Homebrew PCBs

Jan, that is a nice looking unit!
http://photos.groups.yahoo.com/group/homebrew_pcbs/vwp?.dir=/JanRowland

I couldn't find any other information in the Files / Homemade PCB Equipment
folder, am I just missing it or is there another place where I could get a
closer look on your design?

I'm especially interested in your description of "cheap" backlash system.
I'd love to see a close up of that.

As far as "error components" goes, has anyone looked at reducing the number
of joints required as a means of reducing error? And interesting idea is
that of using only two joints in a radial system rather than the 4 or more
required in a linear system.
http://www.massmind.org/techref/idea/lccam.htm scroll down
http://freeandeasy.sourceforge.net/ at the end of the paragraph about
"prototype XY table"
http://freeandeasy.sourceforge.net/rotary.php

Roller-skate bearings top and bottom on the main shaft were sloppy (as
expected) but when they are pressed towards each other, the compression
takes all the slop out and the result is very good. It reminded me of what
you were doing, but I don't think it is really the same.

---
James Newton, massmind.org Knowledge Archiver 
james@... 1-619-652-0593 fax:1-208-279-8767 
All the engineering secrets worth knowing:
http://techref.massmind.org What do YOU know?




Message: 21
   Date: Mon, 14 Jun 2004 01:53:10 EDT

   From: JanRwl@...
Subject: Re: bit cleaning while drilling PCBs

In a message dated 6/13/2004 7:52:33 PM Central Standard Time, 
dg140@... writes:
How did you account for backlash in your homebrew CNC rig?<<

You DID see the photo in the PHOTO section under "Jan Rowland"?  

I built that in '84, and was feeling CHEAP, so I did NOT spring for the 
preloaded Turcite nuts sold by Ball Screws and Actuators, San Jose, CA.  I
bought 
the "Single" Turcite nuts which have a 15/16"-16 threaded end for fastening 
into the carriage or flange or whatever.  I then carefully bored/tapped a
hole 
for 6-32 cap-screws through one side, and then I band-sawed through that one

side with the screw-hole so that inserting and tightening the screw would 
"reduce" the I.D.  This would tighten-up on the screw nicely, and obviously
a "feel" 
for HOW-tight was necessary.  It worked, and has not been a problem needing 
re-adjustment.  These nuts are also more than twice as compact as the
(spring-) 
preloaded nuts, and MUCH less expensive.  These screws are 7/16" dia. 
1/8"-pitch, 4-start.  Thus, one turn = 1/2" movement.  So each step of the
axis 
steppers is 0.0025".  Thus, I had to write my "Coordinate recording program"
to 
round-off each X and Y typed in to the nearest 0.0025" (not 0.001").  I
realized 
how cumbersome this would be (got USED to it!), but wanted the 1/2" lead for

SPEED of operation between holes. 
  What about those PCB drilling machines, what do they do for backlash and 
other error components?<<
Of course, all lead-screws on such professional CNC $100,000 machines are 
covered with bellows ("way-covers", etc.), so they are not viewable by a
visitor 
not willing to exhibit violence.  I am however POSITIVE they are preloaded 
ball-screws!  AND, the screws are surely "precision ground", not the ROLLED
type 
also available.  

"Other error components?"  There are MORE?  Uh . . .  I'd guess MUCH care
and 
careful workmanship by folks who can READ (if you get what I mean!).

Hi Jan,

> Jan, that is a nice looking unit! 
>
<http://photos.groups.yahoo.com/group/homebrew_pcbs/vwp?.dir=/JanRowland
>

I agree too!

I couldn't help but notice the Commodore Pet sitting next to it. Do you
still use this?

I started my programming career on one of these :)

--
Dave Hylands
Vancouver, BC, Canada
http://www.DaveHylands.com/

In a message dated 6/14/2004 7:14:14 PM Central Standard Time, 
dhylands@... writes:
I couldn't help but notice the Commodore Pet sitting next to it.   Do you 
still use this?<<

Dave!  A brief "Yes." would not be UNtrue, but it WOULD be inaccurate, so 
this needs a response.  The scene in the photo is still exactly as it is in that 
photo.  That PET is the first "2001/16N PET" I ever had (new in '78!  The very 
first was an "8K Chicklet Keyboard PET!").  IT still works!  I have another, 
a 4032, on my first-ever home-brew CNC Lathe which I bought for $100 about 
'85, and which still works but for some bits in high-RAM that seem to be doing 
that "alpha particle thing" now and then.  A sudden and totally unpredictable 
unreliability problem.  I had an offer from a local "PC Shop" that were I to 
come take away ALL the old PET stuff, working or not, I could HAVE it all, no 
charge.  I did that.  Two or three of the dozen "kinda worked" and I moved parts 
around until I got two more working, but they all gave out in short order, so 
I now have NOTHING "in the office" upon which to program the PET's.  A newer 
home-brew CNC lathe has a now-obsolete-but-working-fine Compaq "Tower" and an 
old 13" monochrome monitor, and that is totally satisfactory.  So I intend to 
replace the PET on the lathe with another "old PC" as soon as I can find one of 
those round tuits.  I have not used the PET to run the PCB-drill you see in 
about a year, but I DID fire it up the other day to check, and it still 
functions well (I still had an old tape in the cassette unit of a forgotten PC-board, 
for testing it).  I had already begun writing the BASIC programs for the PC, 
basically a word-for-word copy of  the PET programs, except for the 
machine-code patch, CLS, etc.  

Any questions, feel free!            Jan Rowland


[Non-text portions of this message have been removed]

James,

Didn't see the link yet, but usually you trade one source of error 
for another. The probnlem with rotary-used-linear machines is that as 
the distance from the pivot changes so does the mech. 
advantage, "step" size, etc. And trying to make a linear move near 
one of the pivots requires a nearly instantaneous change in direction 
and requires acceleration far in excess of what most inexpensive 
drives can achieve...

There's more, but these are the bigge's

Ballendo 

In Homebrew_PCBs, "James Newtons Massmind" wrote:
>As far as "error components" goes, has anyone looked at reducing the 
>number of joints required as a means of reducing error? And 
interesting idea is that of using only two joints in a radial system 
rather than the 4 or more required in a linear system.
> http://www.massmind.org/techref/idea/lccam.htm scroll down
> http://freeandeasy.sourceforge.net/ at the end of the paragraph 
about
> "prototype XY table"
> http://freeandeasy.sourceforge.net/rotary.php
> 
> Roller-skate bearings top and bottom on the main shaft were sloppy 
(as
> expected) but when they are pressed towards each other, the 
compression
> takes all the slop out and the result is very good. It reminded me 
of what
> you were doing, but I don't think it is really the same.
> 
> ---
> James Newton, massmind.org Knowledge Archiver 
> james@m... 1-619-652-0593 fax:1-208-279-8767 
> All the engineering secrets worth knowing:
> http://techref.massmind.org What do YOU know?
> 
> 
> 
> 
> Message: 21
>    Date: Mon, 14 Jun 2004 01:53:10 EDT
>    From: JanRwl@A...
> Subject: Re: bit cleaning while drilling PCBs
> 
> In a message dated 6/13/2004 7:52:33 PM Central Standard Time, 
> dg140@f... writes:
> How did you account for backlash in your homebrew CNC rig?<<
> 
> You DID see the photo in the PHOTO section under "Jan Rowland"?  
> 
> I built that in '84, and was feeling CHEAP, so I did NOT spring for 
the 
> preloaded Turcite nuts sold by Ball Screws and Actuators, San Jose, 
CA.  I
> bought 
> the "Single" Turcite nuts which have a 15/16"-16 threaded end for 
fastening 
> into the carriage or flange or whatever.  I then carefully 
bored/tapped a
> hole 
> for 6-32 cap-screws through one side, and then I band-sawed through 
that one
> 
> side with the screw-hole so that inserting and tightening the screw 
would 
> "reduce" the I.D.  This would tighten-up on the screw nicely, and 
obviously
> a "feel" 
> for HOW-tight was necessary.  It worked, and has not been a problem 
needing 
> re-adjustment.  These nuts are also more than twice as compact as 
the
> (spring-) 
> preloaded nuts, and MUCH less expensive.  These screws are 7/16" 
dia. 
> 1/8"-pitch, 4-start.  Thus, one turn = 1/2" movement.  So each step 
of the
> axis 
> steppers is 0.0025".  Thus, I had to write my "Coordinate recording 
program"
> to 
> round-off each X and Y typed in to the nearest 0.0025" (not 
0.001").  I
> realized 
> how cumbersome this would be (got USED to it!), but wanted the 1/2" 
lead for
> 
> SPEED of operation between holes. 
>   What about those PCB drilling machines, what do they do for 
backlash and 
> other error components?<<
> Of course, all lead-screws on such professional CNC $100,000 
machines are 
> covered with bellows ("way-covers", etc.), so they are not viewable 
by a
> visitor 
> not willing to exhibit violence.  I am however POSITIVE they are 
preloaded 
> ball-screws!  AND, the screws are surely "precision ground", not 
the ROLLED
> type 
> also available.  
> 
> "Other error components?"  There are MORE?  Uh . . .  I'd guess 
MUCH care
> and 
> careful workmanship by folks who can READ (if you get what I mean!).

Sure, but all you have to do is limit your working area to avoid the center
pivot. Step size is smaller near the pivot and larger away from it, but with
a nice linear microsteppng controller and some drive reduction (see the
pictures) neither extreme should be a problem. The calculations required to
convert x,y into polar are not terribly difficult, and the cost of the unit
goes WAY down because only a couple of components need to be precision.

Next objection? <GRIN>

Now back to those nuts... Did you measure the slop before and after
tightening? And did the friction increase?

---
James Newton, massmind.org Knowledge Archiver 
james@... 1-619-652-0593 fax:1-208-279-8767 
All the engineering secrets worth knowing:
http://techref.massmind.org What do YOU know?



Message: 10
   Date: Wed, 16 Jun 2004 11:02:26 -0000

   From: "ballendo" <ballendo@...>
Subject: Re: Turcite nuts, was: bit cleaning while drilling PCBsf

James,

Didn't see the link yet, but usually you trade one source of error 
for another. The probnlem with rotary-used-linear machines is that as 
the distance from the pivot changes so does the mech. 
advantage, "step" size, etc. And trying to make a linear move near 
one of the pivots requires a nearly instantaneous change in direction 
and requires acceleration far in excess of what most inexpensive 
drives can achieve...

There's more, but these are the bigge's

Ballendo 

In Homebrew_PCBs, "James Newtons Massmind" wrote:
>As far as "error components" goes, has anyone looked at reducing the 
>number of joints required as a means of reducing error? And 
interesting idea is that of using only two joints in a radial system 
rather than the 4 or more required in a linear system.
> http://www.massmind.org/techref/idea/lccam.htm scroll down
> http://freeandeasy.sourceforge.net/ at the end of the paragraph 
about
> "prototype XY table"
> http://freeandeasy.sourceforge.net/rotary.php
> 
> Roller-skate bearings top and bottom on the main shaft were sloppy 
(as
> expected) but when they are pressed towards each other, the 
compression
> takes all the slop out and the result is very good. It reminded me 
of what
> you were doing, but I don't think it is really the same.
>

System stiffness, or lack thereof, due to long moments and singular 
pivots involved...

Ballendo


--- In Homebrew_PCBs@yahoogroups.com, "James Newtons Massmind" 
<jamesmichaelnewton@y...> wrote:
> Sure, but all you have to do is limit your <snip>
> Next objection? <GRIN>