  designer  can  rapidly  spot  trends  and  adjust  member  and  
bearing  sizes  (stiffness)  accordingly  to  converge  on  a  workable  design.    THEN  the  design  
can  be  solid  modeled  and  checked  with  FEA  against  the  and  the  model  considers    
 
To  illustrate  the  latter  approach,  this  document  will  guide  users  through  using  error  
apportionment  and  error  budget  spreadsheets  by  applying  them  to  a  bridge  style  machine  
intended  for  use  as  a  wood  router  in  a  home  shop.  
 
Where  to  start?  
 
A  designer  often  starts  at  the  tool  and  works  backwards  to  the  base.    The  rational  is  that  the  
designer  should  have  an  idea  of  the  stiffest  tool  that  will  be  used  and  the  power  and  speed  
of  the  spindle.    
 
Figure  1  shows  some  stick  figure  concepts  for  the  to-­‐be-­‐designed  router.    Keeping  early  
stage  designs  in  the  hand-­‐drawn  phase  enables  the  designer  to  spend  more  time  with  
young  family  members  creating  concepts  in  Crayon,  as  shown  in  Figure  2,  as  well  as  
preventing  business  overlords  from  assuming  the  design  is  finished  and  so  they  start  taking  
orders…  It  also  helps  the  designer  to  relax  and  more  freely  think  of  design  options…  
Drawing  animals  and  fun  shapes  within  a  design  helps  keep  young  helpers  happily  engaged  
and  promotes  free-­‐spirited  creative  thought  in  the  designer.    It  is  like  a  dither  signal  that  is  
sometimes  used  to  help  a  servo  overcome  friction.      
 
 

  2  
©  Alexander  Slocum,  last  update  2016.03.27  

 
Figure  1:  Early  Stick  Figure  concepts  for  routers  as  motivated  by  consideration  of  
Functional  Requirements,  Design  Parameters,  Analysis,  References,  Risks,  &  
Countermeasures  (FRDPARRC1)  
 

 
Figure  2:  Crayons,  big  sheets  of  paper,  blocks…,  and  a  young  helper  do  wonders  for  
creativity  and  happiness  of  all  involved.  
 
 
                                                                                                               
1  Big  Awesome  (J)  Design  acronym  (created  to  make  fun  of  TMS-­‐TLAs).    See  for  example  

Graham,  M,  Slocum,  A,  Moreno  Sanchez  R.,  “Teaching  high  school  students  and  college  
freshman  product  development  by  Deterministic  Design  with  PREP”,  ASME  Journal  of  
Mechanical  Design  (Special  Issue  on  Design  Engineering  Education),  July  2007,  Vol.  129,  pp  
677-­‐681.  

  3  
©  Alexander  Slocum,  last  update  2016.03.27  

Given  that  a  full  sheet  of  plywood  is  ideally  to  be  cut,  a  bridge  style  machine  is  selected  as  
the  approach  to  first  consider.  Even  in  the  stick  figure  phase,  it  is  a  good  idea  to  start  to  
think  about  coordinate  systems  that  will  represent  the  major  elements  of  the  machine.    The  
best  way  to  assign  coordinate  systems  is  to  think  of  the  centers  of  stiffness  of  the  system,  
particularly  where  the  bearings  attach,  which  are  simply  represented  in  the  stick  figure.  
Start  simple  and  think  in  terms  of  cantilever  and  simply  supported  beams,  so  you  might  
have  more  coordinate  systems  than  moving  degrees  of  freedom.    The  bridge  concept  path  
that  has  been  decided  upon  can  be  envisioned  as  being  modeled  with  7  Coordinate  
Systems:  
1. Tool  in  Tool  Holder  
2. Spindle  
3. Z  axis    
4. Y  axis    
5. Vertical  Riser  
6. X  axis  
7. Cutting  point  on  workpiece  
 
As  discussed  in  more  detail  below,  every  Coordinate  System  N  has  an  origin  to  where  it  will  
attach  to  the  next  Coordinate  System  N+1,  and  a  Point  Of  Interest    (POI)  to  which  the  
previous  Coordinate  System  (N-­‐1)  is  attached.  
 
Figure  3  shows  a  first  pass  isometric  view,  still  hand-­‐drawn  to  prevent  too  early  lock-­‐in  to  
design  details  as  can  be  the  case  when  solid  models  are  too  early  created  in  the  design  
phase.      
 

  4  
©  Alexander  Slocum,  last  update  2016.03.27  

 
Figure  3:  Whiteboard  sketched  bridge  machine  concepts  for  a  router  (created  at  white  
board  based  on  inspiration  from  early  sketches  created  as  in  Figure  2).2  
 
As  the  bridge  structure  in  Figure  3  was  sketched,  the  designer  started  thinking  about  how  it  
might  be  built  and  thus  also  modeled:  where  should  coordinate  systems  be  placed?    Indeed,  
many  corrections  and  adjustments  were  rapidly  made  along  the  way  using  an  eraser  and  
markers.  A  big  DO  NOT  ERASE  note  ensured  that  the  design  could  remain  for  several  days  
so  it  could  mature  with  the  help  of  peer  review  by  colleagues  who  wandered  into  the  office  
were  very  valuable.    Sticky  notes  were  sometimes  added  and  used  to  capture  ideas  that  

                                                                                                               
2  Can  you  find  all  the  smiley  faces?    Ideally  there  would  also  be  small  wild  animals  hidden  

within.    Staying  a  little  silly  helps  keep  the  designer  free  thinking  so  as  to  not  get  too  
enamored,  and  hence  resistant  to  change,  with  their  own  design  too  early.    

  5  
©  Alexander  Slocum,  last  update  2016.03.27  

were  not  quite  ready  to  replace  existing  lines.    The  process  was  catalyzed  and  lubricated  by  
lunches  and  coffee  breaks.    Throughout,  the  process  enabled  thinking  about  coordinate  
systems  that  represent  motion  axes  and  connections  between  major  structural  elements.    
 
The  entire  process  shows  that  it  is  good  to  use  a  whiteboard  for  early  development.    This  
“fuzziness”  is  an  important  part  of  the  creative  phase.    Analysis  (as  in  equations  and  also  
critical  thinking  about  obtaining  components,  manufacturing…)  can  be  thought  of  as  a  lens  
that  helps  bring  a  design  into  focus.  
 
A  note  about  assigning  coordinate  systems,  which  will  be  discussed  in  more  detail  below.    
Every  Coordinate  System  (CS)  has  a  Point  Of  Interest  (POI)  and  a  Point  Of  Attachment  
(POA),  which  by  convention  here  is  at  the  origin  of  the  CS.    Accordingly,  one  CS’s  origin  
(POA)  is  located  at  another’s  POI.    Meanwhile,  errors  in  machines  are  dominated  by  
systematic  errors  (e.g.,  axes  are  not  bolted  together  exactly  square,  or  repeatable  
straightness  error  in  an  axis),  random  errors  (e.g.,  bearing  noise),  and  deflections.    The  
latter  are  due  to  limited  stiffness  at  joints  and  bearings  and  compliance  in  structures.  
 
It  is  thus  logical  to  follow  the  structural  loop  of  the  machine,  which  traces  the  path  of  forces  
through  the  machine  from  toolpoint  back  through  the  machine  and  up  into  the  workpiece  
where  the  tool  is  acting,  and  place  a  CS’s  POA  at  a  connection  between  major  elements.    
Hence  each  CS’s  POI  will  be  collocated  with  the  next  CS’s  origin  which  is  at  a  connection  
between  major  elements…    The  stiffness  of  the  connection  between  coordinate  systems  can  
thus  be  represented  by  springs  between  one  CS’s  POA  and  its  adjoining  CS’s  POI.    Within  a  
coordinate  system,  the  compliance  of  the  structure  that  connects  the  POA  to  the  POI  can  
then  be  represented  by  a  full  (6x6)  compliance  matrix  as  discussed  below.    Figuratively  the  
sequence  is  shown  in  Figure  4.  
 

 
Figure  4:  Modeling  a  machine  as  a  sequence  of  compliant  beams  and  discrete  spring  
connections.  
 

  6  
©  Alexander  Slocum,  last  update  2016.03.27  

Once  the  overall  concept  has  mostly  congealed,  different  views,  close  ups  and  cross  
sections  of  regions  can  be  sketched  to  help  the  designer  better  understand  as  shown  for  
example  in  Figures  5  and  6.    One  method  of  design  synthesis  is  to  start  at  the  tool  tip,  
because  that’s  where  the  action  is,  and  keeping  in  mind  the  forces  and  accuracy  desired,  
create  elements  along  the  structural  loop  to  enable  the  desired  motions  to  occur,  and  
conceptually  size  them  in  accordance  with  their  distance  from  the  tool  point.    This  is  the  
phase  where  one  moves  from  a  stick  figure  to  at  least  quasi  reality  J.  
 
Following  this  logic,  consider  the  machine  in  Figure  3  and  the  cross  section  in  Figure  5.  
CS_1’s  origin  (POA)  is  located  at  the  center  of  the  collet  that  holds  the  tool.    The  POI  is  the  
tool  tip  and  the  compliance  of  the  tool  (1/stiffness)  can  be  easily  determined.    The  stiffness  
of  the  tool  held  in  the  collet  and  the  collet  in  the  tapered  receiver  of  the  spindle  shaft  would  
be  modeled  as  a  set  of  linear  and  rotary  springs  where  resultant  forces  and  moments  on  the  
CS’s  origin  cause  linear  and  angular  displacements  respectively.    This  might  require  
experimentation  or  a  literature  search  as  surely  this  has  been  tested  before.      
 

 
Figure  5:  Whiteboard  sketched  cross  section  of  the  tool,  collet,  and  spindle  on  the  end  of  
the  Z-­‐axis  structure.  Note  Y  axis  is  positive  into  the  page  so  a  positive  X  direction  force    
would  cause  a  negative  rotation  about  the  Y  axis.  
 
CS_2  needs  to  be  able  to  model  the  stiffness  of  the  spindle  bearings  as  well  as  the  spindle  
shaft.    If  it  were  located  at  one  bearing  or  the  other,  the  angular  error  created  by  the  
resultant  moment  from  the  force  at  the  tool  point  may  not  be  completely  captured.    Yet  the  
spindle  shaft  may  also  be  a  significant  source  of  compliance.    The  joint  between  the  spindle  
housing  and  end  of  the  Z-­‐Axis  structure  is  likely  to  be  far  more  rigid  than  the  spindle  
bearings,  and  hence  we  will  focus  our  efforts  on  the  spindle  bearings.  
 
Placing  CS_2  at  the  center  of  the  spindle  shaft,  and  modeling  its  attachment  stiffness  to  the  
POI  of  CS_3  (the  end  of  the  Z-­‐Axis)  as  the  linear  and  angular  (pitch  and  yaw)  stiffness  of  the  
spindle  bearings  and  housing  will  accurately  reflect  the  effect  spindle  bearing  deflections  
have  on  the  motion  of  the  tool  tip.  

  7  
©  Alexander  Slocum,  last  update  2016.03.27  

 
But  what  of  the  deflections  of  the  spindle  shaft,  the  compliant  structure  connecting  CS_2’s  
POA  to  its  POI  which  is  the  origin  of  CS_1?    The  spindle  shaft  conservatively  looks  like  a  
simply  supported  beam  unless  it  is  a  slow  speed  spindle  with  a  significant  set  of  back  to  
back  mounted  rolling  element  bearings  at  its  front.    In  either  case,  linear  deflection  of  the  
simply  supported  beam  is  not  significant  because  CS_1  is  considered    located  at  a  support  
point  (the  front  spindle  bearing)  but  angular  deflections  are  important:    The  slope  
deflection  of  the  spindle  shaft  at  the  POI  directly  affects  angular  deflection  and  hence  Abbe  
error  of  the  tool  tip.    Since  the  relative  position  of  CS_2  and  CS_1  does  not  change  we  can  
locate  CS_2  at  the  center  of  the  spindle  shaft,  and  model  the  compliant  beam  that  connects  
CS_2’s  POA  to  its  POI  as  a  simply  supported  beam  whose  length  is  that  of  the  entire  spindle  
shaft.    
 
Moving  on  to  the  rest  of  the  system,  consider  Figure  3  and  Figure  6  which  shows  a  side  
view  of  the  machine,  where  it  should  be  clear  that  CS_3’s  origin  is  at  the  Center  of  Stiffness  
(COS)  of  the  Z  bearings  (e.g.,  two  linear  guide  (profile  rail)  cartridges  (generically  called  
bearing  blocks)  on  each  of  two  profile  rails).    The  stiffness  of  these  bearings  with  the  local  
structure  will  dominate  the  attachment  stiffness  of  the  CS_3’s  POA  to  CS_4’s  POI.    The  
compliance  between  CS_3’s  POA  and  POI  will  be  that  of  the  Z-­‐Axis    structure,  sometimes  
called  the  ram,  which  can  be  modeled  as  a  cantilever  beam,  whose  length  changes  in  
accordance  with  the  position  of  the  spindle,  which  is  controlled  by  the  Z-­‐Axis  actuator.    
That  Z-­‐Axis  actuator  will  thus  be  the  Z-­‐Axis  compliance  value  in  the  {6x6}  compliance  
matrix,  where  the  other  entries  are  due  to  those  of  a  cantilever  beam.  
 
Note  in  this  design  the  Z-­‐Axis  bearing  blocks  are  attached  to  a  chunky  structure  to  which  
the  Y-­‐Axis  bearing  blocks  are  also  attached.    In  this  case  CS_3’s  POA  (origin)  is  attached  to  
CS_4’s  POI  and  there  really  is  no  compliant  structure  between  the  two.    Hence  the  
compliance  matrix  for  CS_4’s  “structure”  would  just  be  the  null  matrix  of  zeros3.    

                                                                                                               
3  In  a  spreadsheet  this  is  easier  to  do,  because  old  values  can  be  deleted  and  the  default  

value  is  “0”  in  subsequent  calculations  that  reference  the  cells.    Else  if  the  analysis  was  
based  on  using  a  stiffness  matrix  (instead  of  the  compliance  matrix)  the  user  would  have  to  
delete  values  and  enter  1E12  or  some  such  super  huge  value  

  8  
©  Alexander  Slocum,  last  update  2016.03.27  

 
Figure  6:  Whiteboard  sketched  side  view  of  the  machine.  
 
The  bridge  of  the  machine  represents  a  closed  structural  loop,  whereas  the  matrix  methods  
that  are  to  be  used  to  analyze  errors  (Homogeneous  Transfer  Matrices,  HTMs)  is  a  serial  
based  approach.    The  solution  to  this  quandary  is  to  visualize  how  the  machine  deforms  
and  consider  the  philosophy  of  POA  springs  and  the  compliance  matrix  that  represents  the  
structure  between  the  POA  and  the  POI.    
 
With  this  in  mind,  place  the  origin  of  CS_5  at  the  top  of  one  of  the  risers  that  connect  the  X-­‐
Axis  bearings  to  the  bridge  beam.    The  bridge  beam  is  considered  a  simply  supported  beam,    
when  the  Y,  Z  axis  assembly  is  all  the  way  to  the  right  and  thus  far  away  from  the  origin  of  
CS_5  one  might  think  this  represents  a  HUGE  moment  that  will  cause  an  inordinately  (and  
inaccurate)  Abbe  error.    However,  note  in  Figure  3  how  the  axes  directions  have  been  
donated  as  CS_5’  and  CS_5’’  (e.g.,  Z’5  and  Z’’5).    This  indicates  that  the  attachment  points  are  
far  apart.    Given  that  they  are  each  very  stiff,  the  moment  stiffness  of  two  linear  springs  
representing  the  attachment  points,  which  are  spaced  far  apart,  is  also  HUGE  (see  the  linear  
springs  spaced  apart  yielding  rotational  spring  analysis  method  on  the  left  side  of  Figure  4)  

  9  
©  Alexander  Slocum,  last  update  2016.03.27  

and  thus  the  resulting  inaccuracy  of  an  Abbe  error  caused  by  angular  deflection  of  the  
coordinate  system  is  very  small,  and  in  the  worst  case  just  adds  to  the  conservative  nature  
of  the  model.  
 
The  simply  supported  beam  model  of  the  bridge  is  used  to  determine  the  compliance  
matrix  for  the  structure  between  CS_5’s  POA  and  its  POI  (which  is  also  CS_4’s  POA).    As  the  
Y-­‐Axis  servo  moves  the  Y,  Z  axis  assembly  from  left  to  right,  the  slope  will  change  from  
negative  to  zero  (when  its  in  the  middle)  and  back  to  positive.    Hence  the  Abbe  error  at  the  
tool  tip  will  be  accurately  accounted  for.  
 
The  same  philosophy  is  used  for  CS_6  which  models  the  X–Axis.    Note  that  the  compliance  
matrix  for  the  structure  between  CS_6’s  POA  and  POI  is  a  little  more  interesting  because  of  
the  two  risers.    This  just  means  that  the  bending  stiffnesses  are  the  combined  stiffnesses  of  
the  two  riser  elements,  but  the  torsional  stiffness  about  the  Z  axis  is  based  on  the  bending  
stiffness  of  the  risers  spaced  apart  (analogous  to  the  method  shown  on  the  left  side  of  
Figure  4).  
 
Note  that  the  effective  center  of  CS_6  is  thus  at  the  center  of  the  table.    However,  in  terms  of  
the  spreadsheet  model  (and  coordinates  entered)  it  will  be  kept  at  the  center  of  stiffness  of  
the  left  side  X  bearings.    The  final  coordinate  system  is  then  placed  on  the  workpiece  on  the  
table  at  the  point  where  the  tool  will  contact  the  work.      
 
To  model  the  effect  of  springs  and  compliances  associated  with  CS_7  takes  some  thought  
experiments.    This  is  essentially  the  same  condition  as  with  the  bridge,  where  CS_4  
(analogous  here  to  CS_7)  moved  along  the  bridge  and  CS_5’  and  CS_5’’  (analogous  to  CS_6’  
and  CS6’’  here)  were  located  at  the  ends  of  the  bridge.    Reciprocity!  
 
The  POA  for  CS_7  is  thus  modeled  as  the  stiffness  of  the  workpiece  from  where  the  tool  
contacts  it  to  where  it  is  attached  to  the  bed  (in  most  cases  this  should  ideally  be  infinite  
with  respect  to  the  stiffness  (1/compliance)  of  the  tool).    The  compliance  of  CS_7  from  its  
POA  to  its  POI  is  the  compliance  of  the  machine  frame  and  bed  and  is  modeled  as  a  simply  
supported  structure.    However,  some  finesse  is  required  to  apply  the  resultant  forces  and  
moments  from  CS_6  to  the  compliance  matrix.    The  resultant  forces  and  moments  on  the  
origin  of  CS_7  are  applied  to  the  compliance  matrix  to  obtain  the  apparent  deflections  of  the  
POI  of  CS_7.  
 
The  structural  loop  is  thus  complete  and  modeled.    Figure  7  summarizes  coordinate  
systems  1  through  7  and  their  origins  (points  of  attachment)  and  points  of  interest.    Such  a  
table  can  be  helpful  for  the  designer  when  building  the  detailed  error  budget.    Indeed,  
recording  sketches  made  and  freeform  writing  to  record  the  thinking  associated  with  
placement  of  coordinate  systems  and  modeling  of  stiffnesses  and  compliances  can  be  
invaluable  to  the  designer  and  associates.    The  act  of  writing,  knowing  others  will  read,  
helps  to  hone  thoughts  and  resolve  quandaries.    It  also  makes  it  easier  for  peer  review  to  
help  identify  and  correct  issues.  

  10  
©  Alexander  Slocum,  last  update  2016.03.27  

   
 
 
Figure  7:  Summary  of  coordinate  systems  1  through  7  and  their  origins  (points  of  
attachment)  and  points  of  interest.  
 
“Stiffness”  (Compliance)  of  elements  between  a  CS’s  POA  and  POI  
 
In  order  to  assign  reasonable  values  to  the  distances  between  coordinate  systems  origins  
for  the  concept  shown  in  Figure  3  so  the  error  spreadsheet  can  be  utilized,  a  sense  of  
proportion  of  the  size  of  the  structural  elements  is  needed.  
 
First  start  with  an  estimate  of  the  loads  on  the  tool,  and  given  the  allowable  error  of  
structural  elements  (Figure  1),  the  stiffness  required  of  each  axis  can  be  estimated.    Given  
that  some  of  the  parameters  are  essentially  a  “guess”  based  on  what  may  seem  reasonable,  
the  resulting  stiffness  guestimates  should  be  treated  as  fuzzy  estimates  subject  to  change  
and  barter  as  one  idea  or  element  is  traded  off  for  another.    The  tool  tip  is  thus  the  point  of  
interest.    A  spreadsheet  is  a  good  way  for  a  designer  to  quantify  initial  guestimates,  as  
shown  for  the  router  for  example  Figure  8.  
 

  11  
©  Alexander  Slocum,  last  update  2016.03.27  

 
 
Figure  8:  Initial  assessment  of  forces  on  the  cutting  tool.    
 
Next  from  the  first  guess  error  apportionment  shown  in  Figure  1,  we  see  for  the  three-­‐axis  
router  we  are  designing  here,  each  axis’  structure  is  allocated  10  microns  of  error  
(deflection).  This  means  that  each  axis  must  have  a  structural  stiffness  of  about  
85.5N/0.010mm  =  8550  N/mm.    
 
Assuming  for  now  the  joints  are  much  stiffer  than  any  one  beam,  something  we  will  have  to  
pay  careful  attention  to  during  the  design  phase.  Given  the  bridge  type  structure  of  Figure  
3,  for  1.3  m  travel  (so  as  to  cut  a  48”  wide  sheet  of  plywood)  the  bridge  would  be  at  least  
1.4m  between  the  uprights.    Remember,  the  stiffness  (compliance)  goes  like  L^3  so  be  
frugal  with  adding  space  for  what  seems  like  ease  of  design  or  packaging.    In  order  for  a  1.4  
m  long  simply  supported  beam  to  yield  the  a  stiffness  of  8550  N/mm  in  the  middle  of  the  
beam,  the  equation  for  deflection  of  a  simply  supported  beam  loaded  by  a  force  in  the  
middle  is  FL^3/48EI,  a  square  aluminum  cross  section  beam  must  have  a  moment  of  
inertia  of  about  7E6  mm^4.    This  seems  large,  but  can  be  obtained  with  a  standard  stock  
extruded  aluminum  square  tube  6in  x  6in,  x  1/8”  wall,  which  would  be  a  very  efficient  
beam  (high  stiffness  to  weight).  
 
To  help  build  design  intuition,  its  interesting  to  explore  possibilities.    Assume  we  were  
wanting  to  machine  metal  with  higher  precision  than  a  simple  router  would  provide  and  
we  wanted  a  stiffness  in  our  simply  supported  beam  of  about    64  kN/mm  (8X).    This  
requires  a  square  beam  of  about  200mm  x  200mm  x  12mm  wall  (stock  US  size  is  about  8in  
x  8in  x  ½”  wall),  or  equivalently  a  250mm  x  250mm  x  6mm  wall  thickness.    Considering  the  
beam  is  1400mm  long,  the  ratio  of  length  to  height  of  the  latter  is  about  5.6.  A  300mm  x  
300  x  5mm  thick  beam  would  have  the  same  cross  sectional  area  and  be  40%  stiffer!      
Think  of  Saint  Venant  for  beams  that  have  the  same  stiffness,  the  length/cross  section  
dimension  proportions  are  7,  5.6,  and  4.7,  and  the  relative  areas  (weights)  are  1.00,  0.65,  
and  0.39  respectively.      
 
How  to  choose?    The  bigger  cross  section,  smaller  wall  thickness  beam  will  be  much  lighter  
and  hence  have  higher  natural  frequency,  but  can  it  be  obtained?    Indeed  there  are  limited  
sizes  of  stock  square  aluminum  tubing,  with  8in  x  8in  x  1/2in  wall  the  largest.    One  has  to  
go  to  steel  to  get  larger  tube.    A  1400  mm  long  8in  x  8in  x  ¼”  wall  square  steel  tube  would  

  12  
©  Alexander  Slocum,  last  update  2016.03.27  

have  a  stiffness  of  6392  N/mm.    What  is  the  “opposite”  of  a  rectilinear  tube?    A  Round  tube!    
There  are  far  more  options  in  round  tubing,  but  what  are  the  implications  for  bearing  
mounting  and  mounting  the  tube  to  the  rest  f  the  machine?    A  10in  diameter  round  tube  
with  ½”  wall  thickness  has  the  same  cross  sectional  area  as  the  8in  x  ½”  wall  square  tube,  
but  is  20%  stiffer  in  bending  and  60%  stiffer  in  torsion.    What  would  you  choose?  
 
Back  to  the  router,  consider  Figure  4  and  Figure  8  below.    The  compliance  matrix  relates  
the  stiffness  of  a  Coordinate  System’s  structure  from  the  POI  to  the  center  of  stiffness  
(Coordinate  System  N’s  origin).    This  matrix  can  be  created  by  the  user  in  the  coordinate  
systems’  worksheets.    A  6x6  compliance  matrix  is  needed  so  the  angular  deflections  of  a  
structure  in  a  coordinate  system  N  are  added  to  the  next  coordinate  system’s  (N-­‐1)  origin  
deflections.    Then  when  the  error  gains4  for  Coordinate  system  N-­‐1  are  applied,  they  will  
effectively  include  the  Abbe  errors  caused  by  a  structural  member’s  angular  deflection  
causing  other  elements  attached  to  it  to  tilt.    If  a  coordinate  system  is  placed  at  the  center  of  
stiffness  of  bearings  that  allow  motion,  the  compliance  in  the  direction  of  motion  is  
considered  to  be  the  stiffness  of  the  servo  (e.g.,  ballscrew,  coupling,  bearings  etc.).        
 
At  the  origin  of  the  coordinate  system  there  will  be  some  sort  of  attachment  to  the  next  
Coordinate  system’s  POI.    At  each  Coordinate  system  origin  there  are  three  resultant  forces  
and  moments  computed  from  the  forces  and  moments  initially  acting  on  CS_1’s  POI  plus  
any  additional  forces  and  moments  added  along  the  way  to  a  Coordinate  System’s  POI  or  
origin.    If  there  are  forces,  such  as  from  center  of  mass,  in-­‐between,  they  can  always  be  
resolved  into  forces  and  moments  applied  to  the  origin.    Hence  only  three  linear  and  three  
angular  stiffness  values  are  needed  to  describe  the  attachment  of  a  coordinate  system  (just  
a  vector).      
 

 
Figure  9:  Image  of  the  attachment  point  stiffness  and  compliance  matrix  for  a  coordinate  
system  
Since  users  will  tend  to  forget  to  enter  values,  and  the  default  in  Excel  for  an  empty  cell  is  
“0”,  the  compliance  matrix  is  used  in  the  error  budget  spreadsheet.    It  also  makes  more  
                                                                                                               
4  Some  positive  angular  errors  result  in  negative  displacements,  thus  since  random  errors  

have  no  sign,  for  all  the  coordinate  systems,  the  magnitude  of  the  amplification  of  angular  
errors  must  be  individually  computed  for  each  potential  angular  error.    A  table  of  these  
amplification  factors,  or  error  gains,  can  then  be  used  to  compute  the  Abbe  errors  
attributed  to  random  angular  errors  about  each  of  the  axes  of  each  of  the  coordinate  
systems.    The  error  gains  for  the  systematic  errors,  on  the  other  hand,  do  have  a  sign.  

  13  
©  Alexander  Slocum,  last  update  2016.03.27  

sense  given  that  we  have  forces  and  want  to  determine  deflections,  so  X  =  C*F,  where  C  =  
1/k.    Note  the  matrix  will  always  be  symmetric  in  this  analysis.  The  elements  for  which  6x6  
compliance  matrices  are  needed  are  shown  in  Figure  7.  
 
It  is  desirable  to  have  a  worksheet  within  the  error  budget  spreadsheet  for  entering  the  
dimensions  of  members  and  to  calculate  the  stiffness  matrices  since  the  formulas  will  be  
similar.    Note  that  when  equating  the  length  of  a  beam  by  referring  to  a  cell  in  the  
worksheet  Axes'  CS  and  Loads  that  sometimes  the  coordinates  may  be  negative,  but  lengths  
of  beams  are  always  positive,  so  use  the  ABS()  function.    Also  note  when  entering  formulas  
for  compliances  remember  to  consider  the  sign.    For  example,  a  moment  may  cause  a  
deflection  in  a  negative  direction5.      
 
It  is  helpful  to  have  the  formulas  for  deflection  and  slope  of  cantilever  and  simply  
supported  beams  subject  to  forces  and  moments  as  shown  in  Figure  10.    It  is  very  difficult  
to  have  a  beam  with  a  true  zero  slope  condition  at  both  ends,  and  it  is  more  conservative  to  
assume  a  simply  supported  condition.    Note  it  is  important  to  be  careful  of  the  sign  of  the  
applied  force  or  moment  and  the  resulting  deflections  and  slopes.    These  formulas  take  this  
into  account,  such  that  if  a  horizontal  beam  were  also  attached  to  the  point  where  the  force  
or  moment  was  applied,  the  resulting  Abbe  error  would  have  the  proper  sign.  
 

 
 
 

                                                                                                               
5  NOTE  Excel  has  a  strange  “feature”  where  entering  a  formula  such  as  “-­‐L/(E*I)”  returns  a  

positive  value!    Entering  “-­‐1*L/(E*I)”  returns  a  negative  value.    The  same  is  true  when  a  
reference  is  raised  to  a  power:    -­‐1*B130^2  yields  –B1302;  however,  -­‐B130^2  yields  B1302.  

  14  
©  Alexander  Slocum,  last  update  2016.03.27  

Figure  10:  Deflection  and  slope  formulas  for  common  beams  used  to  model  machines.    
Position  of  the  force  on  a  simply  supported  beam  is  from  left  to  right.    A  right  hand  
coordinate  system  is  assumed.  
 
It  is  very  important  to  create  sketches  of  beams  being  modeled  with  their  assigned  
coordinate  systems,  and  compare  the  orientations  to  the  right  hand  systems  used  for  
shown.    A  check  of  the  equations  shown  in  Figure  10  corresponds  to  the  shapes  shown  in  
the  sketches,  which  can  be  verified  by  supporting  and  loading  an  object  and  observing  how  
it  deforms.    As  shown  in  Figure  11,  these  values  represent  the  compliances,  and  it  is  good  to  
see  that  they  are  symmetric:    the  slope  caused  by  a  force  has  the  same  sign  and  value  as  a  
deflection  caused  by  the  moment.  
 

 
 
Figure  11:  Normalized  results  of  deflection  and  slope  formulas  of  Figure  11  applied  to  
common  beams  used  to  model  machines.  
 
For  a  servo  controlled  axes,  which  controls  the  position  of  a  CS  origin  (POA)  wrt  another  
coordinate  system’s  POI  CS  origin  (POA@Center  of  Stiffness),  the  axial  stiffness  
(compliance)  of  the  system  at  the  Point  Of  Attachment  (POA)  will  be  dominated  by  the  
actuator,  such  as  a  ballscrew  or  belt.    The  compliance  of  the  shaft  (or  belt)  can  be  computed  
and  then  a  factor  (typically  3  for  the  attachment  (e.g.,  nut)  to  the  moving  member  and  the  
support  bearings)  accounts  for  the  compliance  of  other  elements  used  to  attach  it.    Figure  
12  shows  an  image  from  the  worksheet  CS_3  Z-­‐Axis  .  
 

  15  
©  Alexander  Slocum,  last  update  2016.03.27  

 
 
 Figure  12:  Results  of  deflection  and  slope  formulas  applied  to  common  beams  used  to  
 
We  are  now  ready  to  apply  the  above  to  each  of  the  coordinate  systems  as  shown  in  Figure  
7  using  the  worksheet  bearings,  servo,  structure.  
 
CS_1  Tool  Holder  
 
Structure  compliance  (from  a  coordinate  system’s  POA  to  its  POI)  
 
For  the  Tool  Holder,  see  Figure  5,  the  compliance  is  that  of  the  tool  which  is  modeled  as  a  
fraction  of  that  of  a  cylinder  to  account  for  the  flutes.    Figure  13  shows  the  Tool  Holder  
system  compliance  calculations  done  in  the  worksheet    CS_1  Toolholder.    Note  how  figures  
are  pasted  to  help  the  designer  check  for  the  sign  of  the  compliances  in  accordance  with  
coordinate  system  placement  on  the  beam  element.    This  should  be  done  for  each  
coordinate  system.    Note  the  “Note:”  on  the  last  cell.    The  designer  should  leave  notes  on  
how  they  modeled  the  system.    Notes  can  also  be  attached  to  individual  calculation  cells.  
 
Note  that  the  worksheet  is  made  generic,  so  if  certain  cells  with  their  entires  are  not  
needed,  those  cells  should  just  be  colored  blue.    No  need  to  delete  or  clear  entries  as  they  
may  be  used  in  the  next  round  of  use  of  the  spreadsheet.  
 

 
 
Figure  13:  Compliance  calculations  for  the  Tool  Holder  coordinate  system.  
Attachment  compliance  (POA  to  next  CS  POI)  
 

  16  
©  Alexander  Slocum,  last  update  2016.03.27  

The  attachment  compliance  for  the  tool  holder,  a  tapered  collet  for  example,  will  be  an  
order  of  magnitude  greater  than  the  stiffness  of  the  tool  itself,  and  even  of  the  spindle  shaft.  
 
It  is  helpful  that  whenever  copying  a  set  of  cells,  pasting,  and  then  updating  the  formulas,  to  
change  the  cell  background  color  as  a  reminder  to  make  sure  to  change  all  the  cells!    The  
compliance  matrix  references  the  cells  to  the  left,  so  if  additional  coordinate  systems  are  to  
be  added  for  a  new  machine  (or  CS),  the  entire  block  can  just  be  copied  and  pasted,  and  
then  checked  for  the  orientation  of  the  structure  with  respect  to  the  coordinate  systems  
and  the  compliance  matrix.    As  cells  are  changed,  change  their  background  color  and  thus  
prevent  missing  one.  
 
CS_2  Spindle  
 
Structure  compliance  (from  a  coordinate  system’s  POA  to  its  POI)  
 
The  spindle  coordinate  system  is  placed  at  the  center  of  the  spindle  shaft  between  the  
spindle  bearings,  as  discussed  in  detail  in  conjunction  with  Figure  5  above,  and  the  entries  
for  determining  the  compliance  matrix  from  the  worksheet  CS_2  Spindle  are  shown  in  
Figure  14.  
 

 
 
Figure  14:  Compliance  calculations  for  the  Tool  Holder  coordinate  system.  
Attachment  compliance  (POA  to  next  CS  POI)  
 
Note  that  the  forces  applied  to  the  simply  supported  beam  are  considered  to  be  at  the  front  
spindle  bearing  and  thus  they  cannot  create  an  angular  deflection  and  the  corresponding  
compliance  terms  compute  to  zero,  as  do  the  off  axis  moment  effects  on  displacement.  
 
Attachment  compliance  (POA  to  next  CS  POI)  
 
The  stiffness  of  the  POA  of  CS_2  to  the  POI  on  the  Z-­‐Axis  is  assumed  to  be  dominated  by  the  
spindle  bearings.    The  attachment  points  are  assumed  to  be  at  the  equator  points  of  the  
four  bearings  
 
CS_3  Z-­‐Axis  
 

  17  
©  Alexander  Slocum,  last  update  2016.03.27  

Structure  compliance  (from  a  coordinate  system’s  POA  to  its  POI)  


 
The  structure  between  the  Z  axis  POA  and  its  POI  (the  origin  of  the  Spindle)  is  dominated  
by  the  tube  that  is  the  Z-­‐Axis  structure  itself:    The  Z-­‐axis  bearing  blocks  are  fixed  to  a  plate  
structure,  which  also  bolts  to  the  Y  axis  bearing  blocks.    The  Z-­‐Axis  profile  bearing  rails  are  
bolted  to  the  structure,  which  is  modeled  as  an  aluminum  square  tube.    A  square  tube  is  
considered  because  the  cutting  forces  can  be  equal  in  the  X  and  Y  directions  and  so  equal  
stiffness  is  needed.    Note  the  length  of  the  structure  is  tied  to    
 
Attachment  compliance  (POA  to  next  CS  POI)  
 
The  stiffness  of  the  POA  of  CS_3  to  the  POI  on  the  Y-­‐Axis  is  assumed  to  be  dominated  by  the  
Z-­‐Axis’  bearing  carriages  (blocks).    
 
CS_4  Y-­‐Axis  
 
Structure  compliance  (from  a  coordinate  system’s  POA  to  its  POI)  
 
The  Y-­‐Axis  POI  is  the  Z-­‐Axis  origin  and  the  two  are  very  close  together  so  there  is  no  
appreciable  structural  compliance  between  the  two:  Its  just  a  chunky  plate,  so  the  
compliances  are  all  set  to  “0”.  
 
Attachment  compliance  (POA  to  next  CS  POI)  
 
The  stiffness  of  the  POA  of  CS_4  to  the  POI  on  the  Bridge  is  assumed  to  be  dominated  by  the  
Y-­‐Axis’  bearing  carriages  (blocks).    
 
CS_5  Bridge  
 
Structure  compliance  (from  a  coordinate  system’s  POA  to  its  POI)  
 
The  bridge  spans  the  risers,  and  to  the  bridge  are  attached  bearing  rails  on  which  the  Y  axis  
bearing  carriages  move.    The  structural  compliance  of  the  bridge  structure  thus  must  be  
calculated.    As  discussed  above,  for  a  closed  structural  loop  machine,  an  open  kinematic  
model  can  be  used  by  considering  the  attachment  points  for  CS_5  to  CS_6  to  be  widely  
spaced  apart  (at  the  top  of  each  riser)  and  then  calculate  the  angular  stiffness  of  the  
attachment  accordingly.    This  gives  a  huge  rotational  stiffness  so  the  angular  deflection  will  
be  very  small.  
 
The  structural  compliance  is  that  of  a  simply  supported  beam,  so  when  the  Y  axis  is  
positioned  in  the  middle  of  the  bridge,  the  angular  deflection  of  the  bridge  will  actually  be  
zero.    Near  the  ends  of  the  bridge,  the  compliance  becomes  huge.    
 
Attachment  compliance  (POA  to  next  CS  POI)  
 

  18  
©  Alexander  Slocum,  last  update  2016.03.27  

The  stiffness  of  the  POA  of  CS_5  to  the  POI  on  the  X-­‐Axis  (the  top  of  the  risers)  is  a  bolted  
connection  which  if  done  correctly  should  be  as  stiff  as  if  welded.    It  is  modeled  as  4  points  
spaced  at  the  corners  of  the  joint.    
 
CS_6  X-­‐Axis  
 
Structure  compliance  (from  a  coordinate  system’s  POA  to  its  POI)  
 
From  the  plates  to  which  the  X-­‐Axis  bearing  carriages  are  attached  up  to  the  bridge  are  the  
risers.    The  compliance  of  these  structures  must  be  taken  together,  so  for  example  in  the  X  
direction,  the  compliance  of  the  two  risers  is  one-­‐half  of  that  of  a  single  riser.    The  torsional  
compliance  about  the  Z-­‐Axis  is  the  effect  of  the  lateral  compliances  K  of  the  risers  acting  as  
a  force  couple  over  the  length  L  of  the  bridge  and  is  thus  2/KL2.    The  roll  compliance  about  
the  X-­‐Axis  is  a  function  of  the  compression  stiffness  of  the  risers  ad  the  bridge  length  and  
thus  is  quite  huge.  
 
Just  before  we  come  to  the  end  of  the  structural  loop  back  at  the  toolpoint,  it  is  important  to  
note  that  there  are  essentially  two  options  for  placement  of  the  bed’s  CS  origin.    The  first  is  
to  simply  locate  it  in  the  center  of  the  bed  and  call  that  the  reference  frame  origin.    Then  the  
entire  machine’s  performance  is  wrt  the  reference  frame  (center  of  bed).  
 
The  second  option  is  to  assume  that  Bed  coordinate  axis  origin  is  also  at  the  toolpoint,  and  
thus  the  position  of  the  X-­‐Axis  coordinate  system  with  respect  to  the  bed  should  be  such  
that  the  origin  of  the  Bed  coordinate  system  is  located  at  the  toolpoint.    This  can  be  
achieved  with  the  product  of  -­‐1  and  the  sum  of  the  coordinates  along  the  structural  loop  
chain.    This  is  what  will  be  assumed  for  this  design.  
 
Attachment  compliance  (POA  to  next  CS  POI)  
 
The  stiffness  of  the  POA  of  CS_6  to  the  POI  on  the  Bed  is  assumed  to  be  dominated  by  the  X-­‐
Axis’  bearing  carriages  (blocks).    
 
CS_7  Bed  
 
Structure  compliance  (from  a  coordinate  system’s  POA  to  its  POI)  
 
The  structure  of  the  bed  is  assumed  to  be  a  sandwich  panel  such  as  plywood  bonded  to  
either  side  of  Styrofoam  insulation  for  a  home  machine  or  1/8”  aluminum  sheets  bonded  to  
either  side  of  honeycomb.    The  bed  really  thus  is  more  of  a  plate  structure,  but  for  
simplicity  sake  here  it  is  assumed  to  be  a  beam  of  length  the  same  as  the  bridge,  and  width  
ten  times  its  thickness.    It  ends  up  being  as  stiff  as  the  bridge.  
 
The  X-­‐Axis  servo  position  thus  ends  up  not  being  a  sensitive  parameter,  what  is  more  
important  is  the  position  of  the  spindle  as  controlled  by  the  Y  axis.    
 
Attachment  compliance  (POA  to  next  CS  POI)  

  19  
©  Alexander  Slocum,  last  update  2016.03.27  

 
The  stiffness  of  the  POA  of  CS_7  to  the  POI  od  the  “next”  reference  frame,  given  that  the  next  
frame  is  to  be  at  the  toolpoint  would  actually  be  the  attachment  of  the  part  being  cut  to  the  
bed.    If  properly  attached,  it  should  be  very  much  ore  stiff  than  the  stiffness  of  the  bed  itself  
for  example.    
 
Discussion  on  the  layout  of  the  spreadsheet  to  perform  the  required  analysis  
 
This  section  is  under  construction…  it  is  not  needed  for  operation  by  a  user  of  the  
spreadsheet,  but  it  is  good  to  discuss  the  analysis  process.    It  will  be  written  in  coming  
weeks….  
 
Error  gains  
 
The  following  figure  illustrates  why  error  gains  are  needed,  and  for  systematic  errors,  some  
of  the  gains  are  negative  
 

  20  
©  Alexander  Slocum,  last  update  2016.03.27  

  21  

Common questions

Powered by AI

Coordinate systems are critical in structural modeling as they align with the centers of stiffness and major connections within the machine design. Each coordinate system has a Point Of Interest (POI) and Point Of Attachment (POA), influencing how errors and compliance are introduced into the model. By carefully placing coordinate systems at strategic locations such as the center of stiffness or at critical attachment points, designers can more accurately predict and manage deflections, systematic errors, and compliance, thereby improving error prediction and the machine's overall performance .

Compliance matrices play a significant role in error budgeting by quantifying the flexibility between different components in a machine. They help in assessing how structural elements and their connections respond to loads, which is crucial for predicting how these elements will contribute to or mitigate errors. The iterative refinement process is influenced by these matrices as designers adjust component dimensions and materials to meet the compliance requirements, thereby ensuring the design meets performance expectations .

Iterative design processes in deterministic approaches are vital for refining and optimizing machine designs. They allow designers to repeatedly test and adjust variables—such as errors, compliance, stiffness—in response to performance predictions. This cyclical evaluation helps identify potential design issues early, refine component specifications, and ensure that the final design meets stringent performance criteria without over-engineering, ultimately leading to cost-effective and reliable machine solutions .

Incorporating peer reviews during early design stages enhances development by fostering collaborative examination of ideas, uncovering overlooked errors, and offering diverse perspectives that can refine concepts. This collaborative approach helps challenge assumptions, integrate interdisciplinary insights, and ensure that the design meets broader engineering criteria, thus leading to more robust and innovative solutions .

Modeling the compliance of a simply supported beam poses challenges such as accurately predicting angular deflections and capturing the complexities of load distribution along the beam. These challenges affect design accuracy as misinterpretations can lead to underestimated or overestimated flexibility, impacting the error budget and resulting in suboptimal machine performance under dynamic loading. Precise compliance modeling is therefore critical to achieving the intended design accuracy and reliability .

Deflections in spindle shaft geometry contribute to machine tool errors through angular deflections that induce Abbe errors, affecting tool tip precision. Strategies to mitigate these errors include optimizing the spindle shaft support, such as using back-to-back mounted bearings, and refining the compliance modeling of the spindle shaft as a simply supported beam to accurately predict and correct angular displacements .

Spreadsheets support the error budgeting process by providing a flexible and accessible platform for calculating, comparing, and visualizing potential errors and compliance across machine components. They facilitate rapid adjustments, trend analysis, and data-driven decision-making, allowing designers to iteratively refine component specifications for optimized performance. This computational tool helps integrate various design factors efficiently, leading to well-informed design decisions .

Error apportionment plays a crucial role in deterministic design by allowing designers to allocate allowable error across different machine axes and components, ensuring that no single axis bears an unreasonable burden of error. In the early stages, designing with error apportionment helps in predicting the overall machine performance and accommodating high-performance requirements by systematically distributing errors and providing a clearer target for machine accuracy and repeatability .

Hand-drawn sketches and physical models in concept development enable designers to remain flexible and engage in free-spirited creative thinking. This approach prevents premature commitment to specific design details and encourages the exploration of diverse ideas. It also allows for an interactive and engaging process, promoting creativity by involving informal elements like crayons and young helpers in the design process, which can lead to creative breakthroughs .

Error management in machine design, like using a dither signal in signal processing, involves introducing slight randomness or adjustments to improve system performance. In design, this involves allocating and handling different error sources among components to smooth out inaccuracies, just as a dither signal introduces noise to help a system overcome static friction and enhance accuracy. Both processes aim to enhance precision by mitigating unintended effects .

You might also like