Repeat V2?

Anyone have some constructive criticism about the repeat?

My findings so far.
-Love the Z belts. Need to use a different attachment system though, too hard to change or tension.
-Actually changing my mind on extrusions, as they seem more dimensionally stable and far easier to adjust than my cut boxes. I calibrate my printers to pretty low numbers. So either built in adjusters or use extrusions.
-The CF beam works great, but I think a Linear guide will make for an overall smaller build and the weight is not all that big of a difference at moderate speeds. Unless I ditch the direct drive extruder, then the CF stays. Tough choice.
-Direct drive extruders are so big…I have not tried a bowden with linear advance. Might need to experiment.
-True bed leveling is awesome and worth the extra hardware.
-Not sure about the BL touch, pretty large, but the accuracy is amazing. I know there are other smaller options.

-Absolutely hate the wire management on the repeat, as well as the filament path. I think the CF rail is too sensitive for whipping around a large bundle, but I will not know until I try a linear guide. I think I will do a quick conversion to see if there is a difference.

That is my 6 month review, and getting itchy to try a revamp.

1 Like

I just bought some 20mm aluminum square tube. A 500mm piece of 20X20X1mm CF weighs 71 grams (I have 2, and I got 68g and 74g for the 2 pieces, so I averaged it). A 500mm piece of 20X20X1mm aluminum tube weighs 88g. (Again, I have 2 pieces, and they were the same at 88g.) The aluminum is more consistent, more dimensionally accurate, and will (I strongly suspect) get rid of the faint pattern that I can see along the X axis of my prints.

The CF looks cool though.

A linear rail for the X axis is certainly easier.

The core belt tension blocks are a little dodgy, I think. I’d have more confidence in something closer to the system that the LR3 X belt uses. Much easier to install and you can make both gross and fine adjustments so much faster. It would be significantly easier to square the printer that way, I think.

The Z belts are good, though I sometimes can see that the teeth of the belt touch the blocks as they go past, and I can’t help but wonder if that’s making some of the Z movement inconsistent. A bit more clearance there would be a good thing. Maybe single screw tension adjustment for the Z belts, too. The current setup with 2 bolts is difficult to remove the last bit of twist to keep the belt flat (Possibly adding to the problem above.)

It bothers me a little that the motor block blocks the hotend fan when the Y axis homes… I fixed mine in firmware, so that when the X axis homes, it moves back out 20mm.

True bed levelling rocks. Can’t say enough good things about this.

I haven’t built one using extrusions, but I have some pretty good calibration numbers from my cut boxes, I think. I should measure the rail parallels again, after a few months and very different ambient humidity winter to summer…

I have never been happy with the output from a Bowden setup, I tried it on the MP3DPv2 and wasn’t happy with it at all, even with the Duet mainboard and dialing in the linear advance. In addition, that much linear advance seems to stretch out the print times. It went from being really close to slicer estimated times to significantly longer at the same movement speed and acceleration settings. I gave up at that point, and installed a direct drive system, so I never did debug it.

I hate wire management. I have no idea how to make it better, since no matter what I do, it seems to be awful.

I would like to see some adjustment to the endstops. M666 doesn’t seem to work for the adjustments to the Z axis, and the belt tooth resolution is still off more than I like, so the G34 bed levelling has to make larger adjustments than I think is ideal. The belt as endstop flags isn’t a bad idea, but it would be nice to have something with finer adjustment.

1 Like

Awesome, thank you!

Ryan, have you tried making a 300x300mm version?

The IR sensors seem pretty interesting to me. My BL touch is still very consistent, but it has scraped on some failed prints, so the pin gets bent and then the leveling needs to be calibrated again.

I also like the idea of those CAN based boards to reduce the size of the umbilical to the hotend. If I were tl try something on my printer, that would be close to the top of the list.

My gridbot has a linear rail for the X. It works great, but I don’t fly at top speeds.

1 Like

Not yet but I am considering it! The LR3 printed parts are a ton of large flat parts. With bed leveling I can finally use the whole bed fearlessly so now I want larger beds.

yup, my main BL touch issue is the extra “servo” wires. Over complicated. TT did a good comparison video, they were all basically more than good enough from what I remeber.

Drool!!! I allllmost bought one a while back.

1 Like

The 300mm thing is pretty common now. CR10 parts are available for cheap, and there is a healthy aftermarket for alternatives that sized.

A lot of the repeat builds have been that size too, which is a good reason for you to try it before you go on a change rampage :slight_smile:.

It makes sense for your farm to be built from 200mm printers. That size is great for everyone. But you should have one bigger one for fun, and personal projects, and occasional odd jobs (like testing a big LR3 part).

I honestly don’t use the whole bed. But I am much more confident printing 200mm things on this printer. Is that irrational?

No that makes sense. I feel like the very edges are almost off limits for a couple reasons.

3 of my 8 printers still have pcb/glass beds. I will probably replace those with 300^2 beds.

The repeats are rad, it took me a while to deal with the CNC’d ones relaxing or moving or whatever happened but now, dam, they are fast and very reliable. I want more, and better. I have some ideas to make them a bit smaller and integrate some wire chains or something. Still up in the air about the extruder setup. On top of that I am way more comfortable with Fusion360 now.

2 Likes

My 300x300 Repeat is functional though not quite complete. I’ll update my build log relatively soon :wink:

I love it. It has been an awesome project and learning experience on multiple fronts.

I could talk about how great it is all day, but if you are really looking for constructive feedback I will say that it is a 9+ from a “design for manufacturing”, but the current design only a 6 or so from a “design for assembly” perspective. Not bad (5 being average), but I would say the “repeat” philosophy has driven a few assembly sequence order constraints and tool access restrictions that made putting it together and getting belts tensioned a bit tedious. I am a little worried about maintenance too, but optimistic that it will be bullet proof.

I would encourage a “standard” electronics package with a bit more headroom for expansion options (High temperature module, multiple extruders, etc.), but overall it is an elegant design with only modest room for mechanical improvement. If there could be a low-cost commodity type CAN wiring solution to the extruder, that would be a huge plus.

I do also highly recommend a 300x300x300 build. It does look like a monster next to my MP3DP-V2, but really only takes up about 10% more bench space and has 3x the build volume.

2 Likes

I use these, I’ve got 2 of them.

  1. PEI is transparent to IR, so you’ve got to be careful with using PEI as a build surface. It’s OK if the PEI is thin, or if you’re using a flexible magnetic bed.

  2. It seems to be somewhat sensitive to colour. Dark colours sometimes seem to register a bit lower than lighter colours. This led to some interesting results until I caught on. My current setup has a PEI sheet directly adhered to an aluminum build platform, with the aluminum spray painted gloss black. This seems to work reasonably well.

  3. Stringing filament is not your friend. Optical interference below the LEDs makes for some interesting sensor errors.

  4. Sunlight in the printer’s home area is not your friend either. While the IR sensor tends to reject interference from stray IR sources, it’s not perfect, and a stray sunbeam can mean something unexpected in the height map. I also sometimes suspect that inconsistencies in the bed temperature have led to some errors as well.

In general, I think that my BLTouch sensors are more accurate. The IR sensor has some advantages. I like that there is no physical contact. The repeatability in a particular spot is excellent, too. I’d like to see what it’s like on the spring steel magnetic beds. I suspect that it would be much better than the surface that I am currently using it on.

Regarding a 300X300mm version: I highly recommend this. I am absolutely delighted in being able to use such a large print area, and I can use the whole thing! The one downside is finding a heated build platform that size with a 3 post mounting surface, the CR10 build platforms use a 4 post mount.

2 Likes

Random notes as a slow starting Repeat builder-

I’m still working through learning Fusion 360 and was thinking of building using 2020 extrusion and targeting a 300x300 bed setup. I don’t mind the learning at all- but it is mildly frustrating that there isn’t a reference build in a common size like 300x300 that could be followed without going through that learning curve.

This plus working on calibrating my TAZ has effectively stopped my build after printing the PLA parts.

The belted Z is one of the great things about the repeat, but I was dreading the attachment and tensioning.

For direct drive with less mass, the H2 as you use is good, as is the Orbiter extruder. I’m finishing a BMG with a pancake stepper and E3D V6 on my TAZ, and it’s 1/4 the mass of the original extruder. Amazing.

For a larger volume repeat, a setup like the BMG X2 might be worth having if someone wanted to use dual extrusion. I’ve been wanting to use dissolvable filaments for a while and some option for a dual extruder on the repeat would be good.

I would love an option for a quick change hotend like the E3D Revo or similar,.

I’ve decided for my use case a 300x300 heated bed will be the size. I was just looking at the various common heatbeds in this size, and an option for a ‘stock’ build using a commodity heated build plate would be good for new builders. I hate that so many of these use 4-point mounting. DUMB! (Giving my TAZ a dirty look as I think about that- it’s the same way.) Easy to put 3 points into a plane. 4 points just leads to infinite ways to warp and deform the print bed.

I realized at the larger build volumes that .4 and even .6mm nozzle sizes don’t make sense for me- I’m wanting to go with a .8 or even 1.2mm nozzle on a high-flow hotend and will run .5mm or larger layer heights for many of my prints. Something to perhaps think about when designing parts.

I have access to machines with smaller nozzles if I should want to print something with fine detail, but for production use like what I imagine your own print farm does, those higher material flow rates seem pretty tempting.

I just fell down the rabbit hole of calculating flow rates as I’ve been trying to get my TAZ better calibrated. As I’ve been working on that it’s been amazing how much faster (in theory) larger parts can be printed with the higher flow rates on larger nozzles. I know that CoreXY allows faster motion, but for production use of parts like those you design, optimizing for a larger nozzle looks like it would make your print farm operations faster and more efficient. It wouldn’t hurt strength either to have thicker parameter walls.

I was wanting to do something better on my Repeat build for wire management. Everything else about V1 machines are so wonderfully well thought out, but wire management seems to be the most difficult thing for builders like me to get right. I build spacecraft, so my expectations for wire management are absurd- and I seldom meet my own expectations on my own personal builds.

As a note, Klipper firmware just recently added support for CAN, including cheap RPi Pico boards with CAN transceivers- so I’m really thinking I’ll want to get my build done with Klipper (Will be my first Klipper machine). Eventually I think we’ll see more use of CAN and something like USB-C for power delivery and communication to toolheads as reducing wiring count and connection types seems like a good idea.

I think a revised repeat should just use linear rails on all the axis and forego the CF tubing on X. Might even open up the possibility of using a compatible extruder mount to other coreXY machines that are using linear rails on X, which would allow more extruder combinations.

Finally, one of the really great things about V1 machines is that you provide source for the designs and support your builder community. Thank you for that. It’s really nice that you reach out for reviews, when so many today are completely deaf to user feedback.

2 Likes

Sounds like we have a volunteer.

That would be awesome. Dui was trying to use USB C for his hotend umbilical, until he found out most cables don’t have most of the pins populated. But using CAN would fix that. CAN uses ±2.5V. I wonder if there is any issue with that. You would certainly fry anything that was actually USBC if you plugged it in.

The firmwares will all support CAN at some point. I thought Marlin did too. But I haven’t looked closely. I use klipper and it does develop fast (by design).

I do like using larger nozzles but the quality suffers with thicker layers. I am sure that is why there is a quest for faster printers not things like 5mm nozzles. I use 0.5mm nozzles and 0.35-0.4mm layer height, I really really do not want to go above that. Since my repeats are so much faster than my V2’s I have been using finer layers to get some of that nice crisp detail back.

Wire management is one of those things I kind of enjoy, so I used to only think about it on a high level when designing. I knew I could make it work nicely as long as the parts fit. So many people complained, I realized I needed to add a bit more…I just do not like adding a bunch of printed parts and hardware to constrain a tiny wire when a cable tie works great. I am learning to incorporate it more though, the LR3 is closer. I knew the wire path and added features, still didn’t add any extra parts though.

As for the 300x300, Two reasons. For me I have been at this since 2013 every bed I have ever touched has been 200x200. literally thousands and thousands. I never think of anything else. The second reason is when I do consider it I know I will go mains power on it. That adds cost along with the larger bed, heater, spring steel, now SSR. This one did get designed at 200 and 300, but I left it at the size I was using.

All good info so far, the next one will be different. I have a bit to wrap up on the LR3 instructions before I get started in CAD so I wanted to start getting info before I run with it.

If I were to switch extruders I would surely pick on that has a door/hatch. The only issue I ever get with the hemera’s is during filament swaps, and occasionally it leaves a piece in there and I have to take the fan and heat sink off just to get it out. The BMG style hatch we used to have on the wade’s extruder is sorely missed.

So many new direct drives.

Well shoot, I’m stuck until august waiting on rails anyway…

1 Like

I have a 24V one and it comes up to temp pretty quickly. It still takes a while to be completely even temp, but the heater shuts off in about 3 mins. The aluminum is what I would have to change to make the heating more even.

Mains might be nice for ABS and 100C beds, but I don’t think it adds enough to justify the potential risk of AC on a moving part of my printer.

Just my $0.02.

4 Likes

I did go with a mains silicone heater for 300x300. I was able to salvage a high quality SSR so the cost was about the same as a 24V heater and it heats & controls incredibly well. Once upon a time I did mange to short out a 12V bed plate heater and that left a pretty good mark on things, so I would never put a mains heater on a bed slinger. However the “vertical only” motion of the bed on the Repeat doesn’t scare me too much - there is nice strain relief and any moving wires are nowhere near any metal. Nonetheless I’ve got a drag chain mostly planned out for the vertical travel to provide even more belly comfort.

I will say though that the cost of a decent 300x300 aluminum build plate and spring steal build surface are surprisingly high. I currently have a 1/8" thick sheet of aluminum mounted that is covered in painters tape that cost $10, but I got what I paid for (0.7mm flatness). The 1/4" cast plate I ordered isn’t in yet and I am auditioning PEI coated build surfaces in 200x200mm form factor before I buy a big one of them. Together and with the heater they are going to cost almost as much as my first printer did all-in. Other than those components though, scaling from 200 to 300mm on the Repeat was only a few dollars for longer rails - and bit more wood.

I haven’t ever used an AC heater. So I guess I was speaking outside of my experience.

My point is/should be that you can enjoy a 300mm bed heated by 24V. I remember mime being reasonably priced (it was designed for a cr10 and I had to deal with the holes).

I can’t speak to the idea that mains is or isn’t worth the risk, since I haven’t tried it.

3 Likes

I’m currently working on building a 300mm^3 repeat after having built a Voron 2.4 and there are some features in the Voron that I really like and am trying to incorporate into the repeat:

  1. Mainsail (instead of Octopi) with Klipper based firmware – there’s a learning curve, but the documentation is great, the user base is growing, and I don’t have to go to my main PC to recompile when making changes to the configuration
  2. Bed mount-based z-min end stop
  3. Klicky (or unklicky) – printed Z end-stop on a sled used for leveling the bed
  4. Auto-Z calibration – combines the last two to
    auto-calibrate any nozzle offset

Current state is I’ve got the frame built and the motion system installed and tested. Working on the extruder and it’s wiring now.

I really like the 2020 extrusions for the frame – you can make slight shifts to things without drilling another hole (for example, for wire management or LEDs or mounting control boards and such).

I do like the repeat design, very printable and the only feedback on the assembly so far matches what Ryan said – the belt tensioning was a bit fiddly – particularly for the z belts trying to get the screw to engage while making sure the belts were out of the way, but not too out of the way that you couldn’t adjust enough. Just my 2 cents on that one since once they’re done you only have to check/adjust them occasionally if at all.

I’m definitely looking forward to printing with it!

1 Like

I agree with you. Maybe a single tensioning screw, with adjustment from the top.

I’m using linear a rail, been printing at 150-200 mm/s with no problem.

My build has a 110V heated bed, so far so good, it heats up to temperature in less than 2 min, has 0.1mm flatness, the downside was the price of the bed

I like the cut box, like @SupraGuy, I got pretty good calibration numbers from it.
With Kippler I’m not using the Z 3 optical sensors, and still getting true bed leveling.

Once I have some time, I will try another build with extrusions, maybe will be a V2

2 Likes

I am fairly certain the Z endstops are not needed in marlin. I added them for future power failure resume function. Turns out I will either invest in larger battery backup or just forget about recovery. If the power goes out long enough the parts pop off the bed so no need trying to resume.

Been looking at cost saving, it is not easy. Those linear guides are expensive.