Manual - How to install TL-Smooth - Zebra Strippes Fix UMO - UM2 - UM3?

[diettelo 5]

i‘ve got the 4 duodes model installed and it good but also habe the 8 diodes modell at home but not installed.

 

so what are better 4 or 8?

[neotko 1,417]

3 minutes ago, diettelo said:

i‘ve got the 4 duodes model installed and it good but also habe the 8 diodes modell at home but not installed.

 

so what are better 4 or 8?

 

I trust @conny_g that they might be just as bad 

 

I have the 8 ones, but I will not test them since next week I get the 3rd duet3d board and some hiwin to finish the mod to G3D corexy and finally forget about the headaches of this um2 boards. Freedom there we go!

 

[diettelo 5]

I hope you would test it.

 

I´ve made a Case for the Tl Smoothers:

 

https://www.thingiverse.com/thing:2770345

[chuck-yanke 10]

Having followed this thread from the start I’m confused. Do these chips make prints better or not?

[mechamecha 64]

They absolutely made my prints better, as you can see from the photos I posted earlier in the thread. They virtually eliminated the visible stepping on the surfaces of my prints. conny_g already did the soldering fix to his printer, that's why adding these smoothers didn't work. I haven't noticed any under-extrusion like neotko mentioned, but if I ever do, I suppose I'll just remove the smoother installed on the extruder. I did add some small heat sinks to each, because I experienced one mishap with a very long printing session where I'm assuming the smoother boards got too hot. I'm still very pleased with this simple, cheap mod.

[gr5 2,170]

@chuck-yanke

Here is a quick summary of the problem:

Ultimaker and many (but not all) printer's stepper drivers are implemented such that every 15th and 16th step is very weak and the stepper usually doesn't move on those.  So if it is moving smoothly it moves steps 1 through 14 then pauses for two steps and then suddenly moves 3 steps on the next cycle (back to the first step).  This causes very very tiny stair-steps in your print and is most noticable on large flattish surfaces that are not parallel (but are almost parallel) to x,y,z planes.

 

For example this is invisible on most cylinders, cubes, boxes, gears, chamfers, holes...   It's invisible on almost all parts.  Unless you rotate your cube by 1 degree from being parallel with X axis.  It's visible on parts like "benchy" which is cartoonish and none of the planes are parallel to anything.  So on Benchy's walls which are slightly not vertical and slightly not parallel it's visible.

It can also be visible as repeating horizontal layer lines because sometimes the Z steps more and sometimes less depending.

For what I print - mostly mechanical stuff - it doesn't matter.  For artistic, organic stuff like an animal - which has no large flattish surfaces - it doesn't matter.  For certain parts it matters.

 

The tl-smoother adds a diode which makes it so the stepper has to up it's voltage.  This really helps the 15th and 16th steps get more power - more oomph.  The tradeoff is the stepper driver has to work harder and gets a bit hotter.  No one seems to know how much hotter so it's not clear if you need to add heat sinks/fans or not.  Most people do not and most people seem fine and happy.  If the stepper drivers overheat their built in circuit turns off for a few milliseconds and you get lost steps in your prints - it's very obvious.  But no hardware will get damaged - instead your print will be ruined.

 

In my opinion it's a waste to put the tl-smoother on the E axis (extruder) because it won't help the quality of your print (in my opnion) and also because that is the stepper that is working the hardest and most likely to overheat as well.  That is why most of these tl-smoother kits only have 3 cables for 3 steppers.

[conny_g 251]

Here are zoom photos of 2016 where I started to do laser exposure of PCBs using the mechanics of the UM2.
You can see the steps "paused" by the stepper.
I think I exposed with 1016dpi here which is half the technical resolution of the steppers, so you see the pause every 8th step.

Now - unfortunately I haven't made a photo of that - the effect is gone - I did the solder fix to connect the ROSC pin to ground.

 

Edited January 28, 2018 by conny_g

[cjs 34]

On 11.12.2017 at 3:18 PM, neotko said:

Duet3D or even a smoothie is the way to go.

Actually the smoothieboard has the drivers implemented the same way as Ultimaker and does suffer the same stripe effect. Did have to do some soldering on my smoothieboard to get rid of the problem. 

A new ultimaker board with the new silent trinamic drivers, like on the duet3d, would be awesome. Wonder how silent the ultimaker would be with such drivers 

[zerspaner_gerd 382]

On 28.1.2018 at 1:02 PM, cjs said:

Actually the smoothieboard has the drivers implemented the same way as Ultimaker and does suffer the same stripe effect.

 

 

The Duet3D board does not have this error, right?

[nilrog 103]

No, Duet3D uses TMC2660.

[zerspaner_gerd 382]

Now I'm confused, it's not true?
That means that Duet3D causes the same zebra stripes?

[cjs 34]

12 minutes ago, zerspaner_gerd said:

Now I'm confused, it's not true?
That means that Duet3D causes the same zebra stripes?

The Duet3D board doesn't have the issue!

One reason are the newer, more advanced, stepper drivers @nilrog mentioned. 

Wonder why ultimaker still doesn't  set the a4988 to low-current microstepping  

[diettelo 5]

8 diodes new or old 4 diodes design whats better?

[neotko 1,417]

18 hours ago, diettelo said:

8 diodes new or old 4 diodes design whats better?

 

Sorry no idea. This weekend finally changed my 3rd machine to duet3D. Can’t be happier. Interesting as an off topic, it handles PID much (MUCH) more precise than um2/marlin (or at least UM old marlin, since I never used new marlins)

 

 

I would save the money of this stuff and just move to a better board. Wifi, better pid (I still need to run PID for the beds to see if it can also improve the um2 pid default values). 

 

Ofc it isnt for an user that doesn’t have do any diy atm. 

 

So long and thanks for all the fish Um boards XD

 

[mutley3d 3]

Maybe this should be a new thread, but posting here as it is directly relevant to this thread.

 

I am Suffering the same issues on my 2 x UM2 machines. Want to finally do something about it. Done all the tweaks and mods, shorting ROSC pins, diode smoothers etc, none really "FIX" the issue. I have Duet boards available to upgrade like Neotko above, but kinda thinking I would like to retain as close to original as possible ie keep existing LCD. I also have a couple spare Arduino/RAMPs boards to hand.

 

So my thinking is use a RAMPS board with TMC2100. I have these drivers on my UMO and it prints wonderful. I have other RAMPS based machines with TMC2100 and they too print same great level of quality. I have a E3D thermistor cartridge, to replace the hotend thermistor, and will also replace the bed thermocouple with a thermistor and run new wires along the loom. This should take care of the main sensor differences. But has anyone fitted a RAMPs board and retained the UltraLCD? or to put the question another way, Has anyone compiled the UM2 firmware to work for a RAMPs board using the UM2 LCD?

 

Ive tried compiling the UM2 Marlin, with the board #define set to RAMPs EFB but as was expected, compile errors occur, for motor current variables, which of course are set via LCD, but not available on a RAMPs board.

 

If someone could help me around the compile errors, or point me in the direction where something like this with TMC drivers might have been done before (I'd be surprised if it hasn't), or suggest another approach in case im looking at this from the wrong perspective, I would be eternally grateful. Thanks in advance.

Edited March 18, 2018 by mutley3d

[neotko 1,417]

6 hours ago, mutley3d said:

Maybe this should be a new thread, but posting here as it is directly relevant to this thread.

 

I am Suffering the same issues on my 2 x UM2 machines. Want to finally do something about it. Done all the tweaks and mods, shorting ROSC pins, diode smoothers etc, none really "FIX" the issue. I have Duet boards available to upgrade like Neotko above, but kinda thinking I would like to retain as close to original as possible ie keep existing LCD. I also have a couple spare Arduino/RAMPs boards to hand.

 

So my thinking is use a RAMPS board with TMC2100. I have these drivers on my UMO and it prints wonderful. I have other RAMPS based machines with TMC2100 and they too print same great level of quality. I have a E3D thermistor cartridge, to replace the hotend thermistor, and will also replace the bed thermocouple with a thermistor and run new wires along the loom. This should take care of the main sensor differences. But has anyone fitted a RAMPs board and retained the UltraLCD? or to put the question another way, Has anyone compiled the UM2 firmware to work for a RAMPs board using the UM2 LCD?

 

Ive tried compiling the UM2 Marlin, with the board #define set to RAMPs EFB but as was expected, compile errors occur, for motor current variables, which of course are set via LCD, but not available on a RAMPs board.

 

If someone could help me around the compile errors, or point me in the direction where something like this with TMC drivers might have been done before (I'd be surprised if it hasn't), or suggest another approach in case im looking at this from the wrong perspective, I would be eternally grateful. Thanks in advance.

 

There’s an user called 3Dper on twitter that a few months ago did send me this picture (no manual or how-to) but basically what he did was rewire the board and connect tcm drivers outside cooling them with a waterbased something. He did pm this picture 

 

I don’t have more info, but... is doable

 

If I would be on your situation I would try the so called 2.1.5 um2 board on aliexpress that some guys sell that has the stepper chips removable and I don’t remember who but someone already told me it works (can’t find the twitter where they did)

 

IMO duet3d wifi, send stuff remotely and better PID tunning is just awesome. But all depends on how much work you want to put on soft or hardware. 

[halfie 8]

1 hour ago, neotko said:

 

There’s an user called 3Dper on twitter that a few months ago did send me this picture (no manual or how-to) but basically what he did was rewire the board and connect tcm drivers outside cooling them with a waterbased something. He did pm this picture 

 

I don’t have more info, but... is doable

 

If I would be on your situation I would try the so called 2.1.5 um2 board on aliexpress that some guys sell that has the stepper chips removable and I don’t remember who but someone already told me it works (can’t find the twitter where they did)

 

 

I'm using the aliexpress 2.1.5 board at the moment, and it's working just fine. 

 

What I did was change the MS1, MS2 and MS3 microswitches to OFF/OFF/OFF, to make the board full step, and then inserted the 2100 drivers. No modifications to drivers, like resistors or whatnot, just plug and play. The printer was completely silent. I got some TMC2100's sent along with the board, but I couldn't get them to satisfactorily print, it was constantly skipping steps. Could be voltage, or could be something else to do with the switch positioning? I don't know.

 

Also, the UM2 LCD isn't supported in the latest 1.1.8 release of Marlin, so I had to jump on the bugfix version, which made the screen work, but I could never get good, sharp corners. Just a ton of bleed (Linear advance on or off). There were a host of other issues too, like I couldn't figure out how to get auto bed levelling working with my piezo, which drove me nuts since it seems like such a cool thing. 

Wish I had more time to experiment, maybe I could have fixed the issues, but eventually i just switched back to A4988 drivers at 1/16 stepping on tinkergnome and the 8 diode smoothers. I'm sad. 

[mutley3d 3]

1 hour ago, neotko said:

 

There’s an user called 3Dper on twitter that a few months ago did send me this picture (no manual or how-to) but basically what he did was rewire the board and connect tcm drivers outside cooling them with a waterbased something. He did pm this picture 

 

I don’t have more info, but... is doable

 

If I would be on your situation I would try the so called 2.1.5 um2 board on aliexpress that some guys sell that has the stepper chips removable and I don’t remember who but someone already told me it works (can’t find the twitter where they did)

 

IMO duet3d wifi, send stuff remotely and better PID tunning is just awesome. But all depends on how much work you want to put on soft or hardware. 

OK so I wont be doing that however I think you just altered my train of thought and pulled me out of a rabbit hole.

 

I will update more shortly but I think I have a solution. Thank you for your response.

[mutley3d 3]

5 hours ago, halfie said:

 

I'm using the aliexpress 2.1.5 board at the moment, and it's working just fine. 

 

What I did was change the MS1, MS2 and MS3 microswitches to OFF/OFF/OFF, to make the board full step, and then inserted the 2100 drivers. No modifications to drivers, like resistors or whatnot, just plug and play. The printer was completely silent. I got some TMC2100's sent along with the board, but I couldn't get them to satisfactorily print, it was constantly skipping steps. Could be voltage, or could be something else to do with the switch positioning? I don't know.

 

Also, the UM2 LCD isn't supported in the latest 1.1.8 release of Marlin, so I had to jump on the bugfix version, which made the screen work, but I could never get good, sharp corners. Just a ton of bleed (Linear advance on or off). There were a host of other issues too, like I couldn't figure out how to get auto bed levelling working with my piezo, which drove me nuts since it seems like such a cool thing. 

Wish I had more time to experiment, maybe I could have fixed the issues, but eventually i just switched back to A4988 drivers at 1/16 stepping on tinkergnome and the 8 diode smoothers. I'm sad. 

 

I am looking into that board and have contacted the seller to request TMC drivers with it, along with some spares.

 

I suspect you may well have had the voltages set incorrectly on the drivers, and cooling them helps as they can get hot. Also if you had them in stealthchop (silent) mode, you get less torque, so possibly cumulative unfavourables working against you. Putting them into spreadcycle mode with a bridging wire increases available torque, but also noise, although my experience tells me you can get them pretty quiet with some tuning of the vref, especially when compared to original 4988 drivers. I will also be sticking with the original UM2 firmware, but might explore tinker fw aswell.

 

So Neotko's post threw my train of thought toward remapping pins with off-board drivers that I have laying about, and drop the TMC drivers onto those.

 

Now my stumbling block at this moment is some uncertainty on remapping pins for the X and Y motors. Can someone possibly advise if the pins on J25 (ADC1) and/or J24 (ADC2) can be remapped and used as STEP/DIR/ENABLE pins. I read that they are analog port pins but can also be used as digital I/O pins. If not, which pins can I use? I need six of them. If not, I notice a combination of 2 available pins on J23 + 3 available pins on J24 + 1 available pin on J26, if J24/J25 wont work.

[halfie 8]

@mutley3d I have them set to 0.9 each, with heatsinks, and with active cooling blowing across them from a 5050 blower, I'm definitely thinking it's stealthchop.

 

With regards to the drivers, the ones that you get from that AliExpress shop you don't actually need to wire bridge, just connect 2 solder pads on the driver, which makes it so much easier

 

Now if I can get any form of Auto bed leveling working I'll be happy

Edited March 19, 2018 by halfie

[mutley3d 3]

Some sort of progress. I have remapped pins in pins.h, and now have 2 external driver sockets wired in. I am using 6 pins from connector J25 (ADC0 ADC1 ADC2 for X axis) along with (ADC5 ADC6 ADC7 for Y axis) for STEP DIR and ENABLE signals respectively. Firmware has been flashed and I have managed to not let any magic smoke out, but i'm not quite there yet.

 

The Y axis motor seems to get enabled, but wont move, the X axis has no response. The 2560 data sheet appears to indicate these pins can be analog input, or signal output. I am not sure if or how I set these pins to become signal outputs. Anyone shed any light on these? I envisage/hope this will be an excellent and easy upgrade once it is resolved, transforming the level of print quality achieved.

 

Thinking aloud, I may remap one axis back to the onboard 4988 and the other to 3 known digital pins in the area of these headers to verify all circuitry and the off board drivers.

 

@halfie yea that's the problem with using the general Marlin release, its updated but may miss some of the UM functionality. I have one of those board in my alexpress cart, just waiting on the seller to get back to me about supplying with TMC drivers.

[neotko 1,417]

@mutley3d

 

afaik marlin2.0 has support for ulticontroller

 

https://github.com/MarlinFirmware/Marlin/pull/9501/files

[halfie 8]

17 minutes ago, neotko said:

@mutley3d

 

afaik marlin2.0 has support for ulticontroller

 

https://github.com/MarlinFirmware/Marlin/pull/9501/files

 

I'm actually wondering something... Would it be possible to use a different controller (ie the duet3d) and just use the ulticontroller plugged in via ext1 and 2 like they are for the ultiboard2?

[mutley3d 3]

OK so I think I have worked out what to do with those analog pins to define them as digital pins. Its so easy.

 

I now have motion on one axis testing with some spare 8825 drivers (instead of blowing tmc2100's). I have to recheck the other axis closely as it is lifeless, might be wiring, might be a dud carrier board.

 

@neotko yea but does it have the bed levelling routine? @halfie seems to be having probs with it.  Ill check it out once I have these drivers sorted, but taking one step at a time, too many changes in one go is dangerous. For now just using the Ultimaker/Marlin from their github repo

 

[tinkergnome 927]

5 minutes ago, mutley3d said:

OK so I think I have worked out what to do with those analog pins to define them as digital pins. Its so easy.

 

This should be already done at startup (if the correct pin numbers are defined):

https://github.com/Ultimaker/Ultimaker2Marlin/blob/master/Marlin/stepper.cpp#L698

 

pin numbers in your case should be 54 to 56 for x-axis and 59 to 61 for y-axis, right?

 

...but there are perhaps some more things to consider:

• https://github.com/Ultimaker/Ultimaker2Marlin/blob/master/Marlin/Configuration.h#L310
• https://github.com/Ultimaker/Ultimaker2Marlin/blob/master/Marlin/Configuration_adv.h#L165