Ok here we go. I reassembled my machine after moving and added a bigger power supply and all of a sudden the machine looks like it is drunk. I am now changing all the control cables to be fully shielded as I believe that this that the unshielded cables in close proximity to the controller are the cause. Power supply is 52 volts at 50 amps.
Go home CNC, you’re drunk! Your machine is probably drinking too much ripple. If it were me I’d run more in the way of a filter capacitor. What’s that one you have there about 6,000 μF ? There’s a formula to calculate how much capacitance you need for X current. But I just tend to guess high myself. Probably not your problem, but some more filtration never hurts. You likely need about 10X what I can see there. Somewhere in the range of 60,000-100,000 μF.
Like Hans and Franz says, All we want to do is pump you up!
@Paul_Frederick Guess again Paul. If you look carefully only the Y axis is affected. I tried different software to eliminate that from the analysis. I am putting money on the unshielded cables being the problem and these will be changed after work tonight. See picture…
I have bigger caps that I can fit so I will let you know and thanks for your input.
@Paul_Frederick So I changed the power supply back to the original 48VDC regulated supply and this made no difference. Then I replaced the Y axis motor cable with the screened cable and that resulted in a slight improvement. I am now thinking I may have to change all the cables to the screen type.
@Robert_Ritchie that does sound like it rules out the PSU as the culprit. When I designed the break out board that I use I made sure I kept some drive strength on the control lines, so it would not be so sensitive to electrical noise. I use one of those Chinese air cooled spindles with a SMPS to run it and those things kill Arduino setups. But it has no effect here. I run it on a pair of unshielded wires, and I never bypassed the motor with capacitors either. I am sure it is broadcasting all kinds of EMI. It never caused me any problems so I never really checked though.
I think I have the power cables a little twisted? Twisting lines makes them not couple so much. That’s something the phone company discovered. They actually patented it when they did. Parallel wires inductively couple with each other. Sort of like how those cordless chargers work? But in signal electronics it is called crosstalk.
Electrical interference and noise mitigation is a somewhat lengthy topic. Other than shielding you can use ferrite beads too. Bypass capacitors. Star grounding. I had to resort to optical isolation on my motor control lines. This is supposed to be a square wave http://i.imgur.com/JVyX6n5.jpg
@Robert_Ritchie when I had a problem with one axis it turned out to be the lead screw nut. I initially thought it was electrical in nature, but it turned out to be mechanical. Examine these two plots http://i.imgur.com/vVdcSLx.jpg That was caused by my X axis lead nut being stripped out.
K M wmm
@Paul_Frederick After work today I will try a few changes to the spindle wiring and see if that improves anything. If there is no improvement then I will go the full shielded rewire of the machine. I have to keep my posts short at the present time as I have no internet at home except my phone. Also next machine I want to run Linux CNC on USB so I will have a few questions for you about that.
@Robert_Ritchie
as of now LinuxCNC does not support machine control over USB. http://wiki.linuxcnc.org/cgi-bin/wiki.pl?HardwareDesign#USB So if that is your goal you will have to use something besides LinuxCNC. GRBL is very popular these days.
@Paul_Frederick ok a bit of misunderstanding here. What I mean is I would like to run Linux CNC from a bootable USB disk.
@Robert_Ritchie yeah even that doesn’t actually work. I mean it works. But you can’t get decent real time performance off USB. People do run real time LinuxCNC instances off MMC flash cards though I think. I’m not entirely sure why that works, or what the major difference there is. They do it to go diskless for environmental reasons. The hot setup is using a vintage Intel Atom mini ITX board and building it right into a control box. For some reason old Atoms have sick latency speed. The only thing I know of faster than them are Xeons. But you have to get an Atom board with the right video hardware on it. Later ones had problems.
There’s a guy named Andy Pugh out there that did it and could tell you why it works. He’s one of the LinuxCNC devs. He’s done insane stuff with LinuxCNC just to do it. Like cutting gears with an encoded spindle. There he syncs two spindles together. He did that on a lathe mill combo machine. Just crazy stuff. But he posted an image of one of the control boxes he made with an Atom built into it that I can recall. And I’ve chatted with him about it too. Here’s crazy Andy https://www.youtube.com/watch?v=ZhICrb0Tbn4
Problem found. Somehow the breakout board has failed most likely due voltage spikes. New boards have been ordered as I am keen to get my machine up and running again.
@Robert_Ritchie I made my own BOB. It is still hanging tough too. http://i.imgur.com/2frKz9O.jpg
