Hi Guys! New here... Love the forums and Open Builds! So I have cheap little 3018 that's powered by a Chinese Arduino / Woodpecker clone running GRBL 1.1h on an ATMega328 with a 500W air-cooled spindle, and I'm building a new machine and using the OpenBuilds electronics package with BlackBox controller (it's on order). I haven't received and tested it, yet, but I see it's also using an ATMega328. I've got to admit that I was a little surprised to see that it's using the same really old $2 processor, but it looks like a very professional and flexible implementation with great reviews. Using the cheap Chinese board, when I'm routing fine detail into a relief and I'm making several small moves using a v-bit, the processing is extremely slow. For example, a final pencil finish estimated processing time was 08:03 (same estimation by Candle and Fusion 360), but it actually took 02:21:27! Typically my toolpaths take 1.5X to 2X the estimation, and I've always chalked it up to the relatively slow acceleration on the wimpy machine. But those toolpaths usually have fewer lines, and the really slow example above had 140k lines. My questions: 1. Is the discrepancy in time due to the acceleration, or due to the limited capacity the processor? 2. Should I expect the BlackBox to perform any better, processing speed wise? I understand the stepper motor drivers will be an improvement, but will the throughput be any higher? 3. Regardless of the above, what steps do people take to reduce the complexity of their toolpaths so that they can get equivalent visual results that don't take so long to run? Thanks!
Sounds more like limitations to feedrates and acceleration - bad settings. Grbl is plenty fast! Definately, also, much better stepper drivers! No need to baby it along (unless of course the machine itself is bad - weak etc - then replace it with an OpenBuilds machine too (; ) Most cam software has an option for something like that (smoothing)
Thanks for your input! So, like I said, it's a wimpy machine. My max feedrates are set to 2250, 2250, and 1000 mm/sec and accelerations are set to 60,60, and 20 mm/sec^2 for x, y, and z, respectively. That's about all I've been able to get without dropping steps on complicated toolpaths. I imagine the accelerations are causing discrepancies between the estimates and the actual results, but do you think if acceleration weren't an issue that the processor could push through the 140k lines in 8 minutes? If I were to run grbl_sim, do you think that would give me a better idea of whether the limitations are accelerations or processing power? Also, I get my BlackBox on Wednesday, so once I'm set up I can try to push through the same code and see how the timing compares. I'm going to use the OpenBuild electronics package on New Carve that's on the way, but I would like to keep this dinky machine running for small jobs, and I'm considering upgrading the electronics if I can get better results. I use Fusion 360, and yes, it has a smoothing option, but it doesn't seem to make much difference. For this file, for example, it reduced the number of lines from 140k to 139k. Any other tips or tricks, or are those files sizes fine for a non-dinky build? Thanks again!
The smoothing settings make a difference. Also think about resolution, if the machine can do a minimum step of 0.025mm then using anything smaller that that in Fusion for smoothing or tolerance is not beneficial since the machine cannot move to that position.
