Parallel Port As Output In Matlab

[crys]

i know it is possible to output something in MATLAB via parallel port.. but i don't know how!! this is in regards to the PCB drilling machine.. we cannot base our project from CNC since we have to make from scratch. what we have in mind is that we extract a gerber file from OrCad, relate it to Matlab which gets our coordinates which sends out x and y coordinates via parallel port (pins 2 and 3). our pcb drill has its own pic which gets the x and y coordinates from the parallel port, and converts it to polar coordinates..

are we a genius or not, ahaha! jowk!

seriously, that's our concept in mind, and i hope it really does work.. we only have less than a month to finalize all of our ideas.. i hope you guys help us out!

thank you!

 

[yngndrw]

Just an idea:
Would it not be easier to calculate the polar coordinates on the PC - Much more processing power means that you don't have to cut any corners.

 

[crys]

a gerber file just calculates the x and y coordinates from your pcb layout. what we actually do is we draw a pcb layout in OrcaD, and through that, we can relate it to MAtlab as above.. ^_^

 

[crys]

there is already a thesis about a pcb drilling machine using x and y coordinates, and it had huge errors once it it activated. so we are optimizing their thesis and would use polar coordinates to be more accurate.

 

[Nigel Goodwin]

there is already a thesis about a pcb drilling machine using x and y coordinates, and it had huge errors once it it activated. so we are optimizing their thesis and would use polar coordinates to be more accurate.

So the previous students were completely inept?, and did a poor job of it, I don't see any reason why polar would be any better, it's just a question of how well you implement the machine and software.

 

[yngndrw]

My guess would be that they had backlash. Remove that and it would probably be perfect.

 

[crys]

So the previous students were completely inept?, and did a poor job of it, I don't see any reason why polar would be any better, it's just a question of how well you implement the machine and software.

it was working, only that it made errors.. what we want is accuracy and precision.. polar is actually much better because of this article from elektor. basically, this is where our ideas stemmed from.

Cartesian coordinates: a heresy
Has anyone every seriously wondered why CNC machines always work using the Cartesian coordinate system? Why are X-, Y- and Z-axes always used? Why, when such machines are so difficult to build? The linear guides must be absolutely parallel, because otherwise the carriage will jam. The axes must be at exactly 90 degrees to one another, or else everything goes askew. The table must be absolutely true and the whole machine must be solidly fixed to a base. These are all disadvantages. But the greatest disadvantage is the linkage between the axes. Consider the X-Y table, the original form of and-operated milling machine. This has two handwheels, one to move the table in the X-direction, the other to move the table in the Y-direction. There are thus two linear guides, one fixed at an angle of 90 degrees to the
other, supporting one another. If the lower mechanism has play in it, this is transferred to the upper one, even if the upper one is absolutely precise. And the lower guide also has to bear the weight of the upper one. This traditional mechanism stems from a time before computers, when positioning along the axes was controlled manually by a technician using handwheels. Technical drawings are normally marked up with XY coordinates so that the successively required positions can easily be reached by use of the handwheels. In the age of automation the technician is no longer employed and the handwheels are replaced by motors under computer control. But the coordinate system has not changed: in the human imagination everything has a length, a breadth and a height. Curious, when most human actions are polar: ‘take three steps in this direction and then turn right’!
Imagine now how you would drill a circuit board by hand without a pillar drill. With the fingers of one hand you would hold the board steady and with the other hand you would hold the drill. Your drilling arm would be rather higher at the elbow than the other arm, since the mini-drill has a certain height. But you do not move your arms in the X- and Ydirections — no, you turn your drilling arm about the pivot at the elbow and turn and slide the circuit board to suit. You optimise your movements using your visual system — not perfectly, however, as sometimes you might miss a hole hidden by swarf. You do not need a firmly fixed base on which to work; your drilling arm is fixed at the elbow pivot, and what is between this pivot and the circuit board does not matter. Even a small tool between the two makes no difference. Your arms and your sitting position need not be absolutely parallel, or even anywhere near, and there are no 90 degree angles to be seen: two pivots are
enough!

 

[Nigel Goodwin]

it was working, only that it made errors.. what we want is accuracy and precision.. polar is actually much better because of this article from elektor. basically, this is where our ideas stemmed from.

Sorry, but I don't really see your optimism?, if you build your machine as badly as the previous team, then it will perform badly - it can only perform as well as it's built - and X,Y,Z machines work perfectly accurately, if built correctly. As suggested, a major problem with either type is backlask - you must cure that.

 

[yngndrw]

I'm with Nigel on this one.

The article seems to suggest a different way of building a machine to minimise sloppyness. It is basically saying "We cannot make two rails parallel, so let's build a work-around". I understand that you have a deadline and may not want to have a change of plan, but I'll comment anyway.

A gantry based machine (Using the Cartesian coordinate system.) will usually be more stable. The gantry for example is supported from both ends and there are no large massive which are supported a long way from their center of gravity. (I.e. Unlike a long arm.) Making mechanical components which are free from slop and backlash is not hard, it is just good mechanical design. Using pairs of back-to-back angular contact bearings on leadscrews and double leadnuts separated by a spring can reduce these. Remember that as far as aligning and setting up goes, the Cartesian coordinate system is actually easier. Making rails parallel and perpendicular to each other is very simple. I suspect that aligning a Polar coordinate machine is not as simple, as the two arm's axises must be parallel. (Am I right about that last comment ?) There are reasons why 99% of the commercial systems use the Cartesian coordinate system. Also remember that drill files are usually specified in the Cartesian coordinate system and the conversion may produce rounding errors if you're not careful. It may also be harder to remove backlash from a rotational axis.

Take a look here for an idea of removing backlash:
**broken link removed**

Just a thought.