Homebrew PCBs

>A 40 pound table would not move, the drill
>bit would break.

LOL! Where is the force trying to move the drill bit coming from? 
Assuming such a force existed, why would this force suddenly 
disappear with a perfect zero backlash setup and not break the bit 
then also? Both a 40 pound table and a zero backlash setup would have 
the same thing in common "THE TABLE WON'T MOVE".

Any slop would be manifested as spurious **movement of the table** in 
the backlash dead zone since the X and Y screws are "fixed" at drill 
time by the holding torque of the motor. You seem to be confusing 
drilling and milling in the same random thought.

>There will be others here that will be using dremels and will be 
>using very lightweight tables with loose construction.

With the right size hole the bit will be piloted to some extent in a 
lightweight setup just as in manual drilling. Most of us are alrady 
familiar with that trick. Long story short....You're making a 
mountain out of a mole hill. However, I would agree with you 100% if 
discussing milling or drilling.




--- In Homebrew_PCBs@yahoogroups.com, "lcdpublishing" 
<lcdpublishing@y...> wrote:
> Considering that most of the drill bits used for circuit boards are 
> so small, they probably would bend or break before they cause the 
> table and part to move.  A 40 pound table would not move, the drill 
> bit would break.  
> 
> There will be others here that will be using dremels and will be 
> using very lightweight tables with loose construction.  So, while 
> you may not agree, fact is fact, software backlash compensation is 
> not the "Solution" to poor machine design and construction.  
> However, sometimes you can get away with it.
> 
> Furthermore, you make the statement "If you know what you are 
> doing", well, considering that most of the people considering 
making 
> a CNC machine will be their first CNC (and probably machine), it's 
a 
> bad assumption.  Again, "trickery", is not a good solution to poor 
> machine design and construction.
> 
> 
> 
> 
> 
> --- In Homebrew_PCBs@yahoogroups.com, "derekhawkins" 
> <derekhawkins@y...> wrote:
> > >When drilling, the position will be off and the
> > >holes will not be on size.
> > 
> > I don't agree with this. Perhaps with a light table and a high 
> runout 
> > drill such as the ubiquitous Dremel one may have an issue. With a 
> > heavy table (the X axis and saddle in my setup weighs over 40 
lbs) 
> > and a low runout drill this is not an issue. Backlash 
compensation 
> in 
> > software is very effective when it comes to precision drilling 
IMO 
> > provided you know what you're doing. Milling and routing is 
> another 
> > story.
> > 
> > --- In Homebrew_PCBs@yahoogroups.com, "lcdpublishing" 
> > <lcdpublishing@y...> wrote:
> > > While I don't fully know the exact details, I can tell you 
right 
> > > away, you cannot rely on software to eliminate the backlash 
> within 
> > a 
> > > CNC system.  For basic positioning, perhaps a little, but in 
> > > practice, no.
> > > 
> > > While software can compensate for the static deviation between 
> > where 
> > > it told the motor to position and where it really is based on 
> > > encoder feedback, the "Slop" is still there.  During cutting, 
> the 
> > > dynamic cutting forces will cause the axes to move around 
within 
> > the 
> > > mechanical slop.  When drilling, the position will be off and 
> the 
> > > holes will not be on size.  When milling, all sorts of 
> dimensional 
> > > problems pop up and in the end, you will have wished you took 
> the 
> > > time the eliminate the mechancial backlash.
> > > 
> > > While there is backlash compensation within all CNC systems I 
> have 
> > > used in the industrial world, it is primarily only used to 
> > > compensate for a very samll amount of backlash .001" or less 
> > > usually. Beyond that and it is time to correct the mechanical 
> > > problems.
> > > 
> > > So, don't waste your time or effort in thinking of a software 
> > > solution to backlash - backlash is a mechanical problem and 
must 
> be 
> > > designed out of the system for even the most basic of machine 
> > > performance.
> > > 
> > > 
> > > 
> > > 
> > > 
> > > 
> > > 
> > > 
> > > --- In Homebrew_PCBs@yahoogroups.com, "Stefan Trethan" 
> > > <stefan_trethan@g...> wrote:
> > > > On Tue, 28 Jun 2005 08:36:23 +0200, mikegw20 <mikegw20@h...> 
> > wrote:
> > > > 
> > > > > Now I have never built a CNC anything and I am not overly 
> > > mechanical.
> > > > > Right that is my qualifications out of the way...
> > > > > I was thinking about high torque motors to drive a linear 
> > thread 
> > > and I
> > > > > thought about those really cheap cordless drills.  That is 
> idea 
> > > pt1.
> > > > > Now of course you need some sort of feedback for the 
> position, 
> > > the
> > > > > normal way would be to put a shaft encoder on the drive.  
My 
> > > thinking
> > > > > is that if you used a linear encoder then you would get 
> absoulte
> > > > > positioning thereby negating the need for backlash 
> > > compensation.  Now a
> > > > > quick look around ebay reveals that linear encoders are not 
> > > cheap.  So
> > > > > has anyone used optical mice for a linear encoder? (idea pt 
> b)
> > > > > Here endith todays musings.
> > > > > Mike
> > > > 
> > > > 
> > > > cordless drills usually use a motor what the model-vehicle 
> > > builders would  
> > > > call similar to a "speed 600".
> > > > Not a bad motor, and you can get replacements really cheap as 
> > well 
> > > as  
> > > > better motors the same size.
> > > > 
> > > > I would try to find drills with metal gears for this purpose. 
> I 
> > > recently  
> > > > bought a very cheap one with plastic gears, it is still 
> working, 
> > > but i'm  
> > > > always expecting it to fail any time. It was cheaper than a 
> > > replacement  
> > > > battery for a better drill.
> > > > There are often ebay lots of many drills.
> > > > 
> > > > 
> > > > As for linear encoder, there are those striped plastic strips 
> in 
> > > printers,  
> > > > but a much easier solution would be to use digital calipers 
> and 
> > > use the  
> > > > data-out, 'cause you would get absolute position not only 
> > > relative.  
> > > > Probably more expensive than shaft encoders and certainly 
more 
> > > expensive  
> > > > than steppers.
> > > > 
> > > > 
> > > > ST