angle meter and pcb drill

Thread Starter

mik3

Joined Feb 4, 2008
4,843
hi guys,

I want to monitor the angle of rotation of a rotating disk (disk speed between 30-100rpm). Do you know any sensors which suit for this application?

I thought to build an automatic drilling machine for pcb's. Any ideas on how to detect where the holes to be drilled on the pcb are located and how to determine the size of the drill bit?
 

Caveman

Joined Apr 15, 2008
471
Do you mean the rate of rotation, or do you want an actual angle at any certain point in time?

As far as the drilling machine, you are effectively making a CNC milling machine. Usually, a leadscrew will allow you to accurately place the drill.
 

Thread Starter

mik3

Joined Feb 4, 2008
4,843
Do you mean the rate of rotation, or do you want an actual angle at any certain point in time?

As far as the drilling machine, you are effectively making a CNC milling machine. Usually, a leadscrew will allow you to accurately place the drill.
I will set a reference point and measure the angle according to it.

Leadscrews are a nice idea for positioning but i dont want to use a computer to drive the machine. I want to put the pcb on it and find the pads itself and drill them. Is that possible? How can i translate my circuit design on Ultiboard for example to instructions for the machine?
 
Last edited:

Pich

Joined Mar 11, 2008
119
Ultiboard will translate drill information. Select file, export, NCDrill it will generate a Gcode (x and y coordinates). Please give more info on rotation disk the rotation disk
 

beenthere

Joined Apr 20, 2004
15,819
Sounds like you would need the X-Y drill coordinates translated to polar for your scheme to work.

Do we understand the method of movement to be swinging the PCB on a pivot, and perhaps a linear positioner for the drill?

Any CAD program that generates Gerber files will output a drill file. The file specifies each hole location in an X-Y grid and also contains the bit size information for the holes. This has been standard for many years.
 

Thread Starter

mik3

Joined Feb 4, 2008
4,843
Ultiboard will translate drill information. Select file, export, NCDrill it will generate a Gcode (x and y coordinates). Please give more info on rotation disk the rotation disk
About the disk, imagine the ordinary x-y axis system. Lets say we start monitoring the angle according to the positive x axis and moving counterclockwise. I want a sensor to measure this angle
 

Thread Starter

mik3

Joined Feb 4, 2008
4,843
Sounds like you would need the X-Y drill coordinates translated to polar for your scheme to work.

Do we understand the method of movement to be swinging the PCB on a pivot, and perhaps a linear positioner for the drill?

Any CAD program that generates Gerber files will output a drill file. The file specifies each hole location in an X-Y grid and also contains the bit size information for the holes. This has been standard for many years.
Ok thanks. I will search the internet to find out how you read this data from a PC port. Do you know any good resources?
 

beenthere

Joined Apr 20, 2004
15,819
Sorry, I can't interpret "how you read this data from a PC port". Do you reger to the Gerber drill file, or the angular sensor?

I gather you will rotate the drill table about a pivot. How will the drill get positioned across the radius?
 

Papabravo

Joined Feb 24, 2006
21,159
.... i dont want to use a computer to drive the machine. I want to put the pcb on it and find the pads itself and drill them. Is that possible? How can i translate my circuit design on Ultiboard for example to instructions for the machine?
These two requirements are diametrically opposed. You don't want to use a computer to drive the machine. "Finding the pads" seems to imply some kind of vision system. These systems are most certainly based on computers. I see no viable method for doing what you want to do except with a computer. I could be wrong, but I'll give long odds on my position.
 

Caveman

Joined Apr 15, 2008
471
I will set a reference point and measure the angle according to it.

Leadscrews are a nice idea for positioning but i dont want to use a computer to drive the machine. I want to put the pcb on it and find the pads itself and drill them. Is that possible? How can i translate my circuit design on Ultiboard for example to instructions for the machine?
If you use a non-cnc milling table, you can dial in the XY coordinates from your CAD program. Usually you have calibrated dials, so each rotation is 100mils, and you have 1mil ticks marked on the dials.

How much do you want to spend?
 

Caveman

Joined Apr 15, 2008
471
I just thought of something a bit more manual. If you overlay a projection of the drill file onto the blank pcb, then using a punch or marking tool, mark where each hole is needed. Then using a standard drill press, you can drill them.
 

Thread Starter

mik3

Joined Feb 4, 2008
4,843
These two requirements are diametrically opposed. You don't want to use a computer to drive the machine. "Finding the pads" seems to imply some kind of vision system. These systems are most certainly based on computers. I see no viable method for doing what you want to do except with a computer. I could be wrong, but I'll give long odds on my position.
I mean i dont want to control the machine with the PC. I want to make it with vision as you said and i will use microcontrollers to control the machine (which is a small computer, you are right). I would like to know how i can implement this vision?
 

Thread Starter

mik3

Joined Feb 4, 2008
4,843
I just thought of something a bit more manual. If you overlay a projection of the drill file onto the blank pcb, then using a punch or marking tool, mark where each hole is needed. Then using a standard drill press, you can drill them.
Yes thats what i want, a type of 'vision' to locate the holes to be drilled and their diameter. Any suggestions?
 
Last edited:

beenthere

Joined Apr 20, 2004
15,819
Only that it's about 4 orders of magnitude easier to locate the PCB to a 0,0 point and drill using X and Y offset to each drill point.

Using a camera is not realistic - it would take enormous processing power (from a microprocessor's perspective) to go through the video feed and distinguish pads from PCB.

There may be a wavelength that makes it very easy to distinguish FR-4 from copper. If not a camera, what visual sensor can find a pad (and then the hole in the pad)? And do it in such a way that a drill would accurately hit the hole.

Using "vision" means having to clear crud off the PCB with each hole drilled.

How does the visual sensor find the next pad? Does it roam like a Roomba, or follow a grid? How does the uC tell if the drill bit is the right size for the located pad? When the right size is changed, does it have to search all over again, or can it hold located pads in some memory map?
 

Thread Starter

mik3

Joined Feb 4, 2008
4,843
Only that it's about 4 orders of magnitude easier to locate the PCB to a 0,0 point and drill using X and Y offset to each drill point.

Using a camera is not realistic - it would take enormous processing power (from a microprocessor's perspective) to go through the video feed and distinguish pads from PCB.

There may be a wavelength that makes it very easy to distinguish FR-4 from copper. If not a camera, what visual sensor can find a pad (and then the hole in the pad)? And do it in such a way that a drill would accurately hit the hole.

Using "vision" means having to clear crud off the PCB with each hole drilled.

How does the visual sensor find the next pad? Does it roam like a Roomba, or follow a grid? How does the uC tell if the drill bit is the right size for the located pad? When the right size is changed, does it have to search all over again, or can it hold located pads in some memory map?
It will scan all the pcb and remember which holes are the same size and drill them. Then it will change bit.
Do you think it is easier to control it directly with a CAM software form a PC rather than to be independent and be controlled by microcontrolers?
 

jpanhalt

Joined Jan 18, 2008
11,087
I thought that might be what you wanted to do. After all, all you need to do is take a video image, convert to pixels, write some software, etc. I am not being sarcastic, because that was an approach I tried in 1983 when I needed to scan a disk that was 10 cm in diameter, pick 37 colored spots on it, and measure their light density. Of course, in 1983, PCs and video capture were slow. We did get it to work. The main problem we had was that although a video image looks nice and straight, it is not. Light reflections/parallax was our second problem. We did not try a flying dot scanner or line scanner. In the end, we mounted the disk to the front of a monitor, used the monitor as a light source to highlight the approximate spot location, and a person read them. It was much faster.

I wonder if today, one could use touch screen-type technology to capture hole coordinates for a board attached to a monitor screen. In the alternative, with a bright light under the PCB, you might be able to detect the etched holes with some sort of offset light detector, center to max intensity, then drill at the location corrected by the known offset.

If you use a program like Eagle, it can make a small starter hole for each drill (use the ULP drill-aid.ulp). That is what I do. The drills center nicely, and unless I am way off, the holes end up centered too.

A sensor- or scanner-based approach sounds like an immensely difficult thing to develop compared to manually drilling or using conventional CNC.

Best regards, John
 

Papabravo

Joined Feb 24, 2006
21,159
I think you have no chance to make machine vision work. There were at least three companies in this area that were started in the 1980's and they're all gone now after burning through several tens of millions in VC.
 

jpanhalt

Joined Jan 18, 2008
11,087
There were at least three companies in this area that were started in the 1980's ...
I know of a forth:D Based on my work just described, I was asked to do some consulting for it. The company actually got a product it could demonstrate at a trade show, but I don't think it sold a single instrument. Later, the whole company was sold.

I think we are all in agreement. It is a lot tougher to do than it first appears to be.

I hate drilling holes too. The easiest solution is to use big SMD parts. John
 

Papabravo

Joined Feb 24, 2006
21,159
My three were

MVI - Machine Vision International
SVS - Synthetic Vision Systems
AISI - Applied Intelligent Systems Inc.

What was yours?

I think it was MVI that wanted me to do some engineering for it at the beginning.

I just rememberd the forth

Perceptron

For all I know it may still exist in some form.
 
Top