Upgrading Mini Kossel to Duet electronics

I've written a couple of blog entries on how I upgraded my Mini Kossel to Duet electronics. Part 1 (Hardware) is complete and at [miscsolutions.wordpress.com]. Part 2 (Software and Commissioning) is part-finished and at [miscsolutions.wordpress.com].



The main problem was that the Duet is too large to fit under the bed where the Arduino/RAMPS was. Increasing the lengths of all the top and bottom extrusions by 15 or 20mm would solve that, and give a little more print area. But I mounted the Duet on the Z tower instead.

Edited 2 time(s). Last edit at 01/05/2015 11:11AM 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].

Nice work Dave, I wasn't sure how close you were to getting the Duet firmware working with Deltas, so I went for a Smoothieboard. Apart from that, my Duet has a problem with the USB connection and I am waiting to here back from Reprap support.( I have a feeling it's going to have to go back or be replaced) How is the printer working compared to the original Mega/Ramps?

Regards,

Les

---

Pointy's Things
Pointy's Blog

Hi Les, the Smoothie looks like a nice board and people seem to like it. The only problem I can see is that the web interface is rather primitive as far as I can tell from the documentation. But I expect someone will write a better one.

How much did you pay for your Smoothieboard? The European supplier appears to charge a fair bit more (Eur110+VAT for the 4XC) than RepRapPro charges for the Duet (GBP69+VAT).

Regarding your USB connector, if you have an early Duet then it may be that not all of the 4 mounting lugs on the USB connector were soldered to the board. See the image in my blog entry. If some were not soldered, then I think that is good grounds for a warranty claim.

I've just sorted out the correct M305 commands for the thermistors used in the Mini Kossel, and I am about to try motor movement and then homing.

---

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
Hi Les, the Smoothie looks like a nice board and people seem to like it. The only problem I can see is that the web interface is rather primitive as far as I can tell from the documentation. But I expect someone will write a better one.

How much did you pay for your Smoothieboard? The European supplier appears to charge a fair bit more (Eur110+VAT for the 4XC) than RepRapPro charges for the Duet (GBP69+VAT).

Regarding your USB connector, if you have an early Duet then it may be that not all of the 4 mounting lugs on the USB connector were soldered to the board. See the image in my blog entry. If some were not soldered, then I think that is good grounds for a warranty claim.

I've just sorted out the correct M305 commands for the thermistors used in the Mini Kossel, and I am about to try motor movement and then homing.

Dave

You are correct the Smoothie is more expensive than the DUET however it is very nicely designed in the driver dept Like the duet setting the current is as simple as telling the config file what current you want but there are also breakout points for all the stepper drivers to allow external add on to be used (Very usefull if you happen to fry one of the onboard ones).

They have lots of capabilities built in to them as well

The reason I bought one was to use on my CNC Router but now thinking it may go on a delta also the fact that they have mature firmware for delta and CoreXY Already which the DUET does not (I know your delta fork is coming along nicely but what about coreXY Kinematics?)

You never know once you get it to a release state I may well give a duet a go on the DELTA I am planning on should be nice for a single extruder machine (My CoreXY will have a kraken on it so the duet is not really an option for that at the moment so I will be using a RADDS Shield with extension on a due board)

Have you had any thoughts about a mixing extruder such as the new E3D one that they have announced?

Doug

Like the printer and what you have done with it mind can't wait to see what sort of performance you get from it once it is tuned?

Quote
dc42
Hi Les, the Smoothie looks like a nice board and people seem to like it. The only problem I can see is that the web interface is rather primitive as far as I can tell from the documentation. But I expect someone will write a better one.

How much did you pay for your Smoothieboard? The European supplier appears to charge a fair bit more (Eur110+VAT for the 4XC) than RepRapPro charges for the Duet (GBP69+VAT)..

Yes on the face of it, it's more expensive, I paid £114.49 for the X5, including shipping. That gives me the options of duel extruders, for the Duet I would need the Duex4 as well, which is £120.

Quote
dc42
Regarding your USB connector, if you have an early Duet then it may be that not all of the 4 mounting lugs on the USB connector were soldered to the board. See the image in my blog entry. If some were not soldered, then I think that is good grounds for a warranty claim.

I had to solder the big lugs as they were loose, but the problem still persists. The soldering on the 4 data/powers pins looks poor, I have sent Ian a picture but I expect he is inundated after the Xmas break.

Quote
dc42
I've just sorted out the correct M305 commands for the thermistors used in the Mini Kossel, and I am about to try motor movement and then homing.

I still might end up using the Duet in the Delta and then use the Smoothieboard in Ormerod when I completely rip it apart and rebuild it, so I shall follow your progress with interest.

Regards,

Les

---

Pointy's Things
Pointy's Blog

Hi Doug, implementing CoreXY kinematics would be a piece of cake compared to Delta. The Duet+DueX4 can handle 5 extruders and hot ends, so it could handle a Kraken. I don't know how the cost would compare with the RADDS+extension+Arduino Due (btw does that give you Ethernet and control over the stepper motor currents?

Duet firmware already supports mixing extruders.

It occurs to me that Smoothieware could probably be ported to the Duet without difficulty, and vice versa.

---

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].

David, this is great news, i was folowing you eagerly, i'm going to try to convert my delta tomorow in the evening. Hopefully i can comlete it soon.
Thank you for your great work, this is a great milestone for DUET electronics.

Pavel

Quote
dc42
Hi Doug, implementing CoreXY kinematics would be a piece of cake compared to Delta. The Duet+DueX4 can handle 5 extruders and hot ends, so it could handle a Kraken. I don't know how the cost would compare with the RADDS+extension+Arduino Due (btw does that give you Ethernet and control over the stepper motor currents?

Duet firmware already supports mixing extruders.

It occurs to me that Smoothieware could probably be ported to the Duet without difficulty, and vice versa.

Dave the radds cost me 100 euros for the shield and 3 selencios driver modules (128 microstepping) and the extension board which gives 8 steppers and a host of mosfet outputs plus about £16 for a due so very comparable in terms of cost and the radds is driver modules rather than hard mounted on the board (This is the thing that worries me with the duet system pop a driver and buy a new board).

My soldering skills are not what they were 25 years ago lol

Doug

Quote
paveldvor
David, this is great news, i was folowing you eagerly, i'm going to try to convert my delta tomorow in the evening. Hopefully i can comlete it soon.
Thank you for your great work, this is a great milestone for DUET electronics.

Pavel

Thanks, Pavel! It's nice to know that someone else will be using this. I've checked in some more work into my delta branch so that the binary supports everything I've written about in my blog entry.

I'm not sure that mounting the Duet end-on was a good idea. I might re-mount it horizontally, so that the USB and Ethernet ports are not at the top. I'll add mounting holes for this option when I revise the enclosure model. The model also needs changing to add a cutout for the hot end terminal block, and to enlarge the cutout for the USB connector.

The heaters and thermistors are all working. The heated bed tends to overshoot the set temperature by several degrees. You might want to try enabling PID for the bed using the M304 command. I'll be too busy to try that for a while.

Individual motor moves are working, so is homing. I still have some more work to do on the firmware, so don't try any "normal" moves yet. This should be working within a few days; then there will just be the Z probe and bed levelling code to modify. I'll probably implement a N x N bed probe option that can be used on the Ormerod too. I'm also considering and auto-calibration function for delta printers, because it doesn't look difficult and could save a lot of time.

---

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

Quote
paveldvor
David, this is great news, i was folowing you eagerly, i'm going to try to convert my delta tomorow in the evening. Hopefully i can comlete it soon.
Thank you for your great work, this is a great milestone for DUET electronics.

Pavel

Thanks, Pavel! It's nice to know that someone else will be using this. I've checked in some more work into my delta branch so that the binary supports everything I've written about in my blog entry.

I'm not sure that mounting the Duet end-on was a good idea. I might re-mount it horizontally, so that the USB and Ethernet ports are not at the top. I'll add mounting holes for this option when I revise the enclosure model. The model also needs changing to add a cutout for the hot end terminal block, and to enlarge the cutout for the USB connector.

The heaters and thermistors are all working. The heated bed tends to overshoot the set temperature by several degrees. You might want to try enabling PID for the bed using the M304 command. I'll be too busy to try that for a while.

Individual motor moves are working, so is homing. I still have some more work to do on the firmware, so don't try any "normal" moves yet. This should be working within a few days; then there will just be the Z probe and bed levelling code to modify. I'll probably implement a N x N bed probe option that can be used on the Ormerod too. I'm also considering and auto-calibration function for delta printers, because it doesn't look difficult and could save a lot of time.

Dave

This is an excellent piece of work and think it will definitely be the way I go on my Delta build which will be starting shortly.

Looking forward to the finished code

Doug

I made major progress today. I had been stuck on the maths for the inverse delta transform for a while (i.e. converting tower motor coordinates to Cartesian coordinates), but I now have that working. I could have skipped doing that, because the inverse transformation is only needed for establishing the XYZ position after aborted and direct motor moves, but that seemed like cheating. Anyway, it is all working now. After that, getting the head to move from one Cartesian position to another was trivial.

At this point I could split long linear moves into lots of short ones to approximate movement a straight line, as Marlin and Smoothieware do. But as I have already worked out the maths to generate the steps for a precise linear movement, and implemented the fast 64-bit integer square root needed to do that in real time, I'll proceed with generating precise linear movements.

I also hooked up one of my prototype touch screen controllers to the Mini Kossel.



The firmware for it could do with some adaptation when it detects that it is talking to a Delta printer (e.g. to have only a single Home button), but that can come later.

Edited 1 time(s). Last edit at 01/11/2015 12:39PM 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].

Dave

A quick Question can the DUET cope with Thermocouples instead of Thermistor's

I have a silicon heat pad that came with a thermocouple built in and would prefer to use that if possible though it probably isn't to difficult to just fit a Thermistor between the pad and the bed if I need to.

Just ordered my Duet so should hopefully get my large Delta up and running within a couple of weeks or so Will want one of your display modules as well if you have any left can you put my name on one please.

Doug

To use a thermocouple, you need to use a thermocouple amplifier chip. Like all the other printer electronics I am aware of, the Duet does not have one on board. The easiest solution (other than using a thermistor instead) would be to use one of the available thermocouple amplifier breakout boards. These come in two varieties: digital output (typically based on the MAX31855 chip, e.g. from Sparkfun), and analog output (typically based on the AD8495 chip, e.g. from Adafruit). Although the digital one would be more accurate, the analog one would be easiest to interface because the output could probably drive the Duet thermistor input directly (there is a chance that the 1K or 4.7K thermistor bias resistor on the Duet would upset the AD8495 output, but I doubt it).

Edited 1 time(s). Last edit at 01/11/2015 12:38PM 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
To use a thermocouple, you need to use a thermocouple amplifier chip. Like all the other printer electronics I am aware of, the Duet does not have one on board. The easiest solution (other than using a thermistor instead) would be to use one of the available thermocouple amplifier breakout boards. These come in two varieties: digital output (typically based on the MAX31855 chip, e.g. from Sparkfun), and analog output (typically based on the AD8495 chip, e.g. from Adafruit). Although the digital one would be more accurate, the analog one would be easiest to interface because the output could probably drive the Duet thermistor input directly (there is a chance that the 1K or 4.7K thermistor bias resistor on the Duet would upset the AD8495 output, but I doubt it).

Dave

Just a quicky to let you know that My Duet is in the post as we speak and I will probably place a pre-order with Tony for one of your Display boards as well (just seems easier than sourcing everything separately and at least will be sure it all works together)

Do you have any of your single head IR bed sensors that may be able to be fitted to a delta (I am not adverse to making a mounting for it just aware that auto cal would be easier with one and I know I would have to make a loom for it though all I would actually need is the Differential sensor info feeding back can do the Thermistors and fan separately)

I have a few single-nozzle sensors left, however I think they are too tall to fit on a Mini Kossel. Also the thermostatic control of the fan isn't really needed, because the fan on the MK is so quiet. However, once I have got the delta firmware fully working and sorted out a few other things, I intend to look into a couple of options for a better Z probe. One option would be a smaller version of the differential IR board, probably without the thermostatic fan control. Another would be a solenoid operated probe, which I considered for the Ormerod but didn't really fit (but it would fit on the MK).

Some people use FSRs under the bed mounts to detect contact from the print head. This appears to be a nice solution, however I have seen it said that this is not compatible with the heated bed option. Presumably this is because the FSRs need to be kept cool and/or because of the extra weight of the heated bed.

---

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
I have a few single-nozzle sensors left, however I think they are too tall to fit on a Mini Kossel. Also the thermostatic control of the fan isn't really needed, because the fan on the MK is so quiet. However, once I have got the delta firmware fully working and sorted out a few other things, I intend to look into a couple of options for a better Z probe. One option would be a smaller version of the differential IR board, probably without the thermostatic fan control. Another would be a solenoid operated probe, which I considered for the Ormerod but didn't really fit (but it would fit on the MK).

Some people use FSRs under the bed mounts to detect contact from the print head. This appears to be a nice solution, however I have seen it said that this is not compatible with the heated bed option. Presumably this is because the FSRs need to be kept cool and/or because of the extra weight of the heated bed.

Ok well I wont be building a MK but a Cherry Pi4 which is larger but in essence is a larger version of the Cherry Pi3s but with a fairly solid effector/nozzle arrangement and the nozzle will only protrude under the effector by approx. 15 mm so something else will be needed to mount one anyway but if you are thinking of doing a simple one that would be Ideal.

I am going to be using a E3D V6 and its fan is very quiet as well and I haven't yet needed a print cooling fan and that can be controlled from the duet anyway can't it?

Doug

You could use my Ormerod differential LED sensor board if it will fit between a pair of parallel rod mounts on your effector. I am away from the office at present so I can't tell you the exact width, but it is about 25mm.

EDIT: I just found that it is 28mm wide.

Edited 1 time(s). Last edit at 01/13/2015 02:27PM 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
dougal1957
I am going to be using a E3D V6 and its fan is very quiet as well and I haven't yet needed a print cooling fan and that can be controlled from the duet anyway can't it?

Doug

Just connect it to Fan0 on the Duet and then it can be controlled by M106 & M107.

Regards,

Les

---

Pointy's Things
Pointy's Blog

Yes that's what I thought Les and Dave there is 30mm between the rods on the CP3/4 so that should be ok just need to print up some sort of mount to go on the effector that won't interfere with the tensioners on each arm pair.

Dave Mail on the way re the single extruder differential sensor.

Hi, i tried to mount the DC42 differential sensor board on my kossel design, and i created effector and fan duct to hold it, bud i havent finish building it yet - and the firmware is not yet ready too.
It fits, but only when insulation cover is not pressent, with insulation cover the rods touch the insulation when in the outside perimeter of print bed. I have about 50mm separation (40mm on effector and the rest on ball-links), im considering increasing the separation or insulate the board with kapton tape only - which i'll try first.
Some pictures attached, i can upload the scad files if anybody would by interested.

Hi David,
today i installed your newest firmware 0.80j and i'm printing my first calibration ortogonal test piece on my Duet delta.
Thank you very much for your hard work.
It is printing nicely and i love it.
One thing i noticed is that you have to have some kind of hardcoded max continuos move of any carriage, my delta has 347mm homed height and if i try to home from Z0 it stops at around 270mm and i have to press the home button again.
The same occurs when homed and I send a G1 Z0 command.
One more question, i have a working differential Z height sensor from you mounted on my kossel, is it already possible to use it for Z homing? If so, how should i enable it in the config?
I tried some experiments my self, but i could not make it work.

Pavel

Hi Pavel, you've beaten me to doing an actual print! Last night I started carrying out and documenting a manual calibration procedure. While doing this I found and fixed the issue with large Z moves that you mention.

I will publish a firmware update and Part 3 of my blog series later today. I haven't implemented support for the Z probe yet. I have almost run out of differential IR sensor boards, so I am considering designing a new version for the Mini Kossel and other deltas. It will probably be just a sensor, with no illumination, fan control or hot end indication and a 3 pin Molex connector. It will be short enough to fit below the effector of the MK. It would be compatible with the sensor on the board you already have.

Edited 1 time(s). Last edit at 01/22/2015 01:46AM 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].

The delta support in my fork of RepRapFirmware is now, I believe fully working as of version 0.80k. As promised, the delta movement is implemented by calculating the time that each step is due, not by chopping up long linear segment into short segments and assuming that a short straight line segment in Cartesian space can be approximated by a linear movement of the tower motors. I've limited the speed in the config file to 250mm/sec for now, until I take some measurements of the calculation time.

The binary is here: [github.com]. I have also made some changes to the config.g file at [github.com].

I've added a Part 3 to my blog series on converting the Mini Kossel to Duet electronics, to cover calibration. It's available at [miscsolutions.wordpress.com].

Next on my list for this firmware is re-testing it on Cartesian printers, then I will move the current code to my main dev branch. After that is some more work to support additional Panel Due features (e.g. print time estimation). So it may be a couple of weeks or so before I start work on supporting a Z probe (probably an IR one) and implementing auto calibration.

---

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].

David, using the 80k firmware the problem with the large Z move is gone when moving with G1 Z10 F10000 or G1 Z340 F10000, but when homing it still stops at around Z264.
Also both prints i made today stopped working, it behaved as if missed steps on one tower (it started to print in air a little off (cca 5mm) from the layer below), but i din't hear any missed steps.
I didn't havy any issues like this with the 80j firmware. I'll probably try to reflash to 80j tomorow to test if it is consistent.
Please let me know if you happen to have the same issue.

Paveldvor, there are only a few changes between 0.80j and 0.80k. One is the bug fix for large Z moves. The second is to limit the range of Z moves to between -0.2mm and the homed height + 10mm. The third is to print an extra decimal place when the M665 and M666 commands are used to display the existing values.

If you think the problem you observed is due to a firmware bug and it is repeatable, then please do the following:

1. Connect Pronterface via USB. Send command M111 S1 P4 and start the print. Capture any error messages that it displays and post them.

2. Identify the location in the gcode files where it happens, produce a smaller sample gcode file that demonstrates the problem if you can, and send it to me.

Otherwise, if your printer is missing steps, it may be that the accelerations or maximum jerk rates are set too high in config.g. I had the values from the Mini Kossel/Marlin firmware in there for a while, but they were way higher than the Ormerod values and the printer sounded really bad. So I reduced them in the config.g file that I checked in with version 0.80k. Or it may be that you need to increase the motor currents to better suit your motors.

If your printer has a Z range greater than about 230mm then you will need to edit homedelta.g. Change the three instances of 320 to a larger value, around 90mm more than your Z range. This figure defines the maximum distance that a carriage will travel up when homing.

Edited 3 time(s). Last edit at 01/23/2015 07:01AM 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].

Great to see you have got this working Dave.

I have my Delta pretty much built and just have to do the electronics. While I want to use the Duet in it, I think for now I will start with the Smoothieboard and switch over in the near future. I am going to need the Ormerod to print a few parts for one thing, such as mounts for the electronics. I also thought it might be a good idea to get it up and running with a known working firmware, which will hopefully help eliminate any build and setup errors when I do the switch.

Keep up the good work.

Les

---

Pointy's Things
Pointy's Blog

David i thing you are right - i changed the accelerations and maximum instant speed a bit more than you, and did not increase the currents.
The homing is of cource just as you said - i didnt realize it is this dependant - i have just the 250mm up and my Z height is aprox 350mm.
I'm not at home yet and i will test it in the evening ang post results - I 'll try to lessen the accel and max speed to your values and than after more testing try to increase it.

btw the step time calculation currently considers there has been an error if the calculated interval between steps is less than about 15us. If this happens, it will abort that motor move, and print a diagnostic message a few moves later to the USB port if you have sent M111 S1 P4. The purpose of this is to catch calculation errors. However, I guess that if you use high maximum feed rates and accelerations, you might be able to generate this situation anyway.

EDIT: I just managed to provoke this software trap on my Mini Kossel by increasing the XY accelerations to 5000 and generating moves at 500mm/sec. I'll check that the calculation is correct, and reduce that 15us figure appropriately if it is.

Edited 2 time(s). Last edit at 01/23/2015 07:38AM 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].

I found the cause of the problem with high speed/high acceleration moves. Under these conditions, rounding errors were causing the calculated step time to be out by up to 9us when the movement transitioned from the steady speed phase to the deceleration phase. As the step interval could be as low as 20us when the feed rate was 500mm/sec, this sometimes brought the step interval below the error threshold.

I've improved the accuracy of the algorithm to reduce the maximum error to around 3us, and I now have moves at 500mm/sec speed and 5000mm/sec^2 acceleration working on my Mini Kossel. I've also reduced the minimum step interval to 10us. I'm doing some test prints and looking for other possible similar problems before I release the changes.

---

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've released firmware version 0.80m. This corrects the rounding issue I found with high speed/high acceleration moves, and a couple of other issues too.

---

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].