Wobblycogs CNC Mk2

[Jonathan]

Occaisionally, I check the squareness of my machine by drilling 3 holes to prescribe a right angled triangle and accurately measure their center distance to determine skewness, and proceed to compensate back to square. I've taken to putting marks on the rotating ballnuts to aid the correction process.

I've never seen the machine loose or gain even just one microstep on either motor, except when I've crashed it. That's including on a job where the machine was left running for the best part of 47 hours.

[Wobblycogs]

Thanks both.

So, I really like the look of the CSMIO/IP-M but the lack of built in support for slaving makes me nervous. From the manual it looks like it could be done via Mach3 but it's not at all clear what the consequences of doing this would be e.g. would homing / squaring work correctly.

I would happily just ditch one of the x-axis steppers (less hassle, cheaper, etc) but I can't see a way to combine an open back to the machine and a belt drive system without ending up with belts all over the place or the set up being impossible to tension. Any ideas would be greatly appreciated.



I should say I'm not totally fixated on having an open back to the machine, it's nice to have and may make it more versatile but it's not the end of the world if it goes.

[Jonathan]

If the output from the breakout board can't source enough current for two driver inputs, then you can buffer it with just one transistor and two resistors and that's guaranteed to work.

[JAZZCNC]

Occaisionally, I check the squareness of my machine by drilling 3 holes to prescribe a right angled triangle and accurately measure their center distance to determine skewness, and proceed to compensate back to square. I've taken to putting marks on the rotating ballnuts to aid the correction process.

Ok but why would you need marks on rotating nuts to aid correction if it never loses a step or position.?? Should be nothing to correct.!!

I've never seen the machine loose or gain even just one microstep on either motor, except when I've crashed it. That's including on a job where the machine was left running for the best part of 47 hours.

But what was you cutting and at what feeds.? . . Aluminium I presume.! . . . Try cutting wood with inertia of rotating screws not rotating nuts around 7-8mtr/min for 47hrs and check it.!!
Also your machine is setup correctly and presumably with plenty of torque left, I'm talking higher feeds and motors tuned close to edge which often happens causing lost steps and much sooner than 47hrs.!!

Again I'll say it " Tune the motors properly with plenty of safety margin and there's no problem with slaved motors" but still it can happen and that's not good and healthy for screws or machine if one motor stalls racking gantry at rapid speeds.!!!!. . . . With belts running out of sync is not even in the frame and stays set just like the day it was set regardless of crashing or motor tuning.!! . . . .And trust me I've crashed hard and plenty so it's well tested. . .Lol

[Jonathan]

Ok but why would you need marks on rotating nuts to aid correction if it never loses a step or position.?? Should be nothing to correct.!!

The marks are there so I can check that it's square when I switch the motors on, as clearly they can move when turned off. I could do the same with switches, but a permanent marker is cheaper!

But what was you cutting and at what feeds.? . . Aluminium I presume.! . . . Try cutting wood with inertia of rotating screws not rotating nuts around 7-8mtr/min for 47hrs and check it.!!

Not aluminium. The feedrates were low and the acceleration set quite high. The result would be the same if I ran my machine for that amount of time with the maximum feedrates, because the motors are tuned properly. Yes, I could set the rapid feedrate slightly higher and loose the odd step here and there, but there's no reason to as I chose the correct size motors for the machine, so the safe feedrates are adequate.

Also your machine is setup correctly and presumably with plenty of torque left,I'm talking higher feeds and motors tuned close to edge

If that's the case then they should buy the right motors and set up the machine correctly in the first place, instead of linking the screws with a belt so it doesn't matter if there's a missed step.

How do you accurately set the machine square when a belt is linking the screws?

[JAZZCNC]

If that's the case then they should buy the right motors and set up the machine correctly in the first place, instead of linking the screws with a belt so it doesn't matter if there's a missed step.

Yep agreed but that's not always the case and some don't have knowledge to setup correctly from beginning and damage can be done before they learn.!! Plus there's the other added benefits using belts give like piece of mind a stall isn't going to turn gantry into a Twizzler.!! . . . Anyway it's bit Linux vs Mach thing we are obviously going to have to agrees to disagree.!

How do you accurately set the machine square when a belt is linking the screws?

Loosen bolts on gantry etc and pulleys then turn ball screws . . . . Simplizz.!! . . . Bit fiddly and time consuming at first but one time deal so no problem.

[Wobblycogs]

Well it's been a busy day for the credit card. The drivers, steppers, ballscrews and nuts, bearings and mounts, spindle and vfd have all been bought mostly from China. I think I might have accidentally bought a collet set as well - I asked for a spindle with a ER20 collet and I think he's read that as "collet set". Not a problem though as I was going to buy a set anyway, I hadn't realized Chai had started sell them. If anyone needs drag chain he seems to be doing some nice looking stuff at a reasonable price.

I've also been tempted over to the dark side, I installed LinuxCNC on one of the old machines I have kicking about last night and I have to say I like what I see. There were a couple of minor complications but over all it seems to be a nicely put together system. I ran the latency tests for a couple of hours with a fully stressed machine and achieved ~7,000ns max jitter so I think it should work quite well. The parallel port supports SPP, EPP, ECP and ECP + EPP modes. I've seen someone mention that one of EPP or ECP is better but I can't find the thread again.

I was looking closely at the electronics in Jonathans "sufficiently strong" build and I can't see a BOB anywhere in the control box. In fact it looks like the drivers are connected directly to the parallel port with the cable being wrapped in a bit of foil. Is that correct? I've not been able to find all that much mention of people running steppers direct from the parallel port, just the odd post here and there but I wouldn't mind giving it a go. As has been mentioned elsewhere I can always move onto using an ESS + BOB if necessary but I might as well give it a shot.

[Jonathan]

7000ns is exceptionally good latency, what motherboard etc are you using?

I was looking closely at the electronics in Jonathans "sufficiently strong" build and I can't see a BOB anywhere in the control box. In fact it looks like the drivers are connected directly to the parallel port with the cable being wrapped in a bit of foil. Is that correct? [...] but I might as well give it a shot.

I left Sasha to wire up the control box and he had problems with the breakoutboard, so for now the stepper drivers are indeed connected directly to the port. The output voltage of the port is sufficient so switch the driver inputs reliably and the driver inputs are opto-isolated, so there's no chance of damaging the port. To get isolated inputs it's easiest to just get a breakout board. In short, it works... but that doesn't mean it the best idea. So by all means give it a shot, but I wouldn't advise using this setup permanently.

Not sure why the foil is there, maybe he ran out of shielded cable but I thought I lent him a 100m reel!

[Wobblycogs]

Yes, I was quite pleased with the latency. Looking at the machine spec I'd forgotten quite how good it was, massive overkill for a control box I know but it was sitting around not doing anything. In fact it's been switched off for so long the BIOS battery needs replacing. I'll replace that and re-run the latency test to confirm the result.

The machine is:

Motherboard: GA-K8NXP-SLI
Processor: Athlon 64 3000+ (not sure which revision, I think it's a 130nm ClawHammer as I bought it quite soon after they were released)
Memory: 3GB DDR400
Video Card: some old nVidia card, can look it up if you're interested.

Initially I was getting a latency of about 50,000ns but I noticed that was caused by a spike when I first played sound through the on-board sound card. Disabling the on-board sound in the BIOS gave me a 7000ns. After performing all the package updates the system required I re-ran the tests and got around 6500ns. I also tried the nVidia binary drivers to see if that would improve video performance - bad idea, whenever anything using OpenGL started it caused a 250,000ns spike in jitter!

I hear what you are saying regarding running direct from the parallel port. Would you also recommend the PMDX-126 as a nice BOB?