LPKF 91S
November 2010
I was reorganizing the lab that I work in, and I came across a CNC PCB router that looked like it hadn't been used in ten years. I really enjoy restoring old machinery, especially if it could prove to be useful once working, so I decided to see if I couldn't get this machine up and running again. After looking up the machine online (LPKF 91s), I found that it was manufactured in 1995 by the german company LPKF. LPKF makes various PCB solutions for industry, including routers for high precision milled circuit boards (mostly used in the RF industry). Turns out, this router was a very expensive router when new, and even though it was a little old and dusty, the hardware (linear guides and lead screws) was still well oiled and rigid.
After searching online I found that I would need a piece of software called BoardCam to run the router. I connected the router to my PC via a serial cable at 9600 bps (this tells you just how old this machine is). After installing the software, I tried numerous times to connect to the machine without success. After a few simple tests with the COM port on the PC, I convinced myself that there was something wrong with the electronics on the machine side. I proceeded to open the machine up and see if I couldn't find a blown component, a cut trace, or something that could cause the symptoms I was seeing. I checked the fuses (they were fine), I check the conductivity of the serial connector to the first buffer chip inside the machine (fine), I checked the power to the buffer chip (fine), I opened up data-sheets for all the components on the PCB used for serial COM and devised tests to make sure they were working (fine)... I went on like this for many hours trying to find something that wasn't working, but everything seemed fine. Eventually I got frustrated and called LPKF customer support and after the usual: is the power on? Can you cycle the power please? Are you sure you've plugged the serial cable in correctly?, customer support said to send the machine in and for 1500 dollars they'd replace the main board. I sure as hell wasn't gonna do that.
I then tried reinstalling the software and realized I had chosen the wrong setup file. Sure enough, the router worked perfectly once I installed the software correctly. The lesson here is you should debug the easy things first. Instead of taking the machine apart and spending hours probing traces (as much fun as this might be), I could have reinstalled the software in 3 minutes and debugged other easier things in a much shorter time. Also, unless you have firm reason to believe that the electrical hardware is not working, it most likely is. This story reminds me of the time when I replaced a light switch for a friend only to find out the lightbulb was blown... (lesson here is never trust a friend's diagnosis)
While I was interning at Apple over the summer, I used their microscope to take some high resolution photos of a sample trace PCB I cut using the router. I wanted to see up close the quality of the traces and whether they were the thickness that I was asking for in my CAD software. The traces should be of the following thicknesses: 0.05", 0.04", 0.036", 0.032", 0.025", 0.018", 0.0125", 0.010" (from right to left). I placed a surface mount 0201 resistor in some of the photos for size comparison. As you can see, they aren't quite the right size. This is because the LPKF routing software uses the width of the routing bit to offset itself from a trace and I didn't know exactly how wide my bit was. I use a 60° single lip engraving bit, which means that as I cut deeper into the copper, my cut path widens. This leads to variable trace thickness depending on how deep you cut. I'm still looking into a way to set my depth of cut accurately every time.