[charliex]

did a scan of about .07 2" x 1.8" at 3ipm, after looking at the point cloud i should have done a smaller more detailed run. oh well.

data.
http://www.goth.am/pc/data/logo.txt

modern browser+webgl you can see the point cloud SRV - Point Cloud Viewer just click on logo.txt at the side , straight out of flashcut (with comma delimit) keys 1,2,3,4 change colour, - amd + change point size.

i guess i should try some tool steel or marble.

[charliex]

Started putting together a SSR control box.

secondary small heatsink + thermal pad




immediate switch crydom 50A


used what i had on hand, i did want a one in, one out, but this will do for testing.


sliced some of the insulating out of the power coord and connected it to the relays terminals, hot side.


and it's too hot to continue in the garage !

[datac]

I have a wierd thing where flashcut just stops for no apparent reason, i can continue the job but it'll just stop as if the feed hold has been pressed, i emailed the tech support guy, see if they have any ideas.

Just stops ? No message box, no nothing ?

The only thing that might stop the machine gently and you are allowed to restart without issue would/should be a feedhold. In preferences, under messages, make sure to turn on the feedhold message option. If it is going into feed hold, you'll get the message. And, if it is going into feed hold, it could be an issue with your keyboard because feedhold is "hit any key".

If you have wired in a dedicated "Feed Hold" switch to an input ? It could be that it needs a little higher debounce set up.

Off the top of my head, nothing about probing jumps out at me that would cause that...... Definitely something Flashcut will get to the bottom of if you call them on it.

[charliex]

Hey Chris,

Yep just stops, no message, no nothing, just like a feedhold

Yeah i just have a keyboard on the computer, don't see any spurious keys. It'll be fine then just stop as if it was a feed hold, then continue i ran a 4 hour+ sample and it stopped about 4 times. Also had an odd one where it would say the probe wasn't triggering within the Z range, connected my meter up to it and it was definitely working, i left the meter on beep unless it probed and flashcut just wouldn't see it.

I emailed flashcut support the day i saw it, no reply yet. Which so far apart from an email about the software update username/password when i first got it, has been my experience with them, no answer to quotes either, well except the automated response saying they'll get in touch, three weeks later, zilch, wierd especially since it explicitly says on the website its the best way to get in touch with them. It sort of works in my favour for some stuff, since instead of buying the things, I end up making them!

i'm getting the automated emails, and the email about the download username/password so its not a spam/address issue.

[charliex]

dropped them a program crash debug log via email and mentioned the lack of comms, i got a return email and then the quotes i'd been looking for their limit switch/touch off probe price, 120 and 500 approx.

[charliex]

My mistake, the limit switch I got the quote at $150 was just for the wires and switches, the G0704 kit is $350 USD for 5 microswitches, wires and mounting plates..... so yeah.

[charliex]

made my own limit switch brackets, i know right..

https://hackaday.io/project/1882-gri...witch-brackets

[charliex]

Testing out 3D machine with a roughing pass for a hemisphere, 1/4" end mill HSS, 5 IPM. 2100 RPM.

It's loosing some steps or something moving so we're tramming and tightening up everything, once the flat surfacing was good, moved on to a side/3d operation.

so after spending a while with it now, the flashcutcnc is ok, the hardware driver box/steppers seem good so far, the windows software is below average, i'd wait for at least version 5 before taking the plunge. Have never been able to get them to respond about the sdk/remote dll that is mentioned on the sales/manual either asked about 4 times so far, they just ignore the question, so either have to reverse engineer the interface driver or wait it out for better software.

the gui is slow and very basic, you can't move around once the machine is started running, even though part of the claim is that box does all the work, so the PC should be free'd up to run the GUI (the zoom resets after every GCODE reparse and it does that often, so you have to keep zooming back in, and if you forget before starting you have to feed hold and then do it, small annoyance.

can't clear the toolpaths so it just ends up being a big single colour blob that is mostly pointless to watch (depending on bit size i guess) rendering is not the quickest either, be nice to have some more modes or zooms/clear/transparency.
it reparses the GCODE everytime you make a change to program positions etc, which takes a while especially on bigger GCODE, and if you've got any warnings to ignore it'll pop up each time ,

too much lag between keyboard and movement,

in continuous movement jog, sometimes it doesn't move on a click, sometimes it moves a bigger amount, sometimes they stack together, so three presses, delays and then fourth press it'll do them all at once

can't change the GCODE feed override to more than 9,999% which seems a lot til you simulate a long run that is at 5IPM.

no safe Z height, if you jump to a line in GCODE, the Z won't retract to a safe working height, it'll just plough through the part and snap the bit, or stall out. you have to change the Z manually, then pick the new GCODE jump point and it'll then move X/Y then Z

jog controls are tedious if you're using the mouse, changing step size and continuous gets old fast ( going to make an external jog controller) i'd personally rather have two buttons one continuous , one slow.

setting the DRO's/program positions is also tedious. drop down, select axis, clear, confirm, etc. again a couple more buttons would be great, especially since the primary viewports don't seem that useful beyond a very basic inspection of the toolpaths,

still would like a real estop, even just for the psychological effects.


the software does get it done, but its like going back to windows 3.11 after using windows 7.

[charliex]

one videos per post, mostly these are just because we wanted to see how the gopro did filming

[mcordeira]

Hi

You mentioned you set the machine type... in the software the G0704 is not an option... I reached out to Flashcut and they sent the G0704 Pro-Series.stp that they use internally. What did you select?

I cannot get the motor to respond for some reason... the LED on the controller react, but no motor movement... any advice?

Best,
Marc

[charliex]

i config'd it myself using the pro box , https://www.dropbox.com/s/bjfnv3jxjd...0Mill.stp?dl=0 i have limits setup so might want to turn those off

be curious to see theirs.

[mcordeira]

Thanks for sending it over... I will give it a shot tonight.

You can find the one they provided me here: https://app.box.com/s/35tw51ohq7b0gr4p558106g6f4dqlxup

Best,
Marc

[mcordeira]

Btw I have the limits as well... any good advice on homing the machine to a zero? I am making 4" x 2.5" aluminum belt buckles - that's all... I could not find any real good resources, videos, etc. on the topic

[charliex]

i added limit/home switches, and just use that. i move the y away from the column a little bit in the limit switch setting you can see in my config., otherwise i just let it do its thing.

if you're running the stock lead screw setup, try cutting a circle of aluminium about an 1" across, i'd be curious to see how it comes out. or measure backlash.

some notes
beware of using the jump to line in the GCODE, it doesn't safe Z, so it'll plough through parts happily.
when you re zero a tool after a tool change the visual tool paths don't reflect that change, the paths are still correct they're just visually off in the editor, annoying when you're trying to watch where it is.
i've been using FANUC gcode style, works good so far.
any key for feed hold is the old 1980s joke meaning any key but the one you panic hit, use the space bar.

those are some of minor things i've picked up using it so far. feel free to email me if i don't see them on here. i can't imagine there are many of us with this setup.

[charliex]

started getting around to installing the ballscrews into the g0704, using the old flashcut aluminium extrusions for a temp pulley based mount instead of direct.







had problems using the backlash correction in flashcut to have enough precision, i thought there was an option to allow it to be more digits but can't find it.

strange thing the Automation Technology double nut ballscrews i bought aren't preloaded, the spacer between was just loose, not sure if that is the normal with them or not, also a noticeable difference betwen the X and Y second nut in movement. its moving about 140in/m now versus the 40-50 i had on the lead screws.

[charliex]

finally got time to clean up the gibs, install the ballscrews tram the head etc and do so test cuts. backlash is 0.0013" on the X and 0.0019" on the Y which is still pretty awful, so going to find some bellevilles that aren't a 1000lbs.



its way more rigid now, and the sounds of the cuts are are lot nicer. finish has improved vastly ( middle hole was an aborted run) 20IPM 3200RPM, can move the XY at 150 IPM now.
Attachment 294218

[charliex]

backlash is down to X0.0002" Y0.0006" added belleville washers between the double nuts and tightened the belts. actually managed to cut a circle that a bearing dropped into with spot on for the X and low on the Y of about 0.0017716545"

[charliex]

this was after putting the bearing in and out a few times, so few slight scratches from that. 20 IPM 0.3" DOC with a 0.250" end mill

[Fastest1]

this was after putting the bearing in and out a few times, so few slight scratches from that. 20 IPM 0.3" DOC with a 0.250" end mill

In the 1st picture at the 3 oclock position, is this from backlash or the photo? I dont see a mark similar at the 12 oclock position. In my experience that would indicate the X is very low BL or well compensated for. And a little Y BL is causing the mark at the 3 oclock position, most likely there is a similar mark at 9 oclock I cant see from the pics.

[charliex]

that is where the tool exited, the lead out was too abrupt= so it left a line.