I was surprised to see how advanced some users are on 3D printing.
I’ve been recently reading with much interest somebody that is putting into practice what I’ve been willing to do for months.
Extruder pressure advance management done right (with proof)… Amazing!
Yes I like Deckingman. He really thinks about what he is doing and how it ends up working.
Sailfish has had really good pressure advance since… like… 2011-2012? What’s amazing to me is how long it took the rest of the 3D printing world to catch up on a feature that makes such a big difference to corner quality at higher print speeds.
The Duet/RepRapFirmware pressure advance and Marlin Lin Advance features are both based on Sailfish JKN Advance. (The actual code implementations are quite different because advance functionality is very intimately connected to the way stepper interrupts and axis synchronization/interpolation are handled.)
MachineKit’s Velocity Extrusion was developed independently but ends up doing pretty much the same thing. In my opinion they’re infringing a Stratasys patent (https://www.google.com/patents/US6054077) but nobody seems to care.
@Ryan_Carlyle … and some player have still to play catch up (Smoothieware for example). There’s still an open feature request since 2012 I think. I just have to learn how to code… Damn!
I’m using pressure advance on marlin, it’s good, but finicky. It helps a bit, but too easy to overshoot/undershoot.
There’s a Smoothie fork with advance code but to my understanding it relies on segmentation for non-linear kinematics like deltas to work… which sounds like a kludge but I don’t know the details. The way they originally implemented extruder motion interpolation by directly slaving it to the XYZ motion makes it very difficult to implement advance functionality. Makes good sense for lasercutters but limited the options for 3D printers. Other firmwares more often run the extruder as a complete fourth CNC axis with its own accel,jerk,etc control. What Sailfish did that was unique was putting the extruder on a separate stepper interrupt so it was running three synchronized XYZ motion axes and a separate extruder motion axis with its own speed control. This makes processor load management way more important (and Sailfish is hyper-optimized for 8bit processors) which is why Marlin’s Lin Advance incurs a pretty big performance hit for high-speed printing. Duet/RepRapFirmware doesn’t have to worry about 8bit power limitations and has a different motion interpolation scheme where there’s no real penalty to tweaking the extruder motion for pressure Advance.
Thanks for the insight… Your knowledge of the ecosystem is up to none.
Nice link. Interesting progress has been made in Firmware for extruder pressure management solutions. Been talking about pressure sensors in the melt chamber for many years. Anyone been able to do this inexpensively? In the mean time there also seems to be a correlation between jerk and acceleration ratio settings that accomplishes similar results without the performance hit at high speeds.
@Jeff_Parish High jerk and low acceleration accomplishes the same corner blobbing quality outcome, but by minimizing flow rate change at corners. But then you get ringing, and it doesn’t really help with travel moves or layer change zippers.
People have measured filament pushing force during live printing, but not melt pool pressure. It’s a really, really difficult pressure measurement application – you’re talking about thousands of psi (yes, really) at ~450F. That’s like refinery-grade specs. The only sensible approach is to use a grease-tube like they use for injection-molding extruders. Basically you drill a port into the melt pool and fill it with high-temp silicone grease, so you can plumb the pressure away from the heat. That’s doable for hobbyists but still kind of complicated and doesn’t measure transients well.
Sounds like something fun to experiment. I wonder if you can do it without drilling all the way to the melt zone, use brass as a sort of diaphragm.
@Ryan_Carlyle Yep, I’ve seen those issues. I’ve been playing with relatively low jerk at 6 - 8 and moderate acceleration at 800 - 1000 and getting really good results in the 40 - 60 mm/s print speed range. Recently tried a print at 300% speed with Skip-Sense installed on a machine and it worked with no skips but it looks like I need to bump up the jerk and/or acceleration accordingly as there was a slight bump on the corners going that fast. This is where the advance pressure management comes in very useful as it can compensate for speed changes automatically.
Solid Drylin bearings have helped greatly with the ringing ( I suspect what I can still see is belt spring) and a really efficient part cooling fan blowing right at the nozzle tip helps nicely with short span ooze control. Agreed that jerk and acceleration settings alone do not help with travel moves as I will still get that little ooze blob from time to time on long travels.
I’m well aware of the issues of measuring local pressure at several thousand psi at 450F. I do it at work on industrial valves rated at 3K psi and operating at over 900 F. It is not cheap or easy. Putting an affordable high temp fast response pressure sensor in a printers melt chamber has not been successful, that I know of… yet. 
Could be an application for nano-sensor (I was thinking of Nanolike for example). I know they’re into tiny deformation measurement for aeronautics.
@Michael_Memeteau It’s probably feasible to use some sort of strain gauge on a heatbreak tube… not sure what you’d use exactly, but it sounds possible. Should be possible to correlate hoop strain to internal pressure. Dunno how much resolution/accuracy you’d get though.
@Stephanie_A All you strictly need is a mixing hot end like the Cyclops or Diamond, then use one of the filament ports for a short grease tube to a pressure sensor. I’ve seriously considered doing this. The problem is that I think the mixing port geometry will change the extrusion pressure enough to make it not a particularly useful measurement. Won’t know until we try it though None of my current hot ends are amenable to drilling in a “sense” port because they all seal the nozzle against the heat break. Something with a one-piece nozzle+hotblock construction (ulgh) would be the way to go here.
@Jeff_Parish Right now I’m working on 20,000 psi @ 250F valves that have to be able to cut 2" OD x 0.245" wall x 135 ksi YS steel pipe and still seal glad I don’t have to deal with 900F though…
One idea is to put a small floating (up and down) retaining ring with pressure sensing piezo devices between the nozzle and the PTFE tube/heat break. Could certainly withstand the temperatures. Wire leads from them are an issue.
@Ryan_Carlyle Sounds like fun! We typically don’t do anything over API 10K range. 250 F is like a nice crisp fall day.
I love the idea of the “floating” bowden style feeder that this printer has on it.
Hi Guys. I noticed on my blog and YouTube channel that this has aroused a bit of interest and that there were a lot of referrals from google+. So I thought I’d stop by and say hello. I’ve never used google+ before so bear with me 'cos this is all new and I’m an old guy.
Ref the “floating” style extruders, they were on their own gantry but passively dragged around by the main hot end. The latest evolution of my printer is 5 colour so now has 5 off E3D Titans which, with motors and mounts, is quite a lot of extra mass to drag around. So I’ve added a second, driven coreXY gantry, stacked above the main one, that holds the extruders and follows the lower gantry around. Essentially, two coreXYs stacked one above the other. All this 'cos I didn’t want 1 metre long Bowden tubes… Anyway, here is a link to another little video that I’ve just put together if anyone is interested https://www.youtube.com/watch?v=0lW0EgNX0iQ
Ian