Really unusual speed issues with new Atomstack A48 Pro V2

Hi guys,

I have a REALLY weird one here. I’ve recently set up a new laser (Atomstack A48 Pro V2). I noticed the laser was homing very slowly and scanning white space very slowly, so I checked my Device Settings to ensure that Fast Whitespace Scan was enabled and set to 7000 mm/min, which it was


I then noticed something even more odd.
If I change the speed of my layer to 10,000 mm/min, the whitespace speed was still slow, but the return to home was much much quicker. If I then run it again at the exact same speed, both the whitespace speed and the homing speed is nice and fast.
If I then change the layer to a slower speed, like 2000, on the first run, the whitespace speed is really quick, but homing is very slow. If I run it again, both the whitespace and the homing is really slow.

This really make no sense at all to me. Does anyone have any suggestions? I’ve also dumped the settings below

$0=1
$1=25
$2=0
$3=1
$4=0
$5=1
$6=0
$10=1
$11=0.010
$12=0.002
$13=0
$20=0
$21=1
$22=1
$23=3
$24=1500.000
$25=3000.000
$26=500.000
$27=2.000
$30=1000.000
$31=50.000
$32=1
$100=80.000
$101=80.000
$102=1600.000
$103=100.000
$104=100.000
$105=100.000
$110=50000.000
$111=20000.000
$112=1000.000
$113=1000.000
$114=1000.000
$115=1000.000
$120=4000.000
$121=1000.000
$122=500.000
$123=200.000
$124=200.000
$125=200.000
$130=395.000
$131=385.000
$132=40.000
$133=300.000
$134=300.000
$135=300.000

Are those the default Atomstack values?

The $110 50000 mm/min = 833 mm/s top speed is extraordinarily fast, particularly as the machine blurb claims only 500 mm/s. The $120 4000 mm/s² acceleration says it will hit 500 mm/s in maybe 30 mm, but that seems aggressive for such a large laser head riding what look like 2020 rails.

I suspect the acceleration & top speed exceed the machine’s ability to keep up, so the motors are stumbling along at a sub-harmonic of that speed. Knock the X axis top speed down to no more than 30000 mm/min and acceleration to 1000 mm/s², with similar reductions to the Y axis.

The homing speeds seem very high for the axis lengths. Set the $24 homing speed to 400 mm/min and the $25 homing seek speed to 100 mm/min; it’ll take at most a minute to home and you can adjust as needed.

The $0 step pulse length of 1 µs seems short, but I don’t know what the hardware looks like. If increasing it to maybe 10 µs doesn’t change anything, then crank it back down again.

Try dialing the pace back to see if the results become more rational. If not, Atomstack folks around here may have other suggestions.

Thanks for looking at this,
Those are the defaults. I even reset them to factory settings just to be sure. The thing is, as I mentioned, it can home and whitespace scan very fast, I can make it do it, but it will only do it if the layer it’s cutting, is also set to run at high speeds. I can run that layer as much as I like, and it will be running as fast as I would expect it to. As soon as I slow down the speeds for the layer though, it suddenly goes back in to the really slow speeds for whitespace scanning and homing

Well, that’s comforting, even if some values don’t (seem to) make sense. :grin:

This may be one of those G-Code interpreter issues, where undocumented parameter values take effect when unknown conditions occur.

All of which is out of my knowledge base!

I think you are confusing Home with jog to start position. You only Home the machine once when powered up. That rate is determined by the $25 parameter, which looks good on yours. I think your $26 parameter (Pull Off) should be about half of the stated value. $26=250 is more reasonable.

About that return to home position: If the head is away from the start position, it will jog at the programmed G00 rate. There is no default G00 rate, so I think Lightburn uses the Speed setting in the Move window. I do not speak with authority, just from my feeling of how my machines behave with Lightburn.

I do not understand this obsession with faster, faster, faster rapid moves. I estimate the difference between 1000mm/m and 20000mm/m on a 16" frame is about 1/2 second. So you would have to jog or fast-move 120 times to lose one minute. We waste more time screwing with parts placement than that. Running the machine at sensible speeds will improve accuracy and reduce errors. Soapbox End