Posts: 359
Threads: 14
Likes Received: 163 in 126 posts
Likes Given: 1
Joined: Jul 2020
Reputation:
12 I just did an upgrade of my ender 3's with SKR 1.4 turbo boards and raspberry pi 4.
Both where setup identically at the same time with identical firmware just loaded into both at once. Basic calibrations like Pid Tuning hot end and esteps (the same anyway) where done for each printer separately.
When I print an XYZ cube from the same g code both finish at very different times and with very different quality, visually you can see one is moving faster than the other. Octoprint gives identical print times for each before I begin. Mechanically there is nothing wrong with either machine that I can see. Belts are tight, linear rails and wheels roll smoothly, steppers arent missing steps or getting hot. tried switching filament between the two with the same machine always having the worse print no matter what I do.
Im using:
Cura 4.9.1 with octoprint plugin
octoprint 1.6.1
Marlin 2.0.x bugfix
looked through everything on the eeprom from the lcd screen
ran M503 and both identical
Any ideas how this is even possible? I know I can tweak the faster machine to slow the print and increase quality but why is it not the same speed anyway? Im missing something Im sure of it.
• Posts: 4,550
Threads: 38
Likes Received: 1,245 in 1,064 posts
Likes Given: 430
Joined: Aug 2020
Reputation:
118 In the M503 out put the things too look for will be Max Feed rates and Max Acceleration and if these are the same then that is a surprise.
How about:
- Hot ends the same? Do they reach a similar temperature roughly in the same amount of time?
- Extruders that same? Same gearing?
- Are the power supplies the same? because this can affect getting the hotend upto temperature and maintaining it there for enough filament to melt and extrude.
- Are they running exactly the same Marlin version? 2.0.x has so many revisions and I would not be surprised that there are differences between releases.
• Posts: 359
Threads: 14
Likes Received: 163 in 126 posts
Likes Given: 1
Joined: Jul 2020
Reputation:
12 25-May-2021, 03:41 PM (This post was last modified: 25-May-2021, 03:42 PM by EVILsteve.) Same everything I'd purposely built both identical so I can swap g codes between them easier.
microswiss hotends on v1 Enders. New extruder motors for microswiss direct drive adapter plates both identical. Both using tmc2209 steppers on uart, current is the same through marlin. Vref on the pins is not used in uart mode.
Marlin was compiled and uploaded to both boards one after another from the same build.
I've narrowed it down to being something in the firmware on the skr boards, board themselves or a phyical issue with the printers. I switched raspberry pi's so it loaded printer 1 profiles to printer 2 and vice versa. It's still the same printer that runs faster than the other.
• Posts: 21,315
Threads: 589
Likes Received: 8,997 in 6,660 posts
Likes Given: 1,426
Joined: Jun 2018
Reputation:
790 If you suspect a difference in the SKR boards or the firmware on them, swap the boards over between the printers and see if the slow one then prints faster (and vice versa). Then you will at least confirm or eliminate the SKR boards as being the cause of the issue.
• Posts: 359
Threads: 14
Likes Received: 163 in 126 posts
Likes Given: 1
Joined: Jul 2020
Reputation:
12 I expect it's got to be something in the firmware. I'd set one up first and dialed in the linear advance to see if it was worth upgrading both machines. Then I'd purchased the second board and re-flashed both so they'd be the same. The best I can guess is the first board did not re-flash and is running the previous build. It's very possible I'd setup the second differently. I should be able to pull the sd cards and see the .bin file to confirm it hadn't uploaded.
I'm not even sure what machine is operating correctly either. Cura shows 29 minute print time and is generally quite close with my previous setup with marlin 1.1.9, octoprint says 23 minutes but is notoriously wrong. It takes 40 minutes on the slow machine and prints quite well and 30 on the other and prints poorly. This is a square corner linear advance torture test so the additional time while seemingly excessive might make sense. Perhaps one does not have linear advance enabled. But I can see the K value with M503 command in both.
I think I will swap boards tonight and load a fresh build of Marlin with a stable version so I can see that it's no longer 2.0.x bugfix to confirm it flashed then run another test.
• Posts: 359
Threads: 14
Likes Received: 163 in 126 posts
Likes Given: 1
Joined: Jul 2020
Reputation:
12 Solved!!
I bought two SKR 1.4 TURBO boards but one of them ended up not being a turbo.
The difference is a 120hz processor in the turbo and 100hz for non turbo other than that the boards and pinouts are identical.
With the turbo firmware loaded it was operating much slower than it should be.
•