alden

You correctly point out the challenges. If you start with the goal that you want to motor motion to adhere as closely as possible to the physics – which dictate that jerk (and even further derivatives) be taken into account, it creates other challenges as you suggest. Laying down a uniform bead being just one of them.

It’s interesting to note that the early reprap machines (including the makerbot) did not have acceleration management at all. They just turned the motors on and off, so the velocity was always constant. This made extrusion computation very easy, but the motors had to run very slowly as they were unable to accelerate to their full speed. They could only move as fast as the initial step would “catch”.

So you are right, you have to change your rate of extrusion.

Guidance on setting the jerk can be found here:
https://github.com/synthetos/TinyG/wiki/TinyG-Tuning

Basically you just crank it up as much as you can until it starts failing and back off. You’ll get a feel for it once you play with it a bit.

This really looks like a serial buffer problem. What flow control are you using? With Coolterm I recommend using XON/XOFF. Make sure both the board and Coolterm are set up for XON/XOFF.

Can you send me your status report setup line and parameters? Or are you using the default status report?

Are you doing anything to cause the SYS report? – like asking for {“sys”:””}, or is this just showing up “on its own”?

I’ll try to reproduce once I have the data.

The green LEDs are just an indicator of stepper driver activity. Electrically they are just the voltage on phase A1 with respect to ground. This means that if the driver chips are activated or changing this phase may change. Since grbl asserts ENABLE to all chips they tend to all go on when grbl is activated.

Mostly the LEDs are there to provide an indication that the motors are running and working. It’s useful for diagnostics to see if grbl is talking to the board, and to see if the drivers are working even if the motors are not (mis-wired, disconnected, blown, etc.)

We expect to push an edge branch by the end of the year. After about a month of soak time it will go to master. Thanks for your patience.

Thanks for pointing this out. It does indeed look like it should work differently. We will add it to the known bugs for fixing.

It doesn’t work like that. All machine tools need to manage acceleration and deceleration. The tool cannot go from zero to the feed rate (F) instantaneously. It must go through an acceleration at the start and a deceleration at the end. Constant jerk acceleration management just does that faster than constant acceleration. You spend less time in accel/decel this way and therefore more time at the selected feed rate.

If you are spending too much time in accel / decel your jerk setting is probably too low. You would have the same problem in a constant acceleration system if your acceleration was set too low.

The other thing to realize is that jerk is the impact imparted to the machine. It happens in start and stop, and around corners. This can cause all kinds of issues like oscillation (bad surface finish), chatter, and even misalignment. Trying to control jerk by controlling acceleration means giving up a lot of ultimate speed relative to controlling jerk directly, as TinyG does.

• This reply was modified 10 years, 11 months ago by alden.

Videos are here:

…there appears to be something really wrong with trying to embed URLs in the page. So try this. Put the pieces together. Sorry for the inconvenience.

build 380.08
http://youtu.be/ _ObnEoOqpmg

build 406.03
http://youtu.be/ yut6X3hdJ_0