Layer shift detection?
-
@mrehorstdmd said in Layer shift detection?:
@NitroFreak But it doesn't "recover" from lost steps, etc., merely shuts down the print?
I like the idea of silent operation- I have a sand table that could use some additional quieting down...
I'm looking at the NEMA-17 motor/driver you linked and the torque spec looks really low. 0.48 Nm can't be right for 52W motor, unless that's torque at maximum rpm. How do you select these things for a given application?
I found the manual here. Are there any special tricks required to interface these to a Duet board?
Do you notice any differences between steppers and servos in terms of things like ringing in prints?
It can, it depends how you configure it. If you set it to never put out an alarm, it will not throw out an alarm and will just try to reach the next commanded position.
I know, it´s perfect, it´s even so perfect that i already suggested that servo to you in the thread that you opened about that sand table
The 52W nema 17 servo has a continuous torque of 0.125 Nm right up until 4000 Rpm, the 78W version has 0.185 Nm. The 180W Nema 23 version has 0.57 Nm right up to 3000 Rpm.
Check page 45 and 47 of the manual posted for the "curves"
You shouldn´t forget that steppers have big torque at basically 0 Rpm, but then fall off really quickly.
This is what a nema 17 torque curve looks like for a really beefy 63mm stepper (what e3d calls a super whopper)
For 24V, you can see that at around ~800 RPM the torque already falls below the value of the 78w nema 17 servo (0.185Nm)
What this means is that you can have a onstant acceleration at any speed without worry of the motors running out of torque. Basically, if your acceleration is fine at 10mm/s, then the same acceleration will be fine at 400mm/s.
The same acceleration at 400mm/s with steppers would result in lost steps, even though at 10mm/s it was fine. -
@Danal said in Layer shift detection?:
@NitroFreak Parts cost is the only driver of price? Assembly labor? Firmware is not a contributor? Trade show marketing (well... not right now)? Support? A dozen things I forgot? Price and cost are only very loosely related, and neither of us really knows how Duet3D chooses to run internally. Therefore, let's turn the question around:
Retailers have posted prices of $230 USD or 250€ to 260€ today for a 6HC, what would you expect to pay for a driverless "base" board?
You´re dragging retailer margins into this.
With your thinking, a thermocouple daughterboard must cost loosely the same as a duet 3 because of firmware, trade show marketing, support and a dozen other things.Let´s look at a real world comparison.
I´m choosing the duex5 expansion board and the expansion breakout board.
https://www.duet3d.com/Duex5
https://www.duet3d.com/DuetAddons/Expansion_Breakout
Both can house a theoretical 5 steppers. I know the Duex has heater mosfets, but let´s just subtract 10 pounds from the price for that.
That makes a price difference (82.50-10)-25 = 47.50 pounds. So roughly 50 pounds for 5 stepper drivers. (10 for one)
This is a real world price difference for the consumer that you can hopefully also comprehend.
So I don´t think its unrealistic expecting to slash 60 pounds off the price of the duet 3 for a version without 6 stepper drivers- especially considering they are the beefier cousins that cost more.
So, Price right now on Duet3d website for the Duet 3 is 185 Pounds. I´d expect a realistic price of around 125 Pounds for a Duet 3 without drivers.
Please let me know how wrong my assumptions and calculations are. -
@NitroFreak said in Layer shift detection?:
You shouldn´t forget that steppers have big torque at basically 0 Rpm, but then fall off really quickly.
This is what a nema 17 torque curve looks like for a really beefy 63mm stepper (what e3d calls a super whopper)That doesn't entirely apply to integrated stepper drivers like TMC used in the Duet. They get very close to constant torque because of what I said previously.
-
@NitroFreak said in Layer shift detection?:
Give me a Duet 3 without any stepper drivers that has the breakout pins right on the board, and I´d be so happy.
It's just a matter of market demand. If there is sufficient demand, Duet will be happy to come with such a product. One thing you can do is committing for large quantities for whatever product you want.
-
@NitroFreak said in Layer shift detection?:
So, Price right now on Duet3d website for the Duet 3 is 185 Pounds. I´d expect a realistic price of around 125 Pounds for a Duet 3 without drivers.
Sounds good to me. And, as I mentioned in a prior post, I'd actually like to see this as well. I support the move of the entire industry to closed loop.
Please let me know how wrong my assumptions and calculations are.
Perhaps my "Are you kidding?" kicked this off on the wrong foot. I apologize.
Seriously, tone is really hard to convey when communications are entirely written, and I used a phrase that might have worked in a face-face conversation, yet that clearly caused miscommunication on a forum.
I apologize.
-
@Danal said in Layer shift detection?:
Seriously, tone is really hard to convey when communications are entirely written, and I used a phrase that might have worked in a face-face conversation, yet that clearly caused miscommunication on a forum.
I apologize.
No hard feelings, i understand
@Edgars-Batna said in Layer shift detection?:
That doesn't entirely apply to integrated stepper drivers. They get very close to constant torque because of what I said previously.
What do you mean? Do you have a source for that claim?
Integrated just means they are soldered on the board instead of pushed in a socket or connected externally.
The torque drop off is largely due to the design of the stepper motor itself.
I´m not going into detail, there´s wikipedia or books for that, but in short Inductance and lack of time for the current to build in the windings is the leading cause for step motors losing torque at high speeds.
You can mitigate that by increasing the voltage which in turn builds current faster. -
@NitroFreak said in Layer shift detection?:
@Danal said in Layer shift detection?:
Seriously, tone is really hard to convey when communications are entirely written, and I used a phrase that might have worked in a face-face conversation, yet that clearly caused miscommunication on a forum.
I apologize.
No hard feelings, i understand
@Edgars-Batna said in Layer shift detection?:
That doesn't entirely apply to integrated stepper drivers. They get very close to constant torque because of what I said previously.
What do you mean? Do you have a source for that claim?
Integrated just means they are soldered on the board instead of pushed in a socket or connected externally.Fixed:
That doesn't entirely apply to integrated stepper drivers like TMC used in the Duet. -
@Edgars-Batna said in Layer shift detection?:
Fixed:
That doesn't entirely apply to integrated stepper drivers like TMC used in the Duet.I just looked through the 2660 datasheet (https://www.trinamic.com/fileadmin/assets/Products/ICs_Documents/TMC2660_datasheet.pdf)
and couldn´t find any reference to a torque curve or generally a flat torque band thanks to the 2660.
Could you elaborate some more? -
@NitroFreak said in Layer shift detection?:
@Edgars-Batna said in Layer shift detection?:
Fixed:
That doesn't entirely apply to integrated stepper drivers like TMC used in the Duet.I just looked through the 2660 datasheet (https://www.trinamic.com/fileadmin/assets/Products/ICs_Documents/TMC2660_datasheet.pdf)
and couldn´t find any reference to a torque curve or generally a flat torque band thanks to the 2660.
Could you elaborate some more?The torque curve is not flat, but it's "good enuf" and absolutely not falling rapidly. Obviously, you're better off with a servo for high speed applications, but within a 3d printer TMC manages really well. That's what I understand from reading the docs and from experience.
-
@NitroFreak Wait just a minute... a "super whopper" stepper has stall torque of 85 oz-in? That's a medium sized NEMA-17 stepper spec, not a "super whopper". I've seen NEMA-17 motors that go up to 110 oz-in.
A 52W servo has torque of 17 oz-in. That's like a pancake stepper. Can you drive a 3D printer mechanism with that?
I'm using 64 oz-in steppers to drive X and Y in my corexy machine, and they don't seem to have any trouble throwing the 1.5 kg moving mass around, but can 17 oz-in manage that? Maintaining torque up to 3000 rpm is nice, but there's no need for 3000 rpm in a printer, unless maybe you gear the motor down to get more torque at the more typical speeds.
What am I missing?
-
@mrehorstdmd I think you and I are on the same page here. I was getting quite excited about these servos until I started looking at the specs and prices. The steppers I use for X and Y are 59N.cm (84 Oz.in) and they throw my 2Kgs plus around with non-print speeds of 350mm/sec by default with no probs at all and I can buy five for 33 GBP (just over £6 each) which is about 1 tenth the cost of a servo it seems. https://www.omc-stepperonline.com/nema-17-stepper-motor/5pcs-of-nema-17-bipolar-59ncm-84oz-in-2a-42x48mm-4-wires-w-1m-cable-and-connector.html
I did go up to slightly bigger Nema 23s for my UV because that weighs in at around 3Kgs but I only run these at 2.4Amps and they are still only about £12 each.
So given my non-print move speed is 350mm/sec and with a 16 tooth 2mm pitch pulley that's about 11 rev/s sec or 660 rpm. If my maths serves me correctly, 3,000 rpm is about 1,600 mm/sec and I ain't never going to use that. So personally I'll stick with the £6.00 stepper (plus say a tenner for the Duet driver) instead of the £70 servo.
-
I'm not really up on my servo motors, but aren't there different types of motor used? As in servo motor refers to a motor with positional awareness through an encoder and PID loop, whereas stepper motors all use the same phased coil setup to produce motion, but there are AC induction servo motors or brushless DC servo motors etc etc.
-
@mrehorstdmd said in Layer shift detection?:
@NitroFreak Wait just a minute... a "super whopper" stepper has stall torque of 85 oz-in? That's a medium sized NEMA-17 stepper spec, not a "super whopper". I've seen NEMA-17 motors that go up to 110 oz-in.
A 52W servo has torque of 17 oz-in. That's like a pancake stepper. Can you drive a 3D printer mechanism with that?
I'm using 64 oz-in steppers to drive X and Y in my corexy machine, and they don't seem to have any trouble throwing the 1.5 kg moving mass around, but can 17 oz-in manage that? Maintaining torque up to 3000 rpm is nice, but there's no need for 3000 rpm in a printer, unless maybe you gear the motor down to get more torque at the more typical speeds.
What am I missing?
I´m not specific about one stepper. I took that datasheet off of Moon´s website. They list it for 1.5A, i think e3d lists their torque at 1.8 or 2A. So there´s the difference.
I think you answered your own question, if the tiniest 52W servo with 0.125 Nm is not enough for you but you dont need 3000 rpm, there´s the answer for you. Also, you can´t quite say that the stepper has 64 "oz-in" because thats at literally 0 RPM. After that it falls off dramatically, just look at the curve.
If the nema 17 servo is not enough, get the nema 23 180w, that one has 0.57 Nm, which is the same as a nema 17 stepper but it goes right up to 3000 rpm.
Everybody is forgetting that for steppers you still need the stepper drivers, which come onboard the servos.
@deckingman sorry but that is no big feat, you use an acceleration of only 1000 mm/s².
If you remember from high school physics, objects in Motion stay in motion unless force is applied.
So just moving the axis at X velocity does not make the motors powerful.
The acceleration is where it´s at. (F = m*a) I want to see you accelerate your gantry at maybe 1g (10.000 mm/s²) up to 350mm/s.This is why servos are great.
My servo has 0.57 Nm torque, with the 20/72 gearing there is 2,05 Nm which means with a tooth radius of 15.35mm a Force of 133 N which accelerates the gantry of 5 kg with 26,63 m/s² (26630 mm/s²) or ca. 2.7g
right up to the top speed of 1389 mm/s at 3000rpm. Yes, all that with my "puny" 0.57NM (80.7 oz in) geared down servo.
That is about 26x the acceleration that you are running. And that is at 100% duty cycle. You can run this motor at 300% for 10% duty cycle, amounting to 79.8 m/s² for a short period (yes, 8g!) I´ve tried it once at 5g and i literally had to tie down the printer to the wall to stop it from shaking itself to death.
I use two e3d super whoppers on my X axis and they can´t keep up.
The added cost is very much worth it for me, my next printer will definitely have servos in all axes.
There´s no way i´m going back. It´s like saying I´m trading my 1000hp F1 engine for a 150hp diesel because it has more torque at 900rpm. It just doesn´t make sense.Honestly, you guys can do whatever you want- I´m not pressuring you into doing anything.
If the other 8 advantages that i listed dont offer any advantage for you, then i can´t help you.
There´s people that complain about skipped steps, loud noise, VFA´s which all go away by using a servo and you complain that you have to gear it differently? You want to tell me that you´d rather have skipped steps, loud noise, VFA´s and more power consumption than just gear differently one time when building the printer? -
A reminder that the actual torque a stepper motor would provide for a 3D printer is Torque * .71 * .098, with 71% coming from the torque drop when using microstepping and 9.8% coming from reducing motion lag to be less than 1/16th a microstep during acceleration. Add in another loss from multiplying by 85% if you aren't running your stepper motors at full current.
The NEMA23s I have provide .74Ncm at 2.8amps, but including the losses, .74Ncm * .85 * .71 * .098 = 4.4Ncm (6.23 oz*in). I then gear that down so I can get ~4000m/s^2 acceleration on my IDEX gantry (depending on how I calculate the mass
), with the gearing providing ~8.3N in the direction of travel. -
@NitroFreak Whoa! No one is saying you're putting any pressure on anyone. I'm just trying to understand. I guess the numbers make sense, just hard to believe. I ordered a couple of the 0.185 N-m NEMA-17 motors to try out. By my calculations, even without gearing it down, in my printer it should be able to accelerate the X axis at 1.7 G and hit a top speed of 2000 mm/sec. All this ignores friction, etc, but it seems there's plenty of margin to allow for that.
Is there anything special that you have to do to drive it from the Duet board (other than grab the step/dir/enable signals)? How does the servo's internal acceleration/deceleration interact with the acceleration in the Duet?
Now all I have to do is wait a month for delivery if the virus situation and trade wars don't screw that up.
-
@mrehorstdmd
Sorry that was mostly directed at @deckingman as it seemed he wanted to convey the message that steppers are superior.If you have any chance to fit the nema 23 version , preferably the 180w get that one because it is only marginally more expensive and you´re going to have all the power you´ll ever need.
You need the duet breakout board to drive the servo input at 5v, as 3.3v is not sufficient.
There´s minimum step and setup timings that you need to account for, they are all written in the documentation and you can put them in the duet using this guideYou need to set the step timing, for the JMC it´s usually 2.5 microseconds, but i recommend going a bit higher, like 5 microseconds.
You also need a usb to serial adapter for programming the servos. You can use them as-is but they will be very soft and not hold any tolerances.
I recommend one with a genuine FTDI chip as these are the best. I use this oneBut this one could work, i don´t know if it´s a genuine FTDI chip
If you have any questions, you can ask me. I´m really happy when more people adapt servos, it drives the community as a whole forward i think.
-
@NitroFreak Thanks for the info. I have a month to wait and plan, get a breakout board, and make cables to connect them to the Duet. I may be in touch again once the motors arrive. I'll be trying them in my printer and in the sand table. If I like how they work I'll order another set for whichever machine doesn't get the first set of motors.
-
@NitroFreak said in Layer shift detection?:
@mrehorstdmd
Sorry that was mostly directed at @deckingman as it seemed he wanted to convey the message that steppers are superior.That is absolutely not true at all. The message was that the steppers I use are more than adequate for my application and they cost less than 1 tenth of a servo motor. A servo motor might well be superior but I can't make use of it's potential on my 3D printer. The reason I use 1000 mm/sec^2 acceleration is that the total moving mass is around 5Kgs and if I accelerate and decelerate that too fast, it puts more forces into the frames than I am comfortable with.
It's a bit like me buying an expensive Ferrari that's capable of 155 mph and using it exclusively for the school run in heavy traffic. It may be a superior vehicle, but what's the point of spending big bucks for something that you can't use to its potential?Edit. I've just compared prices from my normal supplier. The steppers I use cost around £7.79 bought individually with no bulk discounts applied. Ignoring specs for now, the very cheapest servo motor they supply costs £75.39. I use 13 motors on my printer. Total cost for steppers is therefore around £101.27. Total costs for servos would be £980.27. An extra £880 for superior performance but which I can't make use of for the reasons given above.
-
Marco Reps has an interesting video about closed loop motors. Very interesting.
https://www.youtube.com/watch?v=p4ltHDpxrbI
Apparently some TMC controllers do support encoders.
