search

mariolukas / FabScanPi-Server

FabScan Pi Open Source 3D Scanner - Server application
https://www.fabscan.org
GNU General Public License v2.0
184 stars 68 forks source link

hardcoded steps/revolution in FSTurntable.py #118

Closed doc2005 closed 5 years ago

doc2005 commented 6 years ago

in line 7 is the hardcoded value "STEPS_PER_ROTATION = 3200" I assume that this prevents us from using different settings than 3200steps/revolution. Settings in default.config.json were ignored for scan but not for calibration.

jwalt commented 6 years ago

The magic constants of 577 and 1022 in FSCalibration are based on 3200 steps per revolution, too. This is a severe bug, as the Watterott kit comes with a 400 steps/revolution stepper motor, i.e. 6400 steps with default microstepping. As a result, the Watterott kit is currently plain broken.

Fixing the above constant and using something like "65/5*self.steps_five_degree" in place of those constants leads to working scans, finally.

awatterott commented 6 years ago

The stepper motor from the FabScan Pi kit is a NEMA 17 with 200 steps/rev.

jwalt commented 6 years ago

@awatterott There is definitely something wrong in your shop. I got a 17HM3410, your website claims it has 200 steps/rev, but the data sheet you offer as download says it's .9 deg/step, which is 400 steps/rev.

Based on my experience in getting the kit to work (it still produces inferior results, but at least the shape is now roughly correct), the data sheet is right. With 3200 steps (200*16 microsteps), output is utter garbage, objects come out twisted and incomplete. With 6400 it is at least roughly the right shape.

awatterott commented 6 years ago

I will look into it. The 17HM3410 is listed in our ERP as 1.8° motor, but this is wrong.

jwalt commented 6 years ago

For the record, I found the reason for my remaining quality problems, they are not related to this bug or the kit itself, it's a calibration problem.

As for the issue at hand, see PR #120

awatterott commented 6 years ago

The 17HM3410 is included with the latest kits and we will update the software and documentation. Thank you for the PR.

mariolukas commented 6 years ago

I will replace the hardcoded values by config values in the new version. Thanks for reporting.

mariolukas commented 6 years ago

@jwalt can you send me some pictures of the garbaged scans. I want to add them to the troubleshoot section of the documentation.

jwalt commented 6 years ago

@mariolukas unfortunately (or luckily), calibration worked fine and I get decent scans. I don't have the distorted scans anymore, nor do I know what changed, another calibration just magically worked.

mariolukas commented 5 years ago

fixed in v.0.5.0