stepper torque vs. geared extruders

Hi guys,
just want to know if there is a "golden cut" regarding decreasing stepper torque at high RPM and the extruder gearing?
I know about current limited BLDC motors, that the torque is constant from 0 to 50% of the RPM range and from 51-100% the torque decreases more or less linear.
How about steppers?
Insanely high gear ratio could put the stepper in trouble, a direct drive slow running stepper might have more torque for printing and a lot of speed reserve for fast retraction.

I´ve seen extruders with as high as 42:1 ratio. Would they still fit the golden rule?
-Olaf

Quote
o_lampe
Hi guys,
just want to know if there is a "golden cut" regarding decreasing stepper torque at high RPM and the extruder gearing?
I know about current limited BLDC motors, that the torque is constant from 0 to 50% of the RPM range and from 51-100% the torque decreases more or less linear.
How about steppers?
Insanely high gear ratio could put the stepper in trouble, a direct drive slow running stepper might have more torque for printing and a lot of speed reserve for fast retraction.

I´ve seen extruders with as high as 42:1 ratio. Would they still fit the golden rule?
-Olaf

dc42 have done some calculations on this and posted on this forum, can't remember which thread though.

His results was that a gear ratio of 3:1 was "golden", higher gear ratio and the extruder chews through the filament to easy instead of skipping steps.

Acceleration is as important as speed, because over a short retraction distance, the max speed is likely to be determined by the acceleration anyway. For example, 5mm retraction = 2.5mm acceleration + 2.5mm deceleration. If the acceleration is 1000mm/sec^2 then the peak speed will be 70mm/sec.

The extruder on my Ormerod works well. It uses a short (34mm long) motor and approx. 3:1 gear ratio.

The extruder that came with my Mini Kossel kit was more problematic. It used a gear ration of approx. 5:1 and a long (48mm) motor. The initial problem I had was that when the nozzle gets temporarily obstructed by the print (e.g. due to over extrusion, or by starting the print too close to the bed), the torque was so great that the hobbed insert would chew a crescent in the filament and lose traction completely, ruining the print. Whereas on my Ormerod, the extruder would skip steps under these conditions, and extrusion would resume when the nozzle obstruction was removed. So I reduced the current in the Mini Kossel extruder to 400mA to get it to skip steps instead of chewing the filament. The problem I then had was that the reduced current was insufficient to accelerate the motor fast enough. In particular,t when loading or unloading filament it would sometimes stall and scream.

One solution would have been to use a lower gear ratio with that long motor, so that I could run it at higher current, more motor torque, and hence more acceleration. Another solution would be to use a shorter motor, run closer to its rated current to produce the same torque as the longer motor did at 400mA. The rotor of the shorter motor would have lower moment of inertia, and therefore accelerate better with the same torque.

In the end, my Mini Kossel extruder broke and I decided I didn't like the design; so I substituted an Ormerod extruder for it - which is working nicely.

So my recommendation for extruding 1.75mm filament is to use one of the following:

1. A long motor run close to its rated current, with direct drive (i.e. no gears)

2. A short motor and a gear ratio of about 3:1.

Direct drive and a short motor seems to be viable for extruding PLA but not PETG, according to one user's experience reported on the Fisher forum.

Higher gear ratios such as 5:1 should be OK if you use a short motor, so that you can get good acceleration without excessive filament drive force.

As for maximum speed, using 12V drive and a long 1.2deg motor with about 4mH inductance and 1.68A rated current, the maximum speed in mm/sec you can before torque runs out is approximately 25000 or 30000 divided by the microsteps/mm, assuming 16x microstepping. Shorter motors with lower inductance do a bit better. Using 24V instead of 12V drive should more than double the maximum speed.

HTH David

Edited 4 time(s). Last edit at 08/04/2015 03:09AM by dc42.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Quote
dc42

Higher gear ratios such as 5:1 should be OK if you use a short motor, so that you can get good acceleration without excessive filament drive force.

HTH David

I have a planetary gearbox with a ratio of about 5:1 and a short motor and this works very well with PLA but not so good with PETG.

When I drive it at it's rated current ~1.7A the PETG filament tends bow to bow and curl up inside either the cold end or in the hotend just below the bowden tube connector where there is a few mm of space before entering the heatbreak tube (chinese E3D clone).
I have to reduce the current to about 850mA for it to work properly but occasionally my motor screams and it misses a few mm of extrusion.
This tends to happen if there at places where there is a "quick retraction", short time between the retraction and the forward push, and I think this is caused by the loss of control of the rotor during acceleration/deceleration due to the loweredcurrent

So right now I'm looking for a solution with a 3:1 gear where I can mount a short motor and my metal coldend (designed to fit directly to a NEMA 17).

Thanks David for the elaborated explanation

Wouldn´t it be helpful for controllers with onboard digital current control ( hint, hint ) to implement a dynamic current control?

Winding resistance, max. current and gear ratio are the known parameters to regulate the desired ( speed depending ) torque.
Full torque on high speed retraction / low torque for average extrusion...
The steppers would stay cooler, too.
All steppers could benefit from dynamic current control.
-Olaf

Quote
o_lampe
Thanks David for the elaborated explanation

Wouldn´t it be helpful for controllers with onboard digital current control ( hint, hint ) to implement a dynamic current control?

Winding resistance, max. current and gear ratio are the known parameters to regulate the desired ( speed depending ) torque.
Full torque on high speed retraction / low torque for average extrusion...
The steppers would stay cooler, too.
All steppers could benefit from dynamic current control.
-Olaf

That's a nice idea. I guess the current could be increased on all retraction moves, and left high on high-speed extrusion moves (because they are probably undoing a previous retraction, or loading filament). Then reduced when a normal extrusion move comes along. The only problem I can see with doing this is synchronising it to the extruder movement. The stepper pulses are generated and new moves kicked off in the step interrupt service routine; but you wouldn't want to do an I2C transaction in the ISR to change motor current because it would take too long..

For the axis steppers, I don't think dynamic current control would be a good idea during printing, because reducing current reduces microstepping accuracy. I guess it might be OK on a screw-driven Z axis, when bed compensation is not being used and the Z axis only moves on layer changes. My fork of RepRapFirmware already reduces current to 30% of normal when all the motors have been idle for 30 seconds, so as to reduce power consumption whole holding the head in the same place.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Quote
o_lampe
Wouldn´t it be helpful for controllers with onboard digital current control ( hint, hint ) to implement a dynamic current control?

This kind of thing already happens in higher quality stepper drivers (and since you can input the coil inductance/resistance and autotune the control loop it is much more effective as well). I think there is the potential for too many unintended consequences if you start screwing with Vref at high speeds while a single chip driver is running.

Quote
691175002
I think there is the potential for too many unintended consequences if you start screwing with Vref at high speeds while a single chip driver is running.

Such as?

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

I didn´t say to change Vref at high speed or during an ISR. There is a print cue in FW, where you can look ahead through the g-code and do the Vref calculations beforehand.

Another feature would be to change the print speed itself according to track length. ( something a very low acceleration would do automatically )
I hate it, when my printer starts to shake, because the print speed is high when the hotend wiggles in tiny moves along the part.
When it prints an arc ( read: lots of short straight tracks ), it should print fast, but at sharp direction changes like 45° zigzag and short tracks, it should slow down.
-Olaf

Edited 1 time(s). Last edit at 08/05/2015 03:48AM by o_lampe.

Quote
o_lampe
I didn´t say to change Vref at high speed or during an ISR. There is a print cue in FW, where you can look ahead through the g-code and do the Vref calculations beforehand.

The problem is not doing the calculations, it's sending the commands to the DAC or digipot at the right time. Moves (except the first one after being idle) are taken off the print queue and started by the ISR, so it's in the ISR that you would need to set the current. However, it should be possible to do it in the ISR if a DAC with SPI interface is used, because as long as the DAC only needs one 16-bit packet, the hardware can deal with sending that without the processor hanging around while it happens. If it needs more than 16 bits, it's probably still possible using DMA.

Quote
o_lampe
Another feature would be to change the print speed itself according to track length. ( something a very low acceleration would do automatically )
I hate it, when my printer starts to shake, because the print speed is high when the hotend wiggles in tiny moves along the part.
When it prints an arc ( read: lots of short straight tracks ), it should print fast, but at sharp direction changes like 45° zigzag and short tracks, it should slow down.

I believe the algorithm used by Marlin (and copied by RepRapFirmware until I fixed it) used to compute cornering speeds is fundamentally flawed. It works OK for junctions with a small change of angle, and at low speeds, but not for sharper corners at higher speed, where it can demand instantaneous velocity changes far in excess of the configured maximum jerk speeds. Try moving the head in a large hexagon pattern at high speed and I think you will see what I mean. Users have commented how much smoother RepRapFirmware is since I made the change.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

I see, I've planted my seed
The rest will be a matter of time

Now I have to build a delta printer to benefit from it or use RepRapFirmware on my Prusa ?!
-Olaf