Cutting double sided with 3018

[Thunder]

Several issues have been found with the 3018 CNC. While it is a good machine for the price the controller seems to let it down somewhat. Inconsistent cutting depth, shifting zero on Z axis and top and bottom layers misaligned are the common ones. So far I have tried 4 different controllers with different results. So far I have decided to use the best one for what I am cutting at the time. Apart from that cutting accuracy seems to be good. I use .1mm mill cutter and down to 0.3mm drill with no issues. It seems that I have to keep trying more controllers until I find one that dose everything correctly.

Any other thoughts?

[abeyer]

I suspect top/bottom alignment is as likely to be due to reliable and repeatable fixturing as to the controller board.

[langwadt]

those issues are unlikely to be cause by the controller, much more likely it is mechanical

[Thunder]

I suspected that too. 3 different controllers with the same setup and the same PCB files but with different results. Had to be anything but mechanical. I have ordered another controller so we will see what happens with that.

[PlainName]

As everyone else says, probably not the controller. Stick the board down with thin double-sided tape (carpet tape is my preference) - any slight bowing will screw up depth-critical stuff, and the tape makes sure it is all flat.

But first, use a sacrificial bed and make sure it is square to the Z-axis. I use half-inch perspex which was fly cut in situ. Good stuff for wiping up oil spills (you'll be wanting to use oil as a lubricant too, if you want your cutters to last any time).

[My mill is a Sable 2015 so although quite old relatively it is still recognisable as the same thing.]

[Kean]

When I was doing double sided PCBs on my Taig mill and 6040 router, I would use 0.8mm single sided PCB and mill them separately.
When placed back to back the registration was perfect. A few 0.35mm holes and some fine solid wire would stitch together the ground planes and vias (just not under SMD parts).
I also used LPKF copper rivets on some designs - they can be punched down quite flat to go under parts, but you need to check continuity and sometimes add a tiny bit of solder.

[Thunder]

The latest update. I ordered another controller and the cuts were at the correct depth and all holes lined up correctly on both sides of the board. This time it appears that I have 3 faulty controllers.

[PlainName]

Well, I never! Just goes to show.. er.. something 

[Doctorandus_P]

I still find it highly unlikely that you have 3 (or 4?) faulty "controllers".

To me it's more likely you have a faulty connector, faulty solder joint in your wiring, a broken wore or a (partly) defective motor or whatever may be the real cause, and swapping a controller just (temporarily) puts some pressure on the fault so you do not see it at the moment.

Intermittent faults can be very hard to diagnose correctly.

[Thunder]

In the end it seemed to be the only logical choice. If it was mechanical or motors then on this 2 month old machine every motor and every connection and every wire would be faulty because the faults were so varied and seemingly unrelated. Losing X,Y 0 on one controller but everything else worked consistently ok. On another controller X, Y was good but loose Z 0. I felt that if it was a machine fault rather than controller then I would have 3 faulty motors and all the wiring would be faulty too. I felt that that was unlikely.

[PlainName]

Not necessarily. The thing about bad connections is they work or don't work on a whim, and doing something unrelated can seem to fix them for a while. Or break what seemed to be a good connection. You would need to replace the controllers and see exactly the same symptoms as each had last time they were in, and preferably do that twice just to be sure.

[Kean]

It does sound more like wiring or power supply issues to me.  Shielded wiring to the spindle and axis motors helped a lot on my machines when losing steps.

You don't say what controller you've been using - it may be under powered or bottom of the barrel quality.
I've almost always used Gecko drivers with Mach 3 in my builds, but a few had Arduinos and grbl.

[Thunder]

One thing that I didnt mention is that I tried the controllers on 2 machines and used the same setup and the result matched each controller.

[Thunder]

 

It does sound more like wiring or power supply issues to me.  Shielded wiring to the spindle and axis motors helped a lot on my machines when losing steps.

You don't say what controller you've been using - it may be under powered or bottom of the barrel quality.
I've almost always used Gecko drivers with Mach 3 in my builds, but a few had Arduinos and grbl.

Haven't thought about shielding the wiring so I will keep that in mind if it happens again. The power supply I have already upgraded from the noisy 5 amp to a very clean 8 amp supply. Checked on a scope and the new power supply has less than a 0.1V noise where as the original has more than 0.4V noise under load. The controllers are grbl and I have sourced them from different places which for now seems to have made a difference. Even though they look identical.

[Thunder]

Blown yet another controller. Changed the controller and replaced spindle cable with shielded type. Also rerouted the cable away from other cables. Will update once I have tested it tomorrow.