Welcome to the SeeMeCNC Community

mhackney wrote:So to change the subject just a bit...

I've been running FSRs on all 7 of my delta printers. I only have 1 Rostock Max though, the others are mini Kossels and Kossel 250s. I get EXCELLENT calibration results on those smaller printers and acceptable results on the Max.

Has anyone else observed this?

That's what I've been chasing, too. That first point visibly deflects the hotend, and it's been frustrating me. It's also somewhat intermittent. (my posts/rant were here: http://forum.seemecnc.com/viewtopic.php ... 974#p91442 )

I was chasing rigidity, and just picked up a set of Brian's aluminum arm mounts (which look awesome, by the way!) because the stock plastic arm mounts have never really set flat on the trick trucks, and I think that they can flex a little bit. I'll install those tonight, check my belts, and re-test. I've checked for loose belts, but I haven't made an effort to match them precisely. I'll let you know if it helps!

Another note, you can add another jumper to the JohnSL board to make it more sensitive. With my 6# aluminum bed, I found that it helps decrease the pressure needed to trigger the board, which on at least a placebo level has helped

Mine has been intermittent too Jim. But I think Mac The Knife is on to something with the belt tension.

I thought the sensitivity adjustments on the JohnSL go like this:

both sets of jumpers open - most sensitive
both sets of jumpers installed/closed - less sensitive

?

I've been meaning to look for a guitar tuning app, and download it to my phone to adjust all three belts to the same tension. Never got around to it yet.

mhackney wrote:I thought the sensitivity adjustments on the JohnSL go like this:

both sets of jumpers open - most sensitive
both sets of jumpers installed/closed - less sensitive

You're correct, I wasn't being clear or specific... Only one jumper is is for the sensitivity, the other one makes the sensor normally closed. Great point.
https://github.com/JohnSL/FSR_Endstop#s ... ensitivity

Actually there are 2 jumper positions for sensitivity and 1 position for NC vs NO on the V1.2 board. SN1 and SN2 jumpers give you 4 possible sensitivity settings. Both open is the highest sensitivity.

IMBoring25 wrote:It's out, for pre-order...

https://www.duet3d.com/DuetWifi

I didn't know. Thanks.

mhackney wrote:SN1 and SN2 jumpers give you 4 possible sensitivity settings. Both open is the highest sensitivity.

What am I missing?
[img]http://i.imgur.com/pVLeYDX.png?2[/img]

When it comes to matching belt sensitivity, this is what I do....

Rig up a "heavy" weight of some sort that won't ding/bang up your towers. I'm using something around the 2-3 lb range.

Home your machine, and set the z-height about mid point. Leave power applied to the motors.
Put a dial indicator on the carriage
Apply the load and watch the deflection on the indicator
Adjust belt tension until all towers deflect the same amount with the same load

I'm currently set at about 0.015" of deflection with the 3 lb load. It's an arbitrary value, but the belts don't feel too tight, but it takes a fair amount of external force to move the carriage. When the belts are too loose, you won't get a repeatable deflection after repeatedly loading and unloading. I'm getting about 0.0015" repeatability with the .015" deflection. I noticed improved FSR repeatability (the correction values) when using a Smoothie board once tension was equal. I'm assuming the Rambo would benefit also .

On a Max Metal frame, it's been too difficult for me to judge belt tension without using this process because the belts are all but hidden within the frame.

That's a great technique ccavanaugh!

g.

Thanks geneb.

I don't think most realize it, but equal belt tension impacts dimensional accuracy of the printer as well. Belt stretch ever so slightly impacts the resultant steps/mm. On large x-y prints where you are using a large amount of the tower travel, the steps/mm error accumulates and dimensional and angular errors can be measured.

The Duet calibration auto-calibration will alter steps/mm if you allow it. It's a neat exercise to run it before and allow steps/mm adjustment and then compare results after making belt tension equal for all towers.

I'm also see some improvement in reduction of moiré patterns after making sure belt tension was equal.

ccavanaugh wrote:Rig up a "heavy" weight of some sort that won't ding/bang up your towers. I'm using something around the 2-3 lb range.

Home your machine, and set the z-height about mid point. Leave power applied to the motors.
Put a dial indicator on the carriage
Apply the load and watch the deflection on the indicator
Adjust belt tension until all towers deflect the same amount with the same load

I don't visualize where the weight goes... On the carriage, too? Then are you measuring the belt deflection with the dial indicator or the movement of the carriage?

You are missing nothing, I'm an idiot! I had made a copy of the JohnSL jumper settings in a log book I keep and I didn't copy it correctly. I had the default - both open - jumpers setting as .95 sensitivity. My bad, sorry for the confusion!

Jimustanguitar wrote:

mhackney wrote:SN1 and SN2 jumpers give you 4 possible sensitivity settings. Both open is the highest sensitivity.

What am I missing?
[img]http://i.imgur.com/pVLeYDX.png?2[/img]

Jimustanguitar wrote:

ccavanaugh wrote:Rig up a "heavy" weight of some sort that won't ding/bang up your towers. I'm using something around the 2-3 lb range.

Home your machine, and set the z-height about mid point. Leave power applied to the motors.
Put a dial indicator on the carriage
Apply the load and watch the deflection on the indicator
Adjust belt tension until all towers deflect the same amount with the same load

I don't visualize where the weight goes... On the carriage, too? Then are you measuring the belt deflection with the dial indicator or the movement of the carriage?

Yes, hang the weight off the carriage and measure the carriage movement.

Jimustanguitar wrote:

ccavanaugh wrote:Rig up a "heavy" weight of some sort that won't ding/bang up your towers. I'm using something around the 2-3 lb range.

Home your machine, and set the z-height about mid point. Leave power applied to the motors.
Put a dial indicator on the carriage
Apply the load and watch the deflection on the indicator
Adjust belt tension until all towers deflect the same amount with the same load

I don't visualize where the weight goes... On the carriage, too? Then are you measuring the belt deflection with the dial indicator or the movement of the carriage?

Yes, hang the weight off the carriage and measure the carriage movement.

mhackney wrote:I've been running FSRs on all 7 of my delta printers. I only have 1 Rostock Max though, the others are mini Kossels and Kossel 250s. I get EXCELLENT calibration results on those smaller printers and acceptable results on the Max. But in the pursuit of perfection it has been driving me crazy as to why the Max doesn't quite calibrate as well (DC42 on Duet returns a deviation value that can be used to determine the fit). Earlier this week I slowed waaaay down and watched and discovered that as the nozzle touches the first point on the outer perimeter of the bed, that it actually slides outwards (laterally) a fraction of a mm.

Has anyone else observed this?

I was suspicious of the stock plastic arm mounts not being rigid enough, so I swapped those out for Brian's aluminum ones last night.
[img]http://i.imgur.com/vcIyJb7.jpg?1[/img]
Man, they're gorgeous.

I can see that the deflection from before is gone. My effector would slide outwards just like you were describing, and while I haven't taken any quantitative measurements yet, it qualitatively looks much, much better.

Of course, swapping the arm mounts requires retensioning the belts, but I gave them a pluck before starting with the disassembly, and I feel pretty good that they were similarly tight.

I'll post numbers of the H values I was using in my firmware before and after the aluminum mount swap soon.