Various PETG brands compared

Hi all,
I experienced some difficulties with the original BambuLab PETG that I bought together with my X1C and see a lot of topics from people who also struggle to achieve nice prints. At some point, I suspected the filament itself and wanted to know if other brands behave better.

In one of those PETG problem threads another user and me started some tests, comparing BambuLab PETG, Sunlu PETG and IEMAI highspeed PETG, tweaking different parameters like nozzle temperature, print speed, cooling settings, speed reduction at overhangs and layer time. As criteria, I mainly looked at tendency to curl up, loose strands, and surface gloss. As I heard from several sources, PETG that becomes matte is being printed too cold or too fast and suffers from reduced layer adhesion and general strength.

I was surprised by the big differences that I found between BambuLab PETG and the other two, which behaved very similar. But just looking at the gloss didn’t seem adequate to rate the strength of the prints. I quickly came to the point that I wanted to build a pull testing machine, e.g. the one that Stefan from CNC Kitchen published as open source. Thankfully I cancelled that as completely unrealistic and reduced my demand to a simple force measurement.

So instead, I bought a force meter with peak hold function and 500N (~50kg) capacity. I built a little jig and designed a test object. To my surprise, even pieces with 6mm diameter withstood the maximum 500N pulling force easily. Making the diameter even smaller didn’t seem reasonable because it would introduce too much uncertainty. Instead I chose a differnt object, that would create an off axis load, a bit similar to Stefan’s test hook.

The whole project got a bit out of hand. So far I have printed and destroyed more than 150 specimen, testing for different brands and print settings. I thought my results might be interesting for others too, so I want to share them here.

This is the simple jig with the force meter I’m using:

and one of the specimen (please ignore the poor picture quality :wink: ) :

The cross section in the weakest spot is 5x8mm rounded, 4 walls, no infill. The mounting holes are offset by 11mm. This was the result from some iterative steps to optimise print time and material vs. overhang angles and a force that sits nicely in the range of the force meter.
For each combination of brand and settings, I printed 4 specimen to counter sample variation. I added an item that is printed at only 10mm/s to ensure sufficent layer time without using lots of material.

In the beginning, I only printed at 120 and 300 mm/s. Later I appended one test with 30 and 70 mm/s and all later tests were at 70, 120 and 300 mm/s. I did not want to go through all possible combinations but selected them based on prior results. E.g. when I saw almost no difference between 30, 70 and 120 mm/s, I dropped 30 mm/s from the later tests. I also did more extensive tests of fan settings only for the filaments I was most interested in. That is why the population of the table seems quite fragmented. All filaments had been dried before the test and then placed in the AMS.

I also tested some specimen printed flat. I tried to set walls and top/bottom layers so that cross section is as close as possible to the standing specimen, buth there might be differences. So please take those measurements with a grain of salt.

In addition to PETG, I also tested PC and three flavors of PLA.

And for completeness, this is what the current build plate looks like:

Ok, sorry for the very long introduction. This is the table with the condensed results. Only the best results for each brand are shown. The table below has all the information.

And here is the full table with all parameters, individual measurements and including standard deviation as a measure for sample variation:

My takeaway:

  • There really seems to be something wrong with the BambuLab PETG. I couldn’t even print the specimen successfully at 300mm/s in more than one attempt, mainly because the overhangs curled up and eventually were catched by the nozzle and kicked the whole piece over. That weak quality already became obvious during the tests for optimal temperature, speed and fan settings. I had zero interest in further investigating it, as the much cheaper contenders were so much easier, more forgiving and stronger.
  • The Sunlu is the cheapest and yet comes out on par or ahead of most others. So it is hot contender for my go to PETG.
  • The only noticeable step up over Sunlu is german brand dasfilament. Besides highest strength even at 300mm/s compared to others at lower speed, it also printed super nice, no matter which settings I threw at it and strength results were extremely consistent between specimen, where all others had much more variance.
  • almost all materials suffer from 300 mm/s while there is almost no difference between 70 and 120 mm/s. It would be nice to have more values e.g. at 180 and 240 mm/s. But I don’t see, starting over all tests. Maybe I will do it later. I will print uncritical pieces at default speeds, but reduce speed to 120 when strength is critical.
  • too much cooling really destroys layer adhesion for PETG. That is common knowledge, but it is nice to see it confirmed. Surprisingly, the material profile provided by dasfilament suggests 100% part fan and 70% aux fan, which leads to one of the worst layer adhesion values.
  • I totally underestimated, how much load those materials can bear! up to now. I have helplessly overdimensioned most parts that have to carry some load. The cross section of those specimen is really small and they are loaded very unfavorably and still hold up against 15 … 20kg perpendicular to the layers!
  • Color seems to have an influence, as can be seen from Sunlu clear vs. white. That too is old news, but it is interesting to experience yourself. I have heard, that especially white often deviates a lot in parameters from other colors. I also have a sealed roll of black Sunlu. I first want to use up all the other open rolls, but eventually will test that too.
  • PC really is a different beast, not so much for layer adhesion, but for strength within layers and the way it fails. Where all PETG samples just snapped at critical load, the PC started to yield, then walls separated from top & bottom and only when the walls were fully stretched, it finally gave up. On top there are all those other properties like heat resistance, impact resistance, stiffness, low creep. I will try to also get my hands on the Bambu PC, to see how it compares.

My choice of brands still is very limited, but I really don’t want to buy another bunch of rolls and unfortunately, only few sellers / manufacturers seem to offer samples. If you want me to test different settings with my available filaments or if you send me 50…100grams, I am looking forward to your input.

My next topic for investigation will be creep behaviour of different materials. In my opinion, this is a heavily underrated material property. Yet, there are almost no systematic tests available.


First off, Kudos for an excellent and thorough analysis and for sharing your finding.

I have broken my own promise to myself in keeping up to date a spreadsheet I’ve been trying to build on this very topic. My spreadsheet however, is based on qualitative metrics and price, I do not have the instrumentation nor time to evaluate strength.

While on the subject of PETG though. My experience with Bambu PETG is that is exhibits properties I’d more expect from PLA than true PETG. And of course, since I am a price-buyer, it is way overpriced for what you get. My observation have been that it produces a very “pretty” buttery smooth looking print. Considering that Bambu’s primary audience is the ease-of-use buyer, that makes sense.


You asked for suggestions to test. For me it is the following that have consistently hit the sweet spot of price and performance.

CC3D and HZST3D are clearly the same company or at the very least, they are made in the identical factory. The outer carton is the same except the print but the interior product is packaged using the same spool and winding methods. Test show that when I tune the filament after drying it first, the profiles yield identical performance.

HZST3D at $15/spool
CC3D at $14/spool

I would welcome another viewpoint on either one of these producers.


A newcomer on the block is Firos at $13/Spool. It performed identically after drying to the CC3d and HZST3D product.

Firos is an outlier with not much history but I will reorder from them if they hold their $13/spool price. So I don’t see it as a good candidate for testing efforts given that their longevity has yet to be proven. Plus, I have a strong bias against paper spools since the paper itself is prone to moisture retention which provide s challenge when drying and more importantly storage.

In addition, I don’t relish the idea of having to over dry a paper spool which presents a potential fire hazard while in storage. Back in my Boy Scout days we had a technical name for dehydrated cardboard: “Tinder”. :rofl:


There are others I’ve purchased, of course, but I won’t pollute the thread with all of the brands I’ve purchased unless there is interest. Since you tested Sunlu I will say this: Sunlu was one of the brands I found to be disappointing. It did not perform well, and at $16 during a lightning deal last October, I found its value to be truly subpar. Consider it’s base price at the time of $20 I bought it solely for evaluation purposes and have since crossed it off my list as a subpar brand.

My recent experience with Sunlu filament dryer and their shoddy tech support only reinforced that perception that they are yet another Chinese company looking to make a quick buck and are more about perception than true quality. As of today, for PETG they are at $22/spool. They are permanently off my list for all these reasons especially when I can get great PETG from between $12-$15/spool.

1 Like

Thank you for the kind words! I don’t feel like buying additional spools just for testing, maybe that wasn’t so clear :joy: There are two brands that I’m still interested in for PETG. Those are Polymaker and Prusa. Maybe I find somebody among my friends who can spare some.
Anyway I had a look into the brands you suggested. It seems like they don’t sell in germany unfortunately.

It is really interesting that many people made exactly opposing experiences on the same brands. I read that all over the place for both the cheapest and the most expensive brands.
Regarding Sunlu, 20$ seems a very high price. At Amazon germany it is 17€, which I consider overpriced as there is an ebay shop that sells 3 spools at 11…12€ per kg depending on color with 3 day shipping. Very odd that there are such big differences across the ocean.

I’m still undecided if I give in to the low price or if I want the better quality from a local brand at a still very fair price.

Ah, one more brand that sounds interesting but doesn’t sell in germany is VoxelPLA. They now also have PETG at a nice price and produce in the US but shipping overseas is absolutely prohibitive, so not for me.

I’ve had some limited experience with VoxelPLA’s PLA and PETG. Of possible interest is that when I do the orca slicer calibrations on it, the result is much closer to the default profiles for BBL filaments than other filaments that I’ve tried. On the downside, the grey has a very blue tint look to it, and the range of colors to choose from seems sparse compared to bigger brands. In regards to speed, it claims both it’s PLA and PETG filaments are “high flow” and capable of printing at 500mm/sec. It does seem to handle high speeds better than cheaper brands which make no claims of being high speed, but that is a subjective impression.

If there were some kind of good benchmark test for assessing filament performance at high speed, then I think it would help a lot in qualifying different filaments for use on the X1C. A lot of us (maybe most of us?) bought theX1C so we could print a lot faster than we could on whatever printer we had before, but without sacrificing much if any quality, and so finding filaments which print at good quality on the X1C at high speeds is important, because not all filaments do. Unfortunately, I don’t have a neat compilation of proof regarding that, but I have a developing impression that it’s true.