Finally some insight into pressure inside the hotend.


    Full paper:

    @deckingman Looks like some researchers have made a hotend with a pressure transducer built in. The paper gets a little technical for me, but it seems to confirm what we've known through experience. i.e., higher temperatures allow for better flow, more back pressure from lower layer heights, nozzle tip being cooler than heater block, etc.

  • @phaedrux Thanks for bringing that to my attention. I need to re- read the paper again but at first glance, it seems to be a discourse on how In-line rheological monitoring could be applied as a tool to obtain measurements, rather than any sort of practical measurement or testing.

    Basically, the generally accepted theory is that pressure is solely due to filament buckling within the Bowden tube and acting like a spring or that the filament itself gets compressed. Personally, I think this is highly questionable and that other factors such as thermal expansion could have significant effect. This is based on some research and testing I've been doing which is leading me towards the belief that pressure is related to melt chamber size (larger melt chambers have more issues with pressure than small ones).

    Hopefully, the good chaps at Massachusetts will use this technique to give us some data. I'd like to see how pressure builds up (and decays) during the course of a print move and how things like nozzle geometry, layer height, print speed etc effect that pressure (as well as melt volume and temperature).

  • Definitely give the full paper a look. There are some actual measurement results for various material types. Section 4, results.

  • Wasn't there some discussion on this forum or RepRap where someone had built a force sensor into a hotend, of hotend mount and were getting some useful measurements back?

    As a slight aside when I get missed steps on the extruder drive it normally results in the filament uncompressing and back driving the stepper motor. Other than during the fast retract and unretract moves have others noted or suspected less obvious missed steps?

    Good to see open access on the paper. I'd like to see open research as a condition of accepting government research grants.

  • Having seen how much a non-linear advance profile changes with as little as 5C of hotend set point change I think having accurate control of the melt temperature with feed forward corrections for sudden changes in extrusion rate will have a significant effect in process control.

    Without a fundamental change in hot end design (not suggesting I've any ideas!) I doubt melt temp will be controllable with a fixed length of melt zone other than under constant rates of extrusion. This in part due to the thermal mass of the hot end heater block, and also the variable (extrusion rate linked) rate of heat loss from the system.

    An interesting thought occurs when you think about multile heat zones. These would have to be thought of in a fourth-dimensional manner (Marty? 😄 ) as a change in a preheat temperature zone will take a while to get through to the final heat zone and nozzle.

  • Thoughts (edit: review like brutality, not meaning to degrade the contibution of their work - they are playing catch up trying to model what we do.) on the paper:

    It talks about filament slipping but doesn't go into detail. Does it really mean slip or does it also mean the sort of behaviour where the bite marks on filament get closer together with increased extrusion rate? The standard non linear extrusion correction formula in Duet firmware limit correction in feed rate to 0.2, which would be 20% 'slip', but this is repeatable rather than a physical uncontrolled slip.

    Would have liked to see measurements of filament diameter to verify how much it changes during tests rather than just manufacturers point of sale limits.

    We have to convert results presented in shear rates back to volumetric extrusion rates. From Wikipedia [shear] = (8 * [fluid velocity]) / [diameter]. Their last graph shows a shear rate of 1000 as beyond their test range. I make that about 12mm3/sec through a 0.5mm nozzle which equates to a print speed of about 80mm/sec with 0.3mm layer and 0.5mm extrusion width. (edit 2: this would correspond to the sorts of extrusion rates where extrusion rate correction become important for E3D v6 like setups - so sounds about right!) Not earth shattering, but that was off the end on a logarithmic scale, so presented tests will be slower.

    Take homes from the paper?

    Real melt temp lags heater block temp - expected.

    Nothing fundamentally unmodellable about the material extrusion deposition process - expected.

    Beyond my expertise but the relevance of specific physical modelling theories to the process were discussed. Whether or not some models become more significant at delta print speeds would be interesting.

    Apparatus has been developed that can accurately measure flow properties in a static, steady state environment, but little discussion was given to it's behaviour in the very dynamic printing process? Results are presented from prints but little discussion on how head accelerations and vibrations effected it?

    A great start, and a useful tool to validate more obsure methods of melt monitoring but not the prize ticket in itself. For closed loop control to be possible it would need to be functional up to the print speed limit, which was mentioned (optimistically?) to be 200mm/sec for this system.

    We can already print parts 'well' with a 0.5mm nozzle at 40-80mm/sec given a decent machine setup. The need at these speeds is a system that can allow us to get closer to full density parts (where we want it, ie 100% fill or upper/lower/vertical surfaces) without needing to drop our extrusion multipliers to create capacity in the process to absorb geometry related variations in extrusion rates without over build colliding with the nozzles.

    Edit 3: Thanks for sharing the link to the paper. Always interesting to see what academic scientists are upto in the field alongside our styles which range from engineering though to pure trial an error.

  • @phaedrux said in Finally some insight into pressure inside the hotend.:

    Definitely give the full paper a look. There are some actual measurement results for various material types. Section 4, results.

    I've had a bit more of gander but I'm still of the impression that this is an exercise to develop and validate the tools and techniques required, rather than the next step which would be to use those tools and techniques to obtain useful data.

    There are two sections of the "Results" entitled "Pressure verification" and "Rheology verification" and in both cases the FDMRheo data is compared with the CapRheo data.

    The "Discussion" section is much the same comparing the FDMRheo method against two others using a Cross-WLF (Williams_Landel_Ferry) model in sections A and B. Section C starts with the sentence "The following paragraphs are incorporated to illustrate how the FDMRheo addresses some of the shortcomings that are common among on-line and in-line rheometers."

    And finally, the conclusions start with the sentences " The in-line FDM rheometer (FDMRheo) designed in this work has been found to provide very accurate viscosity measurements as confirmed against off-line rheometers. The FDMRheo can collect data across a wide range of temperatures and shear rates to generate a successful Cross-WLF model for analyzing continuous viscosity curves as a function of temperature, shear rate, and pressure.

    So all in all, this doesn't actually give us any insight into what happens to the pressure in a hot end during printing. It does show that the tools and techniques have been developed to provide that data. Hopefully, someone will take the next step. (I'd love to get my hands on the kit.................☺ )

  • @doctrucker said in Finally some insight into pressure inside the hotend.:

    This was the thread I was thinking about:

    Ah yes. I'd forgotten about that. It was great bit of work. The problem is that it was only measuring the pressure of the filament going into the hot end and not the pressure inside the hot end itself. So it's only part of the story.

    Not that anybody takes any notice but from the work that I have done (see my blog) it is clear to me that the current accepted theory that hot end pressure is purely a function of filament pressure going into the hot end is incorrect. If it was true, then retracting the filament at the end of a move should relieve all the pressure but it doesn't. In fact, if you print fast enough you can reach a point where no amount of retraction will relieve the pressure and prevent ooze.

    Not many people can get up to 300mm/sec without hitting the melt rate limit or experiencing extruder slippage, but with a 5 input Diamond hot end you can. Because you have 5 melt chambers and thus increase the melt capacity by a factor of 5, but also each extruder runs at 1/5th the speed so at 300mm/sec each extruder is only running at the equivalent of 50mm/sec. Under these conditions, no amount of retraction will relieve the pressure at the end of a long move and the only solution is to use pressure advance compensation (which works very well indeed).

    As I said, nobody takes any notice of me and I don't really care. But if you pull the filament back at the end of a long, fast move it will relieve the pressure on the filament. If then filament is still being forced out of the nozzle, clearly (at least to me) there must be some residual pressure inside the hot end which is not a function of the filament acting like a spring.

    Which is why I'd like to see some real time measurement of hot end pressure under actual dynamic print conditions.

    Just my twopence worth..........

  • @deckingman No problems with what you're saying there and I said it was a great start, rather than completed work. Edit: Also highlighted that there was little discussion on how print accelerations and vibrations effected it.

    With regards to retract not relieving all the pressure that would be logical. The polymer melt is both compressible, and and highly viscous. In order to relieve the pressure in the melt chamber the polymer would have to decompress and move back. In the moving back action it would need to flow and over come it's own viscosity. Equally pull the filament back too fast and you will induce a tension in the solid filament rather than instantly retracting from the melt. The melt itself would also stretch.

  • @doctrucker said in Finally some insight into pressure inside the hotend.:

    With regards to retract not relieving all the pressure that would be logical.......................

    Hi Wes,

    Yes that's my belief too based on the test work that I've conducted - I think we are in agreement.

Log in to reply