Honest Review: SeeMeCNC Eris Delta Printer

[626Pilot]

Background
================================
I decided to get into 3D printing in late 2012. After reviewing many printers, I went with the original Rostock MAX, mostly because of its huge build envelope and how cool it looks. I've been a customer of SeeMeCNC ever since. I decided to order an Eris so that I could have an extra printer to do little jobs, and to have a backup in case one of my other printers should fail.


First Impressions
================================
The box containing the printer was shipped to my place in about a week, for less than 50 bucks. Not bad, considering that SeeMeCNC is over a thousand miles away! The box has a nice carry handle on top. Opening it, one finds custom-molded foam packing materials, the printer itself, a power brick, a USB "B" cable, a SeeMeCNC palette knife for part removal, and a 1-pound spool of blue PLA. Aside from the build plate reading "CAUTION: Please read user manual located at seemecnc.com before use", there is no documentation at all.

The machine is fairly impressive for a $550 3D printer. Most current offerings in this range (like, say, the DeltaPrintr) are just a bunch of unpainted laser-cut plywood that has been bolted together, with exposed wiring and other things that make them look like they were put together in the garage. Conversely, the Eris is all injection-molded plastic. Nothing looks off-the-shelf about it. The printer was obviously built by people with considerable talent and resources.


Frame and Drivetrain
================================
Up to this point, SeeMeCNC's printers have all used linear rails. Cheapskate bearings ride on these rails with acetal wheels. The Eris is a complete departure from that. It uses pairs of smooth rods at each tower, for a total of six. The carriages are very simple compared to the older Cheapskates, and they are moved by standard GT2 belts. The arms are SeeMeCNC's usual injection-molded plastic, done in a truss for rigidity. The universal joints are SeeMeCNC's newer ball-cup design, rather than the older style that was retired in 2015. The frame seems very stiff to me, which is a good sign. Not all $550 3D printers are so rigid.


Hot End and Build Surface
================================
This is one of the printer's most notable features, and like the smooth rods and simplified carriages, represents a departure from the old. SeeMeCNC's Rostock and Orion hot ends have traditionally been hybrids of metal and PEEK, and use two heating resistors. The new hot end is all metal, uses a single heating resistor, and has a really tiny heater block. They are still using copper-silicone RTV gasket maker to interface the heating resistor and thermistor to the heater block. One thing to note: The heating resistor's leads are exposed. This means that if you use the part removal tool or anything else metal to get plastic off the nozzle, you risk shorting the resistor. I did this, and the printer shut off for a second before coming back on. I'm lucky it didn't blow a fuse, or worse. The earliest Eris printer hot ends had shorter heating resistor leads. Mine came with longer leads, apparently because the shorter ones occasionally heat up the PCB so much that solder could drip from it. It would be nice if SeeMeCNC put fiberglass sleeves on the resistor leads. It seems to me that that could make the machine a little bit safer, and less prone to the sort of accident that requires replacing electronic components.

The PCB is a pretty big deal, and adds a lot to the integration of the printer. It hosts a pin header for power and signal, an accelerometer for automatic calibration, and several blinky LEDs that tell you when the power and heat are on. The hot end is cooled by a single blower fan that sucks air across the cooling fins and into a chamber made of injection-molded plastic. The top of this chamber can be rotated around the axis of the hot end, and depending on how it's rotated, waste air from cooling the hot end is either blown up and away, or directed down into three air ducts for part cooling. An LED is hidden in one of these ducts, and is used for part illumination. All in all, a very clever system, although I wish they'd used a more powerful blower fan - or used more than one of them. SeeMeCNC's 10mm blower fans are excellent, and it would be interesting if there was a mod to add one or two of those to the fan duct system. They vary speed with PWM far better than some cheaper 12V blower fans I tested, and as you may already know, PWM speed control is critical for making sure the filament isn't over-cooled. With just the one blower fan, I got mediocre overhang performance at 45 degrees of overhang, and it got worse the steeper the overhang was. I would say the LED hidden in one fan duct volute could be blocking some of the air, but there just isn't enough air coming out to matter much.

The build surface is an unheated borosilicate disc, about 6 inches across, held down by three tabs that are easy to move out of the way. The top surface of the disc is laminated in something called FABLAM, which is said to work fine with PLA and PETG, and (with a small amount of bed heat) ABS. The surface is VERY tough, and doesn't mar easily. It's said that bed heat will be a future option, but if all you care about is PLA/PETG, it won't matter. It grips PLA like a champ, and removal is pretty easy, especially with the included part removal tool. The Eris' store page states that the build envelope is "4.9" (124mm) in diameter and 6-1/2" (165mm) tall".

I noticed a little bit of stringing at 205C. The hot end seems like it can run cooler, like 200 or even 195. Those temperatures might improve things.


Base and Controller
================================
The printer's base contains four short-can stepper motors. According to Jetguy, it might be a good idea to replace these with motors that are better suited for the drivers used by the controller, but I haven't tried that and can't comment on their efficacy. I would certainly be interested in hearing from anyone who has. The printer is controlled by the familiar RAMBo, although this is a "mini" version. All fuses are standard automotive blade type, so you can get replacements at any auto shop. (The regular RAMBo relies on two Littelfuses, which have to be ordered online.) The RAMBo is not screwed down - it sits in channels that were cast directly into the plastic. These hold it pretty firmly. It would've been nice to see a 32-bit controller, like a Smoothieboard or Duet, but SeeMeCNC really likes the RAMBo's design features, and they've never released anything with a 32-bit controller to date, so I'm not surprised. Still, an Azteeg X5 or similar would've been a good choice, I think. It would've cost about the same.

There is a cooling fan at the bottom of the base that turns on automatically when the motors are energized. It's a bit loud - I think a blower fan would have been much quieter. The noise could get annoying after a few hours. Also, it will tend to suck up any stray filament scraps and make an irritating noise as the blades hit the filament, so keep all that stuff away from the base. I had to disassemble the bottom to tweeze the filament out from between the fan blades. A fine mesh grill would solve this issue. Finally, there's a small LED panel that lights up the Eris logo when the printer is turned on. Nice touch.


Setup
================================
I plugged the printer into its power brick, connected it to a computer with the supplied USB cable, and fired up Repetier Host. I had to run configureFirst.sh to get the required 250000 bits/sec port speed to run under Linux. After doing that, I had no trouble connecting. This printer has an auto calibration system, and the script that makes that work can be found here. Calibration doesn't work perfectly (see below), but after massaging some settings, I got it to work "Good Enough" for a test print.


Auto Calibration
================================
Fun to watch, but imperfect. It taps at all three towers to calibrate the endstops, then homes and taps at the center and Z tower to do the delta radius, then homes again and taps the center to get the Z height. Results are automatically stored in EEPROM when it's done. In the southernmost third of my print surface, the nozzle gets so low that it blocks, which is no good. It does help to run the calibration with the printer on a completely level surface, but even so, it still does the same thing.

I suspect that they could improve on this by adding some code that would tap halfway out to the towers in a triangle pattern, and using the results to do plane compensation. Judging by the spew in the terminal when the auto calibration runs, Repetier Firmware does have plane compensation ability, so most of the required code is probably already there. I think doing the endstops, then delta radius, then plane compensation - run one after another, several times, so the results converge better - would be a better strategy. Z height would be set after the last run through.

It could also be that the axis steppers should be replaced, per Jetguy's suggestion, with ones that are a better match for the RAMBo's drivers.

I notice that movement is sometimes jerky during auto calibration, particularly when accelerating on the XY plane to get from one probing point to another. All three axes stutter at the same time, which suggests to me that the controller is running out of steam. Using a Smoothie- or Duet-based controller would solve that nicely.

I tried adjusting the probing feedrates in EEPROM. Lowering them to 30mm/sec caused the head to crash during probing! I think this is either from the CPU running out of overhead, and needing the probe to hit during an open time slice (which happens to be massaged just right by running at a faster speed), or from faster speeds resulting in less resonance that could be throwing off the accelerometer.

I did have to adjust the probe offset from -0.2 to -0.3 to get the first layer to print successfully, if not perfectly. The filament towards the north end of the build surface tends to be overly round, and towards the south end, it's overly squishy. This results in undesirable artifacts. I also had to use a bed roughness of 0.1 in KISSlicer to fatten up the first layer.

It's important that there isn't any excess plastic on the nozzle tip during probing. To address this, I ran some extra filament through the hot end to get a "handle", then shut off the heat. Once the temp got down to around 150C, I pulled on the "handle" and all the excess PLA sitting inside the melt chamber came out, leaving the nozzle tip totally clean.


Comparison to the DeltaPrintr
================================
I bought a DeltaPrintr in their Kickstarter campaign, which was way back in 2013. It was delivered in early 2015, and I have posted my review here. I got mine for $500, which is $50 cheaper than the Orion. I think their similar cost makes it fair to compare them.

I have to say that the difference in quality between these two printers is huge. The DeltaPrintr has two advantages over the Eris: a larger build envelope, and a menu controller with SD card slot, so it can run stand-alone. However, these are the only two advantages it has. The Eris has a better frame (both in terms of how polished it is, and how much stiffer it seems), better arms, better hot end, better effector platform, better probing solution, and the ULTRALAM print surface does away with the annoying painter's tape. Eris' auto-calibration is not perfect, but it still works better for me than the FSR-based solution used on the DeltaPrintr. Also, you don't have to put the Eris together, which will save the average builder a few days.


My Recommendation
================================
If you want a small-footprint 3D printer that weighs less than ten pounds, I'd take a serious look at this one. The auto calibration system isn't perfect, but by massaging some settings, it can be made to work. If you need perfect auto calibration right out of the box, you might want to wait until the dust settles on the current implementation.

If you want a larger build envelope, I would advise staying away from the DeltaPrintr, and getting something like a SeeMeCNC Orion (also factory built), Rostock MAX (available either in kit form or factory built), or building your own printer on the Trick Laser MAX METAL frame. Or, you can be a square, and get a square, Cartesian printer.

[Tincho85]

Great review, I always enjoy reading your posts.

[626Pilot]

I tried shimming to improve the 1st layer adhesion. It helped a little, but not much. It seems that my printer has a lot of trouble printing close to the edge.

[Mac The Knife]

I got tired of shimming the bed, reverted to printing out a test piece, and adjusting the endstop offsets in eeprom.

[joe]

I got tired of shimming the bed, reverted to printing out a test piece, and adjusting the endstop offsets in eeprom.

So after reading this I am led to believe the accelerator probe does not work consistently.
The one they just released for the Max?
Mac, 626, will you be buying the new hotend/probe assy for your Rostock Max?

[Mac The Knife]

IMO they need to revisit the auto-cal for the ERIS. After using their auto-cal routine, I have to shim up the bed at the Z-tower location. It may work better for the Max since it has a larger area.

As for the new hotend, I'm in now hurry as I've modified my stock hotend to be able to print at higher temperatures. When I ever finish my Metal Max build, I'll consider it.

[626Pilot]

One possibility is that the printers work perfectly when assembled, but then something gets shaken loose during shipment. My Eris came with what looked like a perfect test print, but I could never get the same results.

I've ordered the steppers recommended by Jetguy. I'll try them, and see if it helps. Could be a different probe speed would help as well.

I won't be using the HE280 hot end on any printer I build because I'm already using an E3D Cyclops. If SeeMe releases just the PCB part, I'll think about it, although the groove mount setup would complicate things in my case.

IMO, a better solution would be to use a tiny AVR (Arduino) controller, one of the tiny chips with only a few pins, use that to read a standard accelerometer, and then tie a pin high or low, thereby generating an endstop signal that could be read by ANY controller running ANY firmware. The AVR and accelerometer would both be on the PCB. No I2C involved, and therefore less chance of electromagnetic interference or wire format overhead introducing delays. This is similar to how the FSR endstop board works.

[JATMN]

Great review.

Yea I have had so many issues with the auto level that I have mostly given up on it and just use software leveling in mattercontrol.

Completely destroyed the print surface sheet that came on the glass.. about 80% of the time when leveling at least one point some reason was under the
print bed so the nozzle just rammed the bed when printing in that region. Leveling with and without shims.. bare glass.. print sheet.. painters tape.. all had same results for me..