bondtech e-steps weird values

[dc42]

How are you doing the calibration? Are you using nonlinear extrusion, https://duet3d.dozuki.com/Wiki/Gcode#Section_M592_Configure_nonlinear_extrusion ?

[arhi]

not using non-linear (no clue what it is yet)

Originally I did calibration by

• remove hotend
• measure how much filament is out of extruder, reset caliper to zero
• extrude 100mm
• measure how much more is out now
• modify E value accordingly

repeat usually 2-3 times to close on exact value

Now, I'm not doing that, I started with "much lower" value, and I'm printing same 3 shell cube sliced with 100% extrusion multiplier, increasing E value a bit every time print finish waiting for a gap between perimeters to be completely gone. (I hoped I'll close in on it faster, now after 10 5mm heigh cuboids I figured I better used something bit bigger like 80x50x5mm so I can actually use those as tray's, the 20x20x5 ones can only go to recycling bin)

[oliof]

IMO you cannot calibrate esteps with a flow calibration procedure and multiple shells will give you wrong values anyways because thermoplastics will affect each other line by line and falsify the layer width.

[arhi]

@oliof yes and no. With semitransparent filament and microscope I can see exactly how they affect each other and how much they overlap so I can get a pretty good calibration that way.

What's happening now is it looks like speed affects extrusion amount, but in a weird way?!?!?! going faster requires modifier to go down not up ?!?!? None of these issues were visible with .3 nor .4mm nozzle .. but .6mm behaves very different.. never experienced that before but normally I used .6 and .8 on hexagon, not e3d... it's probbly something stupid I'm overlooking... need to redo the whole thing from scratch and it will probbly go away

Anyhow, found some nice tricks in IM to go around issues I was experiencing ... (I can set separate flow% for first layer, top, bottom, infill ...)

[arhi]

https://reprap.org/forum/read.php?262,802277 ... When I read that thread first time (when it was first written, 2-3 years ago) I did the test myself and did not have that problem.. probbly shallow toothed big diameter screw I was using was handling this better, or my nozzle was big and ABS is not behaving same, dunno .. but now I'm reading it and testing and .. damn .. that's the problem.. I calculate at one speed, print at another ... major issue I had printing first layer where it was overextruding a lot (I print first layer slow, 10-15mm/sec or rarely 20mm/sec) and then I had perimeters not touching later on when it switches to 100mm/sec .. no skipped stepps but .. and looking at filament you can see the teeth go differently... looks like dual drive bondteck is affected more than old single drive extruders... maybe time to make again a good big high precission extruder with 2 huge wheels like I seen on professional machine

[bot]

If you're saying you need to reduce the e-steps/mm for a larger orifice nozzle, that would suggest the theory about increasing back pressure = lost motion is correct.

What I think may be happening, is with smaller orifices, there is a quicker rise in pressure, and the filament needs to go somewhere (if the drive is actually driving it) so it is forced to go up the hotend, back where it came from, creating that characteristic bulge at the end of filament after being driven through an e3d extruder.

The less pressure there is resisting the forward flow of filament, the less (in theory) the filament is flowing back up the hot end.

[arhi]

@bot said in bondtech e-steps weird values:

If you're saying

I'm saying it's confusing to me attm.

I found one reason for the $%^%$#^* that was happening; I had an oily spot on the printbite and it was doing something weird with the filament after head goes trough there. Cleaned it with acetone and now that part of the problem is gone.

Big vs small nozzle, talking about this particular setup here (BMG+E3DV6 is not something I used in past 10+ years, I used Wade's extruder and number of extruders forked from Wade's extruder, some designed by me, some by others, also Flex3Drive (old version, not modern G5, I used G5 only a bit) with hobbed bolts machined by Arcol in Budapest that are a lot different than regular hobbed bolts or modern drive gears (they have super sharp teeth but the pitch is rather large, comparison M4 tapping there vs M3 tapping on your regular hobbed bolt, only this was machined using some cutting rig so teeth are not angled but straight), I used either heaxagons, or 1730 or makergear hotends or stuff I machined myself and I'm mostly printing ABS/HIPS from 3mm filament .. so what I'm used to is very very very very different setup than this BMG+E3DV6..

So what I noticed is that on this setup E3DV6+BMG with .3mm nozzle I did not notice a significant error in e_step calibration with the change in speed. Of course, the speed with .3mm nozzle is a tad limited so we are talking .2mm max layer height, .35mm line, and 40mm/sec top speed. So same e_stop values worked great for both 20mm/sec and 40mm/sec. With .6mm nozzle bottom speed is still 20mm/sec but top speed is 100mm/sec (or more) and what I see is that if I calibrate 60mm/sec to be "perfect" my bottom layer printed at 20mm/sec will be heavily over-extruded, while the rest of the object, where I bump the speed to 100mm/sec will be underextruded (the perimeters don't touch). I'd say that nozzle size mostly allows for a larger range of speed and that speed is what's causing the extrusion error and not the nozzle size.

I was originally confused, especially when I started using 1.75mm filament, why "perfectly calibrated extruder" need modifier when changing material, and then I figured out that teeth dig into filament differently and that they dig in the soft filament fully while they only scrape the surface of PLA making the circumference of the drive gear effectively different (larger with PLA, smaller with TPU and everything in between with other filaments), and even with same material, teeth will dig into translucent PETG less than into white non-transparent PETG.

Thinking that this might be the only explanation to the difference in what is extruded vs speed (as there are no skips) looks like BMG will dig itself into filament differently depending on the force required to push in the filament effectively making the circumference of the drive gear dynamic / dependent on the pressure.

Reading the forum post from few years ago this all makes sense but it seems to me that dynamic here is too high and that something else is in play here as I'm getting to much difference between 20 and 100mm/sec... I have idea how/what to test, and it's great there is dynamic extrusion to test stuff with too, but that will require some electronics and time that I don't currently have so we'll see .. in few weeks maybe

[deckingman]

@arhi It sounds to me like your bottom layer is simply being too "squished" so it looks like it's over extruded. You really shouldn't be using the first layer as a basis for checking the extruder.

I'm not surprised that you get a difference between 20mm/sec and 100 mm/sec with an 0.6mm nozzle. If you calculate the melt rate with that nozzle at that speed, it works out at about 28 mm^3/sec - far above what an E3D V6 is capable of.

In my experience, I found that the BMGs are much more consistent than extruders which use a spring loaded tensioner. I once did an extreme retraction torture test using BMGs and Titans - the BMGs could retract and un-retract the same piece of filament. The Titans completely ground through the filament after about 8,000 cycles but the BMGs were still going strong after 26,000 cycles and looked like they would go on for ever. They kind of bite into the filament so that it acts like a rack and pinion with the "rack" staying the same size, whereas the spring loaded Titans bit deeper with every cycle. https://somei3deas.wordpress.com/2018/05/11/bondtech-bmg-vs-e3d-titan-extreme-retraction-torture-test/

I don't think your problems have anything to do with the e-steps or extruder calibration but are more likely due to layer height/width at low speed, and melt rate limits at higher speeds.

[arhi]

@deckingman said in bondtech e-steps weird values:

@arhi It sounds to me like your bottom layer is simply being too "squished" so it looks like it's over extruded. You really shouldn't be using the first layer as a basis for checking the extruder.

I'm getting gross overextrusion even with no squishing and "round" filament on the printbite (petg does not want to be squished at all, especially not on printbite). I guess it's all normal and expected just bit unexpected to me.

I'm not surprised that you get a difference between 20mm/sec and 100 mm/sec with an 0.6mm nozzle. If you calculate the melt rate with that nozzle at that speed, it works out at about 28 mm^3/sec - far above what an E3D V6 is capable of.

I like to print hot, so I get lot more mm3 than rated, I get easy 40mm3/sec from this nozzle before bmg start to skip steps / grind filament. What I did not have ever taken into account is how the drive gear circumference changes with the "bite pressure", and how "bite pressure" changes with "back pressure" from the hotend. TBH I'm still not clear how this happens. It's obvious bite depth (that makes the circumference change) depends on the spring tension. The bite depth changing on the pressure needed to push the filament is new to me, but thinking about it, after the teeth pass the "horizontal" position, there will be a force downwards and horizontal, the horizontal one is defined by the spring but the downwards one makes the teeth go deeper if there's more resistance to filament movement

In my experience, I found that the BMGs are much more consistent than extruders which use a spring loaded tensioner. I once did an extreme retraction torture test using BMGs and Titans - the BMGs could retract and un-retract the same piece of filament. The Titans completely ground through the filament after about 8,000 cycles but the BMGs were still going strong after 26,000 cycles and looked like they would go on for ever. They kind of bite into the filament so that it acts like a rack and pinion with the "rack" staying the same size, whereas the spring loaded Titans bit deeper with every cycle. https://somei3deas.wordpress.com/2018/05/11/bondtech-bmg-vs-e3d-titan-extreme-retraction-torture-test/

I know your test, it's a very good read.
The extruders I'm using for past 10 years can all be called "fixed" as I don't have a good spring source here I normally just use a screw to fix the idler bearing in place, no "variable" pressure there, and what I mostly used, flex3drive, also uses excentric arm that positions the idler so again no springs... my teartime printers (upgraded to smoothieboard) also have fixed idler (all teartime extruders do as far as I know). I have one titan on one small printer but I used it for maybe 20kg of filament total, not some super useful experience that I have with it so this is my first real experience with springs, and looking forward, I think I'll try to modify the BMG to remove spring from the equation, I think it should be simple, with few washers.. maybe a shim

I don't think your problems have anything to do with the e-steps or extruder calibration but are more likely due to layer height/width at low speed, and melt rate limits at higher speeds.

Well same printer, different extruder, no problem, many other printers, different extruder, no problem. The only difference between this one and ones without a problem is that gears are not fixed.. so attm I'm taking that as possible root cause. Since it should be easily "fixed", I'll see how will it behave after I fix it

[deckingman]

@arhi What do you mean "the gears are not fixed"? Is this a genuine Bondtech or another clone?

[arhi]

@deckingman said in bondtech e-steps weird values:

@arhi What do you mean "the gears are not fixed"? Is this a genuine Bondtech or another clone?

It is a genuine Bondtech. (I have few originals that I use and some clones that I don't use that are not geared, and also some clones that look similar to original but are also not in use, yet). It has 2 drive gears that mesh together, one that's also linked to a motor trough big gear is fixed, the other one that meshes together is on a swing arm that's pressed with a spring. It is identical to what you have (looking at your imagery). If you fully close the arm it would deform the filament as the gears go too close to one another, that's why there's a spring to allow them to come apart to allow filament to enter. What I'm thinking is to add a shim / washer between the main body and the arm to not allow arm to fully close and then to tighten the screw without the spring. Now while I'm thinking about it, no need for the shim, since I just need to fix it from "opening" and remove force pushing it more close, so just removing a spring and using shorter screw should get me exactly what I want (or use the same screw just replace the spring with piece of stiff tube).

[arhi]

@deckingman as for the clones, I'm thinking about ordering these to test them out:
https://www.aliexpress.com/item/4000608604486.html

[deckingman]

@arhi I found on my (6) BMGs that the spring doesn't really do much - unlike the Titans for example. I think it's only there to allow you to pull the arm back for fast feeding of filament. The way the teeth of the two gears mesh together means that it is impossible for the hobbed part to bite too deeply. So once there is enough tension on the spring to bite into the filament, further tensioning of the sprint has no effect. You might as well just tighten it as far as possible - no need for shims or short fixed screws.

As you know, I use a mixing hot end and when I print something like this https://www.youtube.com/watch?v=9zXkNvqrDbA, I vary the mixing ratio in 1% increments every "n" layers. So individual extruder speeds vary between 0 and 100% throughout the print. Never had any problems with speed dependent under/over extrusion.

Why don't you calibrate the steps per mm the "normal" way by extruding say 100mm of filament at low speed (say 5mm/sec). Then without changing the steps per mm, increase the feed rate in say 1mm/sec increments and note the amount of extruded filament. Repeat as necessary and tabulate the results. Then we can see if there is some sort of linear relationship between extruder speed feed amount, or if there is a step change, or something else.

Ref the clones - https://www.bondtech.se/en/2020/01/31/original-bmg-vs-clone/

[arhi]

@deckingman said in bondtech e-steps weird values:

@arhi I found on my (6) BMGs that the spring doesn't really do much - unlike the Titans for example. I think it's only there to allow you to pull the arm back for fast feeding of filament. The way the teeth of the two gears mesh together means that it is impossible for the hobbed part to bite too deeply. So once there is enough tension on the spring to bite into the filament, further tensioning of the sprint has no effect. You might as well just tighten it as far as possible - no need for shims or short fixed screws.

I have very nasty sounds from it when I tighten the spring completely.
I tested the idea today, used piece of PTFE tube cut to size instead of spring and there's almost zero difference in extrusion rate now between different speeds, BUT, now I have grinding of filament at 100mm/sec .. interesting, will need to investigate further.

Why don't you calibrate the steps per mm the "normal" way by extruding say 100mm of filament at low speed (say 5mm/sec). Then without changing the steps per mm, increase the feed rate in say 1mm/sec increments and note the amount of extruded filament. Repeat as necessary and tabulate the results. Then we can see if there is some sort of linear relationship between extruder speed feed amount, or if there is a step change, or something else.

I plan to do that the moment time allows as this seems to be very important thing that I never had to deal with so need to know what's going on

Ref the clones - https://www.bondtech.se/en/2020/01/31/original-bmg-vs-clone/

Well aware of that page, but, I can't say I agree with it completely. Would not dismiss it right away as my clones are still in a drawer so till I have printed with them I can't judge but from the imagery on that page I can say that

1. the el-cheapo ungeared clones I purchased match all the "problems" mentioned on that page, but since I purchased those for a completely different reason (I wanted to use them as encoders for the filament passing and as unspooling devices, to pull the filament from heavy spools if necesary to create slack for real extruder to not have to deal with that .. they are attached to DC motors and I'm super happy how they perform)

2. comparing gears between 2 clones and 2 original BMG extruders I can tell you that there's no way anyone would notice a difference, no difference in color nor weight nor shape, the teeth on these clones are cut straight like original BMG and not with a tap (like on those el-cheapo ungeared ones) and teeth are hardened (tested with a file, the file just slips over teeth, can't bite)

3. the logo is not there of course, not a functional difference, the case on the other hand on the clones is bit different, clearly not 1/1 copy of the original and clearly tougher than the original, very good professional finish, I don't mind the clone case at all!

4. "black POM gear" - my originals have same white gear as clones, I cannot differentiate them. dunno what these black ones on BMG site are but mine originals are WHITE and all original BMG's I seen have white gears. Also a local shop told me that they have stripped those white gears on original BMG's on bunch of them inside 6 months testing and that they printed new ones on their SLS machine that behave much better so...

5. threaded metal pushfit - I don't use bowden so I could not care less but again on my original bmg I don't have this "metal push fit" ... "black thumbscrew" - I would not even comment on this

6. secondary drive gear - now, VISUALLY they are identical as original ones. I can't say how they perform, have not tried yet!!! so when I do I can say if they generate debris or not, tare trough filament or not.. visually - no difference, hardness - no difference (file test)

7. clunky and cheap - well, I seen clones that really look cheap an clunky, but I'v seen ones that look awesome and super professional (I won't show pictures as I do not want to market them, neither the nice nor the crappy ones) .. just 'cause they are clones or PRC made does not mean they are bad...

I understand they found probbly the nasty clone to compare with and for those that need 1 or 2 extruders I'll always push toward original ones.. but.. for experimenting, breaking, melting.. I'm going with clones

the double-drive system is now used by hemera too, is used on 5-6 different extruders these days so .. I'm all for helping the new design and going original (hence my 2 originals) but I need another 4 extruders on the e3d toolchanger, I'm going with 2 bmg clones and 2 original flex3drives, we'll see how will these clones perform, but pricewise I'm pretty sure they will be delightful ... now the two I linked, the full alu frame.. for 10 years I'm talking about extruder like that, I'm sure e3d guy's can find in the email archive when they pushed water cooled kraken I shot them a message that it would be super cool to have a whole aluminium extruder water cooled with kraken nozzle shoved into it... so similar to these aluminium dual gear extruders from PRC that you can push something similar to kraken heatbreak in only water cooled and not air cooled... note these are available for few years already!! e3d just recently published hemera and it is not fully metal nor water cooled