Why isn't horizontal radius calibrated this way?

[bob64]

Step 1) Make all corner axis's the same height by calibrating with the end stops.
Step 2) Calibrate z-height at center. so z=0 is calibrated at center with the paper between the glass and nozzle.
Step 3) Check height at one of the axis. If it's too high and you need to lower the nozzle, decrease horizontal radius. If it's pinching the paper and you need to raise the nozzle, increase the horizontal radius.
Step 4) Recheck that same axis until it's the same height as the center. Increasing/decreasing horizontal radius as needed. Re-run the same axis script after each radius change. Repeat this step until the pressure is exactly the same as the center.
Step 5) Make the other axis's match.
Step 6) recheck center. Repeat from step 2 if needed.

Reasoning: Horizontal radius only changes the height as it moves away from the center.... so there's no point in making the center hit 0 by adjusting all 3 endstops. Am I thinking this through or is there something i'm overlooking?

[Xenocrates]

I'm fairly certain that the reason the endstops are tweaked at all is to compensate for the arms, and for the geometry, and that horizontal radius is tuned somewhat the way you describe. (Although you tune for the smallest offset, and then make the other endstop asjustments, rather than picking a random axis) It wouldn't change your effective H-radius to make the printer think an arm is slightly higher/lower, that would only affect the centering of the printer, and not the concavity of the geometry (It does allow you to even things out before you adjust H-radius, as if you just tweak one end stop to match, it should throw the others off)

[nebbian]

Bob64, I do things exactly the way you describe, it's the fastest way to get the delta radius that I've found. I can normally get it within 0.1mm in 5 minutes or so.

[Jimustanguitar]

That's how I do it too. Make it level first, set your center height perfectly, and then make everything else match last.

[bob64]

Uh, why doesn't the assembly guide do it this way?

[mhackney]

History, which repeats itself.

[Jimustanguitar]

Uh, why doesn't the assembly guide do it this way?

The manual probably has you set the center point first so that you don't crash the machine too badly if you don't know what you're doing yet.

[nebbian]

The most common way of doing this that I've found is:

1. * Calibrate endstops so that nozzle is at 0 when brushing a piece of paper near all three towers
   * See how far off the nozzle is when at the centre
   * Adjust delta radius
   * Go to step 1.

The problem with this is that you have to calibrate the end stops every time you adjust delta radius, which is inefficient.

The much quicker way is to get your end stops calibrated enough so that they are horizontal, then give them an offset so that the nozzle is at 0 when in the centre, and then iterate over the delta radius while testing the nozzle near the edge. It's basic troubleshooting -- isolate your variables.

[Jimustanguitar]

The problem with this is that you have to calibrate the end stops every time you adjust delta radius, which is inefficient.

That probably indicates some other problem then. Changing the radius will move the perimeter points equally. Level the machine first, and you shouldn't have to touch the adjustment screws again.

[geneb]

I'm really hoping the new hot end with the accelerometer board will make the whole manual calibration thing a moot point.

g.

[Windshadow]

I'm really hoping the new hot end with the accelerometer board will make the whole manual calibration thing a moot point.

g.

in my case I would use the word praying than hoping

as an insider Gene do you get the new hardware for things like this several weeks before release so you can write the manual revisions before it ships to the great unwashed multitude?

[mhackney]

Duet with FSRs make the whole manual calibration thing a moot point! It takes more than just the probing system, the calibration code also needs to perform fast and well. To date, RRF is the only firmware I've found that calibrates accurately in less than 30 seconds - including probe time. An accelerometer probe is icing on the cake.

[Xenocrates]

I think Gene gets it a week or so before they start going in kits, dependent on stock on hand (Sometimes longer, sometimes shorter), but that on occasions it's been released as a retrofit part very soon after Gene gets it. (And he sometimes just finds stuff on his doorstep with minimal warning), given his past griping about some of it.

Somehow though, I've been very lucky on the calibration front. It produces good parts, and I haven't needed to re-calibrate in months, despite many, many hardware changes (I didn't even have to re-tension the belts at the tops of the towers when I did the new carriages, I just pulled the belt to approximate tension and clamped it). Just reset the Z-height, and sometimes PID, and I'm good to go. Which is really amazing considering it's in an unheated, uncontrolled shop. Not that I know how I managed that at all, and I'm really hoping any future printers go half as smoothly. (I'm going with beginners luck and a fluke personally)

[Jimustanguitar]

I think Gene gets it a week or so before they start going in kits

If he's lucky

[bubbasnow]

I'm really hoping the new hot end with the accelerometer board will make the whole manual calibration thing a moot point.

g.

I enjoy these conversations resurfacing every couple months... auto calibration will make these forums boring.

[geneb]

The earliest I'll see one is at MRRF.

g.

[Windshadow]

Duet with FSRs make the whole manual calibration thing a moot point! It takes more than just the probing system, the calibration code also needs to perform fast and well. To date, RRF is the only firmware I've found that calibrates accurately in less than 30 seconds - including probe time. An accelerometer probe is icing on the cake.

My wish list includes the duet and the DC firmware as well as the 7" screen etc but as I read about the nest version of the duet due in a few months I thought I would struggle on so I could do a one stop rebuild this summer as a birthday gift to my self
Today I had one of my endstop screws work itself out while printing.... strangely that print finished ok but the next one to put it mildly did not.... I never heard the screw fall out so I never thought to look before I started the next print.

if frustration continues i may just go with your fsr fix and push up the whole project and not wait for the rumored new version of the duet (this is not my first hassle with the endstops)

[3D-Print]

Any idea what the rumored "New Duet" is in reality??

[cope413]

It's likely this...

Basically a RAMBo with an ARM chip...

[Jimustanguitar]

Looks like Johnny's design work. Same connectors as the other Rambo's, etc. Very nice. Ethernet and SD are interesting. I like the automotive fuse holders, too. Looks like different silicon for the stepper drivers...

Which board is that, actually?

*edit* Now I can see that it says Ultimachine on it... Are you testing the new Rambo?

[DeltaCon]

Today I had one of my endstop screws work itself out while printing.... strangely that print finished ok but the next one to put it mildly did not....

That makes completely sense, because endstop screws are only used while homing.... when your next print starts...

[cope413]

Yes, we got one for testing. I don't know what the name is. I've seen "Archim" - which is printed on the board - but it's unclear what the actual name will be.

[cope413]

Also, there isn't ethernet on this unit, and the drivers are DRV8825

[HumanLiberty]

Duet with FSRs make the whole manual calibration thing a moot point! It takes more than just the probing system, the calibration code also needs to perform fast and well. To date, RRF is the only firmware I've found that calibrates accurately in less than 30 seconds - including probe time. An accelerometer probe is icing on the cake.

Is there a place in this forum that describes how to implement what you've described here?
I've had FSRs installed for months, to try using them with OpenDACT, but firing them never affected the nozzle movement, which just kept crashing into the bed. I'd hate to have to buy and install a new board with my RAMBo still functional, but also hate the FSRs being useless - anything to have reliable printing without turning any GD screws or removing the nozzle to attach a probe....

BTW, I came here looking for info on any EEPROM/Firmware adjustments needed for having installed the 325mm Trick laser rods - all I did so far was change Diagonal Rod Length from 269 to 325. Anyone know?