First milled PCB

[pminmo]

Phil, I continued to read the pages after the one you've mentioned. At some point you write about homing the machine to x0,0 and y0,0 by using limit switches. I was thinking about that problem, homing the machine, and thought about the limit switches but I had my doubts about them. Is there not an over-shoot that can be different every time. Not by much, mind you, but still and over-shoot. Am I right about it or am I wrong? I hope wrong because then I'll use it in my upcoming machine.

Yoram

The repeatabilty of the switches is really good, much better than I ever imagined. I've had no problem using them alone.

As for pcbgcode efficiency, it's pretty bad. There are some folks working on path optimizers, one is posted in the forum. They help considerably!!

[acondit]

Phil, I continued to read the pages after the one you've mentioned. At some point you write about homing the machine to x0,0 and y0,0 by using limit switches. I was thinking about that problem, homing the machine, and thought about the limit switches but I had my doubts about them. Is there not an over-shoot that can be different every time. Not by much, mind you, but still and over-shoot. Am I right about it or am I wrong? I hope wrong because then I'll use it in my upcoming machine.

Yoram

It depends on how you set up your limit switches. With optical home switches you can get very precise repeats. With a home switch plus an index on the leadscrew you can also get very precise results. Probably on the order of a couple of tenths. Again it all relates to how precise you build your machine to be.

Alan

[pminmo]

I got it down to only 2,200 lines from over 11,000 but it is still not efficient. I had made PCBs using the ink transfer and had the board prepared with photoshop to look like the attachment. If I knew how to convert that file to g-code I'll have a file that will cut in less time.

Yoram

I've not persued this route, but I'm pretty confident that you can convert the image to dxf and then use something like cambam or ncplot to CAM your cutter paths. You can also use polygons in eagle, but it takes some design time and thought. There are also some eagle ulp's for image import, but I suspect they make an ugly set of paths, but I don't know. Doing the polygons in eagle leaves you with precise drill placement.

[pminmo]

It depends on how you set up your limit switches. With optical home switches you can get very precise repeats. With a home switch plus an index on the leadscrew you can also get very precise results. Probably on the order of a couple of tenths. Again it all relates to how precise you build your machine to be.

Alan

I've put a dial indicator on my setup with plain jane limit switches and never seen more than 2 mils. A good friend has done the same with similar results on his machine. And in truth, I think backlash is probably the biggest contributor to that.

[acondit]

Phil,

I am generally in agreement with you. I was merely making the point that good design and careful implementation are important. Repeatability is important if you are faced with stopping the machine in the middle of a run and restarting the same run. For many of the things I do a couple thousandths is no problem. However, for something like a bearing fit it is important. And for cutting bearing pockets on opposite sides of a piece it is critical. I suspect the same is true for cutting double sided pcbs (just guessing).

If one wants to get the home positioning repeatability down to the order of 0.0005 or less, it is do-able on a diy machine, though it does cost more. Generally, the more precise the more the cost. I also agree that you probably have to deal with issues of machine rigidity, bearing fit and backlash if you want super precision on your machine.

I think that at the moment the backlash on my machine is probably well in excess of 0.002" but when I adjust it periodically I am able to get it down to about 0.001". My problem is that I have a nut that works loose. Because I am in the process of rebuilding the machine I don't have a lot of incentive to fix the problem with the old design.

I guess the most important thing is to be able to analyze what your needs are and, if the machine is not meeting those needs, be able to determine what needs to be fixed. If it is meeting your needs then who cares if it has 0.010", 0.001" or even 0.0001" repeatability (other than bragging rights).

Alan

[pminmo]

Phil,

I am generally in agreement with you. I was merely making the point that good design and careful implementation are important. Repeatability is important if you are faced with stopping the machine in the middle of a run and restarting the same run. I suspect the same is true for cutting double sided pcbs (just guessing).

It is, although I'm amazed at how easy it is to make minor alignment tweaks visually, and I've got 56 year old eyes with bifocals on. (magnifying headset helps though)

If one wants to get the home positioning repeatability down to the order of 0.0005 or less, it is do-able on a diy machine, though it does cost more. Generally, the more precise the more the cost. I also agree that you probably have to deal with issues of machine rigidity, bearing fit and backlash if you want super precision on your machine.

Alan

I think we are in complete agreement, I just wanted to point out that my experience at least is that the simple lever limit switches are much better than you would think. I don't know the internals of the switches, but obviously by the external design they get a mechanical amplification of movement to the actual mechanism via the arm.

For milling pcb's, absolute accuracy isn't as much of a problem as sloppyness of the mechanicals. Simple 1/4-20 all thread is plenty accurate, but if there is mechanical sloppyness it's all irrelivant. I notice it when nuts wear to the point you get near 6 or 7 mils of backlash and you are trying to mill a pcb that has pretty small stuff. 20 mil traces and clearance and above, you can have 7 mils. 7 mils isn't pretty, but it will still produce a working pcb.

[pminmo]

Using your machine to make a smd stencil:

"http://www.youtube.com/watch?v=Mq-JrfIW24s"]YouTube - DIY SMD Stencil

[TigerPilot]

Sorry if I sound like an ignorant, Phil, but what do I do with the stencil?

Yoram

[pminmo]

Sorry if I sound like an ignorant, Phil, but what do I do with the stencil?

Yoram

Apply solder paste for smd parts for reflow. (soldering) You need to apply paste only in area that have pads that smd devices will be placed prior to reflow. For small boards it's not hugely necessary, but as the number of components increase, the easier it makes the process.

[TigerPilot]

smd devices

Isn't this a redundant redundancy?

Do you put the stuff then in an oven or do you use a heat-gun or what? In the video you first cut the mdf board. Why? Couldn't you glue it without milling the board first? I guess the mdf board has to be replaced for every new design. In my neck of the woods I didn't find a source for mdf boards yet so that I have to treat each piece that I can lay my hands on as if it is made out of gold.

[pminmo]

Do you put the stuff then in an oven or do you use a heat-gun or what? In the video you first cut the mdf board. Why? Couldn't you glue it without milling the board first? I guess the mdf board has to be replaced for every new design. In my neck of the woods I didn't find a source for mdf boards yet so that I have to treat each piece that I can lay my hands on as if it is made out of gold.

Oven or heat gun depends on quantity and size of pcb. You can skip on the mdf slots, but the glue disperses more and takes more effort to remove the brass because of it. The brass is very thin and flimsy. The channels also give you the area that is going to be milled, and thats the most important areas for the glue. MDF scrap is all over my workshop,,,,,,,,,,,,,

[electrokean]

Phil,

This is very interesting - something I hadn't even though of, and will have to try.

I enjoyed the background music! And it is kind of "holey".

Kean