Print 3D

Edit:03 mars 2016, Cre:21 juil. 2015


What is it ?

PETG stand for Polyethylene Glycol.
It is an improved material compared to PET as it is less brittle, but its characteristics are different. It is supposed to melt at a lower temperature.
I successfully printed it on the Fisher Beta and my own D-Box delta printer, and it is the material I use at 90%.

Note that as a lot of plastic, PETG is a material which may be produced with a wide range of characteristics.
Unfortunately, basic things as material specification are way too often neglected by filament suppliers. I have no data about thoses I bought (Reprapper and esun). It is really a shame, especially considering that the filament is sold at more than ten time the raw material cost, what are we paying for ?

One characteristic (which is what interested me first) is that the maximum usable temperature is much better than for PLA and may go from 70° to 100°, the ability depending from chosen PETG nuance (which is never documented...).
So this is a good material for printer parts : extruders, motor supports, enclosed machine parts.
There is an important note in the wikipedia page, which tells that if the filament is not dried, the water will hydrolyse with the material while heated, then reducing its mechanical resilience. So, you won't notice damp filament, but your parts will be more brittle. As for other filaments, you shall store filament in containers with driers.

I started my first PETG prints with an enclosed DC motor (for a garden train monorail):

Material tested

I tested (all China made):

  • 3 reprapper PETG spools, 1 translucent, 1 red translucent and 1 green translucent (bought at imprimante3D France)
  • 1 esun PETG blue translucent spool (bought at
  • 1 esun PETG yellow translucent spool (bought at reprapworld as 'budget' PETG, they have other PETG under own brand)

I shall be clear here: while they share the same chemical name, The reprapper and the esun PETG are NOT the same material and they behave very differently.
Transparent Reprapper need much lower temperature and is much easier to print. Adhesion is much better and part are less brittle.
Having started with the reprapper spool, I had a lot of problems with the esun till I understand that it shall be deposited very hot, till they became stringy. In addition, the first spool I get (the blue one) had serious difficulties for unwinding and could not have been unspooled by the (weak) original extruder of my Fisher delta printer. The second one (the yellow) was working nicely for its first quarter then started making knots and totally blocked the filament, even with the geared extruder.
Those suppliers give no numbers, but I assume that, needing a very high temperature to melt, the esun resist better to high temperature. The esun have the reputation to be brittle. In fact, if deposited at high temperature, it not as brittle as you experience at lower temperature, but it is anyway more fragile than the reprapper.
One advantage is that the esun is more slippery and then, may be more useful for sliding stuff (printer carriages ?)
If you try to deposit at high speed, you see the filament became whity. This is because it needs a lot of energy, without a long heating zone, and cannot be heated at high flowrate. Slow down ! It could be wise to define a flowrate limit in your slicer. I have yet to make tests to get numbers, but the reprapper flowrate possible with my hotend is much higher.
With the esun:

  • If your layer is whity-> no adhesion, increase temp or decrease flowrate
  • If you don't see the underlying layer by transparency -> no adhesion, increase temp or decrease flowrate
  • If your print is not shiny-> no adhesion, increase temp or decrease flowrate
  • If its is not stringy, you are not at sufficient temperature
  • If it leaks a lot, you are too hot. In fact, you need the temperature to be adapted to the flowrate, but this is a bit over the capabilities of the slicers.

With the reprapper:

  • If the part is not somewhat translucent, you shall increase temperature. It is not as shiny as the esun, so this is not a valid indicator. I had temperature control problem on one part (too near to the limit) and you see the variation of the temperature in the layers by the whiteness of them. White is insufficient temperature and drive to poor adhesion.


I succeeded to print on a non heated bed covered with 'buildtak'.
To have proper adhesion on a non heated bed, you need a first layer with a sufficiently wide extrusion. I set height to 0.35 and width to 0.8. If I left Slic3R set its default for 0.35mm first layer, the filament did not adhere.
Good level calibration is imperative. Hopefully, it is automatic on my machine.
However, considering the difficulties I have on the first layer, I think that, if available, a heated bed is preferable.
To print on a non heated bed, the bed shall be not thermally conductive (no aluminium), as this cool down the buildtak and drives to warping. Other than that, there is no warping with PETG.

The PETG is much softer than PLA (elasticity is a bit less than half) and so the filament extrusion length output may be slightly lower. On my machine, I found that the extruded length was 2% less than for PLA. Less significant than most diameter variations.

Part fan shall be shut down. If not, the adhesion is really poor as it cools the filament during deposition. In addition on the Fisher, the part cooling flow very efficiently cools the hotend block, which could not go over 250°C if not insulated.
On the Fisher, you cannot shut down the part fan (it also cools the hotend), so I removed the part cooling top and bottom ducts (there is only one screw, but a screw shall be reinstalled as this maintains the effector assembly) and taped the holes in effector (this is required). On the Fisher 1, the top duct is not removable, so you shall install a bottom deflector to send the flow outside of the hotend area, may be redirect it upwards.
As no cooling drive to thermal management problems, some very limited part cooling may help, but this shall not be directed toward the nozzle. If there is no forced cooling at all, the minimum layer time shall be set at approximately 25 sec and for small parts, you may somewhat decrease the temperature.
Any cooling directed to the filament extrusion area will severely impact adhesion.
If your cooling duct is not directed toward the hotend, you may use limited cooling. On The D-Box, I set 0% for a layer time of 60 sec and 35% for layer time 20 sec. Between these two threshold, the cooling is proportional to time. On Slic3r I set 20sec. as minimum cooling time, which made the printing slow down to comply with this time. I set the minimum speed to 10mm/sec, lower may create jams. Slic3r does not slow down below this speed and may not comply with required time, as it don't stop the print between layers. For very small parts printed alone, it is recommended to print simultaneously a 'thermal discharge' column of 10x10mm, same height as your part.

Extrusion temperature shall be high. I have not measured the real temperatures on my machine, so the temperature I gives you are just for information and shall be adjusted according your machine and your filament (I assume that my machine gives 10~15°C temperature over the real, but I was not capable to measure accurately with contact thermometer):
- Transparent Reprapper extrusion temperature for first layer : 268 °C, next layers : 265°C (manufacturer recommend from 210°C to 260°C). Colored reprapper PETG needs higer temperature, say approximately 285°C.
- esun extrusion temperature for first layer : 285 °C, next layers : 295°C (manufacturer recommend from 210°C to 260°C).
295°C is a bit too hot, but it allows a larger extrusion flowrate. NO COOLING is imperative for esun.

With these parameters, the cohesion of the piece is good for reprapper. I have made a break test part and if in the direction of the filament, the reprapper PETG is not brittle, it remains somewhat brittle in the interlayer direction. esun PETG tend to have adhesion problems, this is related to its high temperature requirement and the flowrate (so, the speed) shall be limited to maintain proper adhesion.

If your printer did have difficulties going that high in temperature, try to insulate your hotend with polyimide tape. On the Fisher the all metal hotend have a relatively short heating zone, so needs more temperature to transfer heat to filament.

My first part with Reprapper PETG at 250° and an extrusion coefficient of 1 had a lot of interlayer adhesion problems and ended in the bin.
At 265-268°C (reprapper translucent), the extrusion is very easy, but the extruder is leaking.
At 295°C (esun), the extruder is leaking a lot.

I tested 0.15, 0.2, 0.28 and 0.3 layer thickness. The lower thickness is delicate because the filament cool too quickly and you have to slow the speed.

For a given speed, the margin for the temperature in thin, within a range of 5°C.
If the temperature is too high, you got problem and cleaning support may be very difficult.
If the temperature is too low, you will have adhesion problem and you also have other problems.

With transparent Reprapper PETG, I tested a wide range of speed, from 50 to 180 mm/s. The later is possible, but does not allow proper adhesion and increase the risk of incidents.
60 mm/s for the perimeters and 80mm/s for the infill seems to be a relatively safe bet with 0.2mm layers for Reprapper PETG. For standard hotend heating zone, you may lower these speeds for 0.3mm layers because of difficulties to transfer heat to filament (which may be desirable as that makes stronger parts).
With esun PETG and standard hotend you may not exceed 50mm/s for 0.2 mm layers and 50mm/s (all speeds) for 0.3mm layers - for my hotend - longer heating zone will allow higher speeds (80mm/sec is the ultimate for esun PETG with very long heat zone).

For dimensional hole accuracy, you shall slow down significantly the small perimeters.

For speeds over 60 mm/s, if your print is going bad, slow the speed to 50~60 mm/sec and you may be able to recover your print.

The main difficulty is that the PETG is cooling fast while deposited and became hard very quickly,and the defaults (hard blobs) propagate through layers.


It is imperative that :

  • Your bed shall be flat
  • your retract is properly configured and you shall retract for EACH move (this may not be the default in your slicer, and this is not in Slic3r).
  • Your bed shall be flat and your calibration shall be well done.
  • You shall not overextrude.
    Overextrusion may creates small filaments parts wich are dragged by the hotend and may creates problems. That means the print need constant survey. Before understanding the above, I had to pause from time to time some prints to remove some blobs.
  • You shall be at high temperature, much higher than where you can extrude. You shall not be too hot because that makes the hotend leaking, which cause problems (blobs, again).
  • Your bed shall be flat (get it ?) and your Z-offset shall be appropriate. If you are too near to the bed, your hotend will 'scratch' a lot the first layer while depositing the second. This is especially important if you do not have a heated bed. To cope with unflat bed and have adhesion of first layer, people 'press' the first layer against the bed. NO WAY with PETG. Large deposit width is ok for first layer (I use 0.8mm for 0.35 mm first layer).
    It shall be understood that a default in a layer (a hard blob) could propagate though ten layer or more and at the end, kill your print. If noticed (the 'scratch' noise is well audible), you shall pause the print and carefully cut the blob with a sharp cutter.
    A poorly started first layer may be saved by slowing down to 50% the speed during the 3 or more next layers.
  • You shall have a high Z-lift, which is easy for a delta, but slow down machines with a Z lift done by screw.
    I use 0.8 mm. More may be better (that's where delta printers show their muscles !).
  • Your print shall be translucent (do not use opaque filament, that will make the temperature adjustment difficult), if not, increase temperature or decrease speed.
  • If you use esun PETG, your print shall be shiny.
  • For some brand (esun), you need a high temperature and a hotend with a relatively long heating zone, or your extrusion rate shall be limited.


Retracting the filament shall be done for any hotend move without printing to reduce the filament strings ('oozing'), and you shall modify your slicer parameter accordingly.
All-metal hotend manufacturers recommends to have a retract length lower than the transition zone. So, they recommend generally a retract not exceeding 2mm. That may be the need for a direct extruder, but that don't work for 'bowden tube' extruders.

On the Fisher and now on the Lily, I have a Bowden tube extruder, which while short (460mm), is a bit tortured. This means that the compression of the filament (and extension of the tube) is significant. PETG is much more flexible than PLA and significantly compress in the bowden tube, even if the extrusion pressure is lower. This means that with a bowden, you shall increase the retract value with an estimated value of the compression.
I get much better results and significant string reduction while increasing the retract from 2mm to 3.5 mm and increasing the retract speed. Also, I retract for each move. On the [D-Box], longer bowden tube (680mm) the retract is set at 4.8mm and tests shall be carried with a slightly longer retract.
With the direct drive, the retraction speed I set was 150mm/s. Now, with a geared extruder, I reduced it to 80mm/s.


Supports are difficult to manage with PETG. You shall not cool in order to not cool the filament. That means that the part remains relatively hot and the support are fused with layers. Also, while properly deposited, PETG is not brittle and you cannot break the supports.
Also, supports tend to be pushed away by the nozzle and they often create major problems.

  • Avoid supports if possible.
  • Increase space between part and support (> 0.3mm)
  • Design your own supports. With PETG, it is much easier to cut a thin wall with a cutter than to try to break a melted pillar
  • Bridge where possible
  • Print supports at low speed (30 mm/s), to avoid them being pushed aside.


You shall not cool, especially with esun.
With small parts, it could be delicate as they have no time to cool. I set in the slicer a minimum cooling time of 25 seconds between layers, and the absolute minimum shall be 20 sec.
Reprapper PETG, again, is more tolerant and I ended to cool the short time layers with 20 to 35% cooling flow, with a directive and low flow duct.
As the part remains hot, it stay flexible and is somewhat displaced by the nozzle, but it normally have cool down to a state where it is elastic and it spring back to its place, without noticeable effect on print.
You shall not set a too low minimum speed as it create risks of jams, so the minimum speed may be 12~15mm/s. That means that small layers are difficult to manage with PETG.


Depending your hotend melting zone length, you may be obliged with short melting zone to increase the temperature. That decrease the overhang capabilities. With my quite high temperature, I am capable to do 40° overhangs, and I design my parts accordingly. The behaviour is a bit frightening, anyway, as you see the overhang displaced while the hotend is moving, but it is elastic and came back. In fact, as you shall not cool, the part remain quite flexible while you print and you see it moved by the hotend, but it print ok, anyway. Overhang shall have a minimum width of 2.5mm, if not, the pointy part will have problems.

What is the problem with a non-flat bed

  • Any bump in your bed will make the hotend scratching the first layer during the deposit of the second layer, that may push aside filament and create blobs. These blobs are very hard and will creates troubles which propagate through layers and could kill your prints. That is especially problematic if you have a hole in your print at this location. Pause your print and cut blobs with a sharp cutter. If the filament perimeters around holes have gone away, restart the print.
  • Any hole in the bed may create adhesion problem of the first or second layer, especially if you are in a large area where the hotend go fast and have difficulties to heat.

It may be useful to have the second layer much slower than the further ones (50~70%). However, as the slicer I use did not have such option, I managed that manually.


  • Bridging works well, but I use an extrusion factor of 0.7 while bridging. Speed 80 mm/s (initially, there was no fan, so no point to slow down). Now I have controlled fans on both printers, so during bridging I cool at 100%.
  • Overhang are ok, but tiny overhang tend to be displaced by the hotend and create some 'scratching'. This scratching is not as deadly as the first layer one.
  • Supports are very difficult to remove, as they are fused with the part.
    Due to the accidental dragging effect, you shall not use pillars for supports. Rectilinear goes better. I try to avoid support as much as possible for PETG.
  • Small holes dimensions are reasonnably accurate if you are not too hot and you do not overextrude.
  • Easiness of extrusion. The PETG extrudes very easily and is not very demanding for your extruder. That shall not fool you, and the used temperature shall be much higher than the temperature where PETG start to flow.

Note about breaking test and printing small parts.

With an simple hotend with limited heating length, the capability to heat the material if flow increase is limited. That means that when you print alone small parts, the speed is reduced for cooling (depending your slicer setup).
Smaller speed increase the real filament extrusion temperature (for a fixed hotend value) and so the part adhesion.
This means the mechanical properties will be much better for small parts than for larger parts print at 'normal' speeds.
So, to be representative of real world, the breaking test part shall be printed along with other parts at full speed.

I typically have adhesion problems on large areas, especially for the esun PETG due to high temperature requirements.
I no longer use rectilinear infills, as the interconneting surface between the infill layers is low and as the speed is high, I prefer honeycomb. With honeycomb, the filament layer is deposited exactly above the other layer, so with a much larger interconnecting area. Also, thats slow down the print, so improve the adhesion. That takes more time, anyway.

Hotend requirements and adjustments

Most hotend are not adjustable, so you may not be able to tune your hotend to PETG printing.
DTA labs sells a fully adjustable hotend, the Prometheus. I have installed it in the D-Box, with an extended heating zone (2 nuts with a hotend 0.4mm). The longer heating zoner helps, however, you still need relatively high temperature and careful checking of adhesion.
They made a quite detailed explanation of the effects and trade-offs of the hotend parameters, which is a must read for everyone.
It is interesting to note that their recommendation for standard heating zone length (16mm) is nearly twice the heat length on the Fisher hotend (designed for PLA at low flowrate).

Smell and burns

Even while overheated, the PETG have less odor than other materials. However, in the D-Box which is enclosed, there is a noticeable odor, which tend to stay in the box. This odor is stopped by the carbon filter (no smell behind the filters) but not removed quickly.
PETG does not burn easily, but if you let it stay in the hotend at its maximum temperature for long time, it became somewhat brownish.
On the esun blue translucent, while the machine slow down (to let layers cool down according minimum required time), the temperature becames high and the color slightly darken. This is not visible on the yellow translucent esun.

Insulation of the heating block

To attain the high temperature given above,I had to insulate the heating block on the Fisher (no problem on the Prometheus hotend of the D-Box). I have done this insulation with three layers of polyimide tape. My first insulation was a poor job and was too near to the print, so blobs were gluing on the tape and I had problems. So, I redone the insulation properly and most problems are gone. You shall cross your tape layers and finish each layer on the fin side. You will need tweezers and to well push the tape on the block and each layer on the former one (use toothpick). The tape shall clear the nozzle.

Where to find PETG in Europe

Imprimante 3D France sells reprapper PETG, which is yet the brand I recommend. I ordered one translucent spool.
La boutique du 3D (France) sells esun PETG. I ordered blue translucent.
Reprapworld (The Nederlands) sell esun PETG as 'budget' PETG and have own brand (real PETG: to be tested soon).I ordered the 'budget' yellow translucent not knowing it was esun. Quick delivery. (Denmark) sells what seems to be the lowest cost PETG in Europe, but my question about brand and origin remains unanswered. Ordered other stuff to them, but no PETG without info. Quick delivery.
The lowest cost PETG could be bought in China (~11 Euros/kg) at Winbo, however, considering the shipping cost, that may be better done within a group buy. Also, toxicity of products from China is a big concern, see Reprap Wiki : Health and Safety


As already underlined, there is no data from filament manufacturer, so you can look to properties of miscellaneous PETG from different manufacturers. You can see that the properties variability is quite large, notably for the deflexion temperature (from 69°C to 99°C)
Matweb gives an overview of properties :
Other miscellaneous


Thread about PET and PETG and PETG retraction on RepRap forum
PETG review and Mechanical tests on 3d-printizer.

(c) Pierre ROUZEAU
Éditer - Historique - Imprimable - Changements récents - Rechercher
Page mise à jour le 03/03/2016 01:28