Diamond hot end potential print speeds and pressure advance testing
-
@bot:
If you have an extrusion multiplier of .01, there could be many times when those extruders never produce a single step. Look at a typical gcode file you print, see the distance the E axis is driven, and calculate how many steps that is at your e steps/mm. If it is less than 100, your other filaments are likely not being driven often.
I was talking mixing ratios, not extrusion multiplier bit I think I see where you are coming from. I guess for very small moves, the 1 % won't actually be acted on as it would be less that 1 micro step, which would potentially lead to maybe a maximum of 2% under extrusion on very small moves. In practical terms, I can't say that this is visible in the final printed object.
-
David,
Assuming you are right, and most of the pressure build up is due to the filament buckling and taking up space in the Bowden tubes, then I'd have thought that as my Bowden tubes are only 160mm long then I'd need a very low number for pressure advance. You say that when it's melted the filament is a fluid and can't be compressed. So the compression happens in the unused tubes (but only because the filament is solid and assuming a helical shape). However, the length of these unused tubes is only 21mm. As there are two of them, that would equate to an increase in Bowden tube length of 42mm. If we add that to the 160mm, we still only have an effective Bowden tube length of 202 mm. Even if we take all 3 Bowden tubes and add 21mm to all three, we still only get a total length of 543mm which is by no means unusual. I'm having a job believing that this is all down to the filament acting like some sort of coil spring (or 3)
Ian
-
I take your point that with short Bowden tubes, you should need less pressure advance than with long ones.
Are you saying that you have one Bowden tube that is 160mm long, and the other two are only 21mm long? I thought they were all about the same length?
You can work out what additional movement pressure advance is causing at the extruder drives. To do this, first work out the extrusion speed. For example, for a move that prints parallel to the X axis, take the E parameter (if using relative extrusion) and divide by the amount of X movement, then multiply by the feed rate for that move and by your speed factor. Convert that extrusion speed from mm/min to mm/sec. Then multiply by the pressure advance. This will give the additional extruder movement in mm.
-
@bot:
If you have an extrusion multiplier of .01, there could be many times when those extruders never produce a single step. Look at a typical gcode file you print, see the distance the E axis is driven, and calculate how many steps that is at your e steps/mm. If it is less than 100, your other filaments are likely not being driven often.
I was talking mixing ratios, not extrusion multiplier bit I think I see where you are coming from. I guess for very small moves, the 1 % won't actually be acted on as it would be less that 1 micro step, which would potentially lead to maybe a maximum of 2% under extrusion on very small moves. In practical terms, I can't say that this is visible in the final printed object.
Oh, no doubt it wouldn't have a visible effect. I was just commenting because I guessed the missed microsteps may prevent any pressure-equalization that driving the other extruders may have. It seems irrelevant now.
-
There's a simple test to see filament compression when cold. Extrude slowly until you hear the motor is about to missatep (on duet this might be harder since is much more silent). Then mark the filament on the feeder and then release the lock mechanism of the feeder. Then you can see how much mm compression you had while cold inside the bowden.
I did this test long ago before moving to directdrive and indeed the filament compress and flex like a spring.
-
There's a simple test to see filament compression when cold. Extrude slowly until you hear the motor is about to missatep (on duet this might be harder since is much more silent). Then mark the filament on the feeder and then release the lock mechanism of the feeder. Then you can see how much mm compression you had while cold inside the bowden.
I did this test long ago before moving to directdrive and indeed the filament compress and flex like a spring.
Did you mean "Extrude slowly and gradually increase the extrusion rate until you hear the motor is about to miss a step"?
-
David,
No the Bowden tubes themselves are all the same length - i.e 160 mm. The extra 21mm is the 2mm diameter hole inside the Diamond nozzle from the heat break down. My point is that, if pressure build up is purely due to the 1.75 mm diameter filament being compressed by being pushed into a helix inside inside a 2mm tube, then the only thing that would affect the pressure build up would be the length of the tube(s). So adding 21mm to length of the Bowden tube makes it 181 mm and if we assume that, even with a single filament, pressure is building up in all three tubes, then the combined length is (3x181) 543mm. The Wiki states that with Bowden tube set ups, a value of 0.1 to 0.2 is usually about right. I'd have thought that would relate to typical Bowden lengths of say 300 to 600mm? In my case, the combines length of all three tubes fits within that range yet I have found that a pressure advance value of 0.5 is required. Therefore, my conclusion is that compression of the filament within the tube is not the only factor that affects pressure build up and possibly has less of an impact than the theory might suggest.
It's not all about compression. There can be a build up of pressure without any compression taking place. If we ignore the fact that the hot end has a hole in the end for now, as we force filament in at say 1N/m^2, it will take some time for the pressure inside the hot end to build up, depending on the volume of the chamber into which we are forcing it. Because pressure is force per unit area, the bigger the surface area, the more force is required to attain a given pressure. So a large volume has more force stored up than a small volume if they are both at the same pressure. So when we force filament into a large chamber, it takes much more to reach the same pressure then if the chamber was smaller. Hence the need for more compensation (in terms of length of filament). The hole in the end of the nozzle simply slows down the rate of pressure build up.
Added to the above, we have the fact that we are heating the cold filament. As the volume of the chamber is fixed, the filament cannot expand so the pressure builds up. The higher the melt rate, the higher the pressure build up.
Personally, I think that these factors have more of an impact on pressure build up than current theory would have me believe. I think it's a better explanation of why the Diamond hot end needs more compensation than "normal". Compression of the filament within the Bowden tubes cannot be the overriding factor because the combined length of all three, isn't enough to explain the fact that the Diamond hot end need 2 to 3 times more pressure compensation than is deemed to be normal (for that length of Bowden tubing).
I guess we could spend time quantifying the effects of all the various parameters but I'm too old. It would be a good thesis for some youngster to explore.
I'm happy to just accept that the Diamond, on my machine, with my extruders, my Bowden tube lengths, my heater, etc etc needs a 0.5 pressure compensation value.
Ian
-
There's a simple test to see filament compression when cold. Extrude slowly until you hear the motor is about to missatep (on duet this might be harder since is much more silent). Then mark the filament on the feeder and then release the lock mechanism of the feeder. Then you can see how much mm compression you had while cold inside the bowden.
I did this test long ago before moving to directdrive and indeed the filament compress and flex like a spring.
Did you mean "Extrude slowly and gradually increase the extrusion rate until you hear the motor is about to miss a step"?
No, he said do it cold. So essentially just push filament into a blocked Bowden and see how much you can force in. Interesting concept - might give it a try but with my Titans, I'd have to choose the point where it starts to grind away the filament.
-
David,
…The Wiki states that with Bowden tube set ups, a value of 0.1 to 0.2 is usually about right. I'd have thought that would relate to typical Bowden lengths of say 300 to 600mm? In my case, the combines length of all three tubes fits within that range yet I have found that a pressure advance value of 0.5 is required...I'm not saying there isn't something about the internals of the Diamond that needs more pressure advance, but I am saying that a 3-in mixing extruder will need 3 times as much pressure advance as a single-filament extruder, other things being equal.
The figure of 0.1-0.2 sec in the wiki was based on feedback from other users. Maybe I should do some more tests on my delta with its 600mm Bowden tube, including high speed tests like yours. I was wary of trying higher pressure advance than 0.2 because my E3DV6 is prone to jamming. If I reduce the retraction distance enough then it may be OK.
-
Hi David,
Please don't take my comments the wrong way - I wasn't being critical of you or the Wiki. I think we are both agreed that the Diamond probably needs three times more pressure advance. Where I amicably agree to disagree is the mechanism behind that assertion. Maybe for now, it might be prudent to change the text of the Wiki to something like "Typical values are 0.1 to 0.2 but can be in the range 0.01 to 0.5 depending on characteristics of the extruder, hot end, and filament path and can also be affected by extruder acceleration and jerk settings", or sum such loosely worded statement.
When I get chance, I'll do some more testing to see if the amount of pressure advance can be reduced at lower (normal) printing speeds. I did a quick test to make sure that nothing was horribly wrong using 0.5 pressure advance at low speeds but didn't do any extensive testing. It is possible that I may need less pressure advance at lower speeds although my initial findings didn't indicate that would be the case.
The effect of pressure advance is not an easy thing to quantify as it relies on somewhat subjective examination of the print quality. I guess also, it will be affected by the extruder itself as well as the acceleration and jerk settings so could vary wildly from user to user.
Ian
-
Since the filament doesn't fit exactly in the bowden tube, there is not only the filament compressing, but the filament MOVING inside the tube. This is probably much greater than the filament compressing, but once the slack is taken up, only the filament compression adds further. This could possibly explain why 3 bowden tubes needs more pressure advance than 1 of the same length: you;re getting 3 times the "backlash" as well as the combined "compression."
-
Yes, possibly but it's all conjecture….....I'm a cynical old sod so need a bit more convincing. Show me some proof that 3 short Bowden tubes need three times as much pressure advance compensation as one long one and I'll believe it.
It just seems to me that everyone is hung up on this theory that the only thing which can increase the need for pressure advance is Bowden tube length and things like chamber size or melt rates have no effect. It may be right but until I see clear scientific evidence that proves the theory, I'll keep an open mind.
-
…..................................................
You can work out what additional movement pressure advance is causing at the extruder drives. To do this, first work out the extrusion speed. For example, for a move that prints parallel to the X axis, take the E parameter (if using relative extrusion) and divide by the amount of X movement, then multiply by the feed rate for that move and by your speed factor. Convert that extrusion speed from mm/min to mm/sec. Then multiply by the pressure advance. This will give the additional extruder movement in mm.I've just done these calculations which was an interesting exercise - thanks for the method David.
So initial X position, 28.475, final X position, 326.575 therefore X distance = 298.1mm (this makes sense given that the object is 300mm long with two perimeters at each end). E amount was 14.75732. Feed rate 6,000 mm/min (100mm/sec). So E speed comes out at 297.03 mm/min = 4.95mm/sec. Using my pressure advance of 0.5 that gives me an extra E amount of 2.475mm. Looking at 2.4mm on my vernier calipers, I could well imagine that filament could buckle inside a Bowden tube by that amount, especially if there is some backlash in the fittings. I assume that I can drop the feed rate in the calculation and leave the distance and E amount the same (can't see why they would be different). So if I drop the feed rate down to 3000 (50mm/sec) which is probably closer to what most people print at, then the extra E amount drops to 1.2376 mm and I have no problem believing that 1.75mm diameter filament can buckle inside a 2.0mm ID tube by that amount - even one that is only 160mm long.
What is also interesting is that if I reduce the pressure advance in the above calculations to 0.1, then the extra E amount becomes 0.495mm at at feed rate of 6,000mm/min (100mm/sec print speed) and 0.2475mm at a feed rate of 3,000 (50mm/sec). A quarter of a mm of filament movement seems to me far too little on a long Bowden tube (come to that, even on a short Bowden tube). So from that, my conclusion would be that for Bowden set ups, the amount of pressure advance required is somewhat higher than the Wiki would indicate. Perhaps 0.2 to 0.3 per 100mm or some such?
Of course, it's eminently possible that I've got this completely wrong so please step in and correct me…......
Ian -
Why not try the thin walled circular test object (not on vase mode)? Its very clear then if its underextruding at the start of a layer or at the end, and tuning pressure advance is fairly easy.
Retraction is complicating the issue as the amount of retraction does affect this behaviour. The advice is "reduce retraction when using pressure advance" but if I may, why does this need to be vague? If the effect of pressure advance is a known and calculable effect then there ought to be a formula for how much retraction needs to be reduced. I accept retraction may or may not be optimally tuned on any given printer, but lets say a tuning routine could be determined for getting the optimum retraction, then it would be possible to say for 0.1 pressure advance retraction has to be reduced by x amount.
-
Simon,
The trouble I'd have with a thin walled circular object and observing the start and end of layer change, is that my Z axis is slow - it has to be as the bed weighs around 7kg so speed is set to 300mm/min and acceleration is set to 10mm/s^2. This means that there is slight pause on layer change and because of that I use retraction on layer change. Therefore, the start and end of a layer change would be affected by both retraction and pressure advance. Also, with circular objects there are all sorts of other variables and factors that come into play, like short segmented moves, jerk being applied, direction changes happening etc etc. For example, if extruder jerk is set too low, this will have an impact on the overall print speed so if the pressure advance is increased and an improvement in print quality is noticed, was that down to the pressure advance change or was it just a result of printing more slowly?
With a longish straight parallel object, the moves are as pure as they can get. Start at zero speed, accelerate up to speed, maintain speed for period of time, then decelerate. All in a linear fashion parallel to an axis. No jerk involved, no retraction involved, no direction change (well apart from the small extrusion width shift in between long moves). So the only thing that can have an impact on the quality at the start and end of the move is extrusion pressure which is just what we want to compensate for.
Ref retraction - here is my take on it. If you think about it, retraction is only used after a print move. So it is likely that some pressure would have built up during the previous print move. Therefore, when a user sets their retraction amount (usually by looking at blobs at the start and end of a move), it is likely that they have built in to that an element of pressure advance compensation. In theory, the retraction should only need to compensate for the "natural ooze" of the filament at near zero pressure. In order to calculate how much the retraction needs to be reduced by, we would need to know if the user had set their retraction to suit a high print speed (which is likely to build up more pressure) or at a low print speed where there may be very little pressure. The calculations above illustrate quite nicely how pressure advance (in terms of filament movement) is affected by feed rate. So I don't think we can have a magic formula, but what we could probably say is that pressure advance should be tuned before retraction. Or possibly, retraction should be set at low print speed so that it is only compensating for "natural ooze" and not the affect of any pressure in the system.
That's my twopence worth anyway.
Ian
-
I like the scientific purity of what your doing and you do need a clean uncomplicated test. But my suggestion is to print a real object, and by that I mean something that is difficult to print if these settings are wrong, which will manifestly fail completely to print adequately and is sensitive to small changes.
Much as the long straight moves test gives you a comparable test from one setting change to the next, you're optimising your machine to print beautifully filled cuboids, which are essentially not something you are going to print many of in reality, and which would be perfectly fine even without the optimisation, unless printing at insane high speeds. I'm with you that printing above a certain speed probably requires a different model to printing at lower speeds, semi molten plastic doesn't behave the same at all extruder pressures and speeds.
I can see retraction and pressure advance are only related in that you might need a bit less after a move with pressure advance. That now makes more sense.
-
Simon
I take your point but must disagree, although if it works for you then that's fine.
If one goes straight into printing a complex object and it comes out wrong, then where does one start looking to find what's at fault or which setting to change? I'm not optimising the machine to make cuboids - just optimising the pressure advance setting. I wouldn't dream of using the same test to refine retraction for example. For that, I print different sized objects spaced varying distances apart. I use different tests again to refine my extrusion steps per mm. Then when I print a complex object and it doesn't come out as expected, I know it won't be due to pressure advance or retraction or extrusion steps per mm etc so it must be something else.
This approach takes me back to my exhaust emission testing days when I used to calibrate test facilities. You can do a total system verification test by injecting a known mass of Propane over a period of time into the pipe that would connect to the vehicle, and at the same time measure the Hydrocarbon concentration with a FID (type of gas analyser) and the volume of gas using a CVS (type of volume measurement). Then you multiply the concentration by the volume x the gas density and you should end up with the same number (as the known mass that was injected). If the measured value is low, then it could be either the gas volume measurement that was low or the concentration, so you'd have to go back and calibrate both of those to find out where the fault lies. BUT the worse case scenario, is if the numbers correlate well but only because there is a positive error in the concentration measurement and a corresponding negative error in the volume measurement. This is bad because we have only checked using one pollutant (Hydrocarbons) so if there is an error with the volume measurement, it'll throw out the mass emissions results for the other pollutants (CO, NOx etc). That is why every aspect must be individually calibrated and only then is the total system verification test carried out.
So the analogy here is that you could have a pressure advance setting that is too low, and a retraction setting that is too high but they cancel each other out for that particular object. However, if you print another object with the same settings that maybe has longer moves, or you get impatient and increase the speed multiplier, you may get blobs or signs of over extrusion at te ends of the moves because the pressure advance setting is too low and the retraction amount being fixed is no longer sufficient to compensate.
So that's why I prefer to do it the way that I do.
-
I print a bunch of different stuff from icicle towers to single wall cubes to single wall cylinders to the ultimate bridging test. I find that I agree with Deckingman here. keeping it simple allows for a lot of itterations in a short amount of time to narrow down where the issues are (single wall really helps shine a light on change and effect.)
Though as stated I do think having a good mix of fast printing objects is the key to fine tuning.
Then when youre all done print up a few benchys and lattice cubes and if you are really feeling dialed in try out "The tower of pi"
You guys know a butt load more than me though so take what i say as a grain of salt.
