Rostock Max v2 HE280/Accelerometer Upgrade - Summary

[bruceg]

Just got around to installing an HE280 (+ wiring whip) on a v2 and thought i would summarize in one place what i gleaned from the various forum posts.

tl;dr - smooth sailing following the guides, calibration couldn't be simpler with the wizard, upgrade well worth the effort.

Thanks as always to geneb, mhackney and all on the forum and particularly in this case JJ Johnson for the HE280 guides and Dave Crocker for the Calibration Wizard.

Machine is a late 2014 v2 ATX power supply upgraded with molded cheapskates, ball cup arms and a variety of home brew extruders. Build platform is a much abused 2mm PEI sheet bonded to the standard glass plate. Was running Repetier 091.

Applications/Firmware Used:
Arduino IDE 1.6.0
Repetier 0.92.2 (2016 12 27) - configuration.h has been spruced up and the definitions are now nicely grouped
Matter Control 1.6.1
Delta Calibration Wizard

Guides/Documents Used:
HE280 Assembly - Starting Step 10
Rostock Max v1/v2 HE280 Hotend Upgrade
Installing the Firmware
Hot End Whip Wiring - Sheet1(2)
Hotend Accelerometer Probe PCB Board Rev 5c PCB layout image marked up with Pin Function

Settings and Changes in Configuration.h: (line numbers for 0.92.2 2016 12 27)

Code: Select all

10   #define PRINTER 2 // used v2 as starting point as v3 contains extra line defining a pin for the fan board
19   #define POWER_SUPPLY 1 // old school ATX supply 
59   #define MIN_DEFECT_TEMPERATURE 15  // chilly basement

192  #define MAXTEMP 290 // increase from 245 to match v3 setting
194  #define INVERT_Y_DIR 0 // required if POWER_SUPPLY 1
196  #define DELTA_DIAGONAL_ROD 291.06  // ball cup arms
199  #define Z_MAX_LENGTH 350 // starting point for calibration note that v3 distance of 395 will be problematic 
201  #define CARRIAGE_HORIZONTAL_OFFSET 26.5  // molded cheapskates
227  #define Z_PROBE_SPEED 60 // note that v3 speed is 90 -  not sure of significance

805  #define UI_ENCODER_DIR 1 // 0 normal 1 inverts encoder direction i.e. 1 is opposite to 091 direction
// also note encoder has been effectively "detuned" a bit in this section - a little better for the old and slow in my view...
834  #define UI_SET_MIN_HEATED_BED_TEMP  25 // for flying close to the sun with Ninjaflex and flexible PLA on PEI
837  #define UI_SET_MAX_EXTRUDER_TEMP   265 //  to allow higher temp adjustment from LCD UI

Delta Configuration Wizard was simple to use (note that the "PROBE-ZOFFSET:-x.xx" number appears a couple of lines or so above the pause, but you can always scroll back). Started out with cold bed and hotend, then with bed heat on and nozzle hot but turned off. No problems. Then tried with the hot end heat on. As advertised this was problematic with a failure the first go but success on the second. If this is an electrical noise problem (presumably associated with the PWM) it would be interesting to see if running the I2C and interrupt lines in shielded cables would help.

With one round of hot calibration, first layer thickness variation over an ~200mm track across the bed was +/- 0.06mm vs > +/- 0.15mm with my best efforts at manual calibration on my lumpy bed. Bottom line, a problematic model that had been giving me fits with bed adhesion printed on the first go. Now on to the Prometheus hot ends...

[img]http://forum.seemecnc.com/download/file ... w&id=14232[/img]

[cyber.shifter]

Maybe running the I2C and interrupt lines through a ferro core would help eliminate the noise? Just a thought. They're fairly cheap and can usually be snapped around the cable. It would be an easier thing to try first before having to switch out cables. Awesome post though! I'm thinking about upgrading my hotend, I have the same model Rmax. Did you just run a new set of wires through one of the pillars?

[bruceg]

cyber.shifter,
Indeed, ferrite cores are suggested as a solution. Perhaps you have already spied this report. ETA However are not applicable to this problem - see mhackney below.

The wiring interface for the HE280/Accelerometer board is accomplished by sniping off the old hot end wiring around the level of the extruder and then making splice connections (which can then live in the top section) and swapping around a few connections on the RAMBo board. So NO new wires need to be pulled (provided you are ok with the hot end fan running all the time). You can get a pre-made wiring whip that you splice into your existing wiring which will give you both the connector for the accelerometer board and the two connectors used at the RAMBo. To make your own bespoke harness, get a plug for the accelerometer board (SeeMeCNC) or (Digikey) and plugs for the RAMBo, one each of 4pos and 2pos and minimum three (for those who crimp perfectly every time) of crimp Sockets . The advantage of the SeeMeCNC whip is that the wire colours will match the guides and documentation.

My dim thinking on the shielding was based on the following:

1) The wiring In my machine has two pairs (Red/Black) of 26ga wires running in the Y-tower extrusion that previously connected the HotEnd/PEEK fan (Heat 1 +/-) and Part/Layer Fan (Fan 0 +/-) and 2 pairs of 18ga wires running in the Z-tower; Red/Black connecting the HotEnd Heater (Heat 0 +/-) and Green/White connecting the HotEnd thermistor (Thermistor 0).

2) The new wiring results in the following connections (listed as old function/RAMBo label to new function/accelerometer board pin connector number) Note: please don't trust me - follow the guides - they will lead you through with no thought required!!)

Y-Tower Wires:
The old HotEnd fan wires become the I2C pair (Heat 1+[Red26ga] > SDA Pin6[Red26ga], Heat 1-[Blackk26ga] > SCL Pin5[Blackk26ga].
The old Part Fan wires change as follows (Fan 0+[Red26ga] > INT Pin7[Blue26ga], Fan 0-[Black26ga] > Part Fan Switched Ground -Pin1[Orange26ga].

Z-Tower Wires:
The old HotEnd heater wires change as follows (Heat 0+[Red18ga] > 12V Constant Supply Pin2[Red18ga], Heat 0-[Blackk18ga] > HotEnd Switched Ground Pin3[Blackk18ga].
The old HotEnd thermistor wires change as follows (Thermistor 0"lower"[Green18ga] > Thermistor Signal "+" Pin8[Green26ga], Thermistor 0"upper"[White18ga] > Accelerometer Board Common Ground -Pin4[White26ga].

3) Under the presumption that the HotEnd Switched Ground is the trouble maker via noise induction in the wiring (vs on the board) when calibrating with the HotEnd on (and the part fan is not running); i thought it might help to run the I2C as a separately shielded pair and the interrupt (probe output trigger) in separate shield in the whip section only as the HotEnd switched ground and the communication signals run in separate towers. This would still leave the problem of making a connection for the shield drains to a good earth ground. One could consider making a connection via grounding one of the towers and connecting the drain in the top, swallowing one's pride and running an external wire or more properly, pulling one drain wire through a tower.

You are most likely right however, that the most practical approach would be to make the "hard" decision to sacrifice one of the half dozen or so VGA cables i have been dragging around for the last 30 years or so and apply the ferrite beads. The simplest option of course is to not worry about the effect of nozzle expansion on Z and calibrate with the bed at temperature and the hotend off (+/- preheated) per mhackney - i will try and link a couple of threads on this topic once i get the driveway plowed.

For anyone applying an accelerometer board to a v2 i recommend looking at the whip wiring and the accelerometer board image (along with the guides) particularly if you plan on making your own harness.

[mhackney]

STOP PROPOGATTING MYTHS please! Ferrite beads have NO effect on the probing problems that were experienced. I and others have done extensive tests/experiments with oscilloscopes, etc and the ferrite beads do not magically fix anything. The data that is transmitted with or without beads is identical. That said, it is CRITICAL that you do not probe with the hot end being actively heated. That will created noise on the I2C and cause problems/crashes.

Shielding is a worthwhile idea. Trash80 and I were literally just talking about that yesterday and I believe he is planning some experiments with shielded cabling.

[bruceg]

mhackney,
I stand corrected, thank you sir! My apologies for not researching before responding. I will edit my previous response to reflect the data.

[mhackney]

No worries bruceg, this happens all the time and is one of the (minor) issues with forums like this that get a lot of readership. There are so many myths and misinformation on 3D printing as it is. Unfortunately (and I'm not referring to you or the gent who originally posted the ferrite beads) there are too many self-proclaimed experts. That and folks do one test or get one good result without measuring, conducting control tests or even formulating a hypothesis to test and then claim victory on one result with no real understanding. That's how I was trained (as a PhD scientist) and that's how I work. I'm a badger when it comes to understanding a problem and deriving a solution. I'm spending an insane amount of time understanding exactly how accelerometer probing works, profiling how it responds to printer configuration changes and accelerometer configuration changes, probe surface and a lot of other factors - all with several hypotheses I'm testing to either validate or invalidate. For instance, here is detailed acceleration (g force) data for the HE280 on a Rostock MAX V3. Each run is 13 probe points. There are 32 bytes of acceleration data in the buffer when the accelerometer triggers. That's what you see in each row (one row of 32 for each of 13 points). Note that the 12th data point (typically 4.06g) is the max event that likely triggered the probe. The 3 runs show the run-to-run consistency, which currently is Ok and not as good as FSR or IR probes. This is the best I've been able to get out of days of testing, tweaking, etc. Note that I can decrease the actual mean and deviation of calibration results with a couple of simple probing configuration changes but I never get better than about .6. Also note, simply decreasing the probing radius or the number of probing points almost always gives a much better deviation. Simply by decreasing the probe radius by 10mm and only probing 10 points (6 outer, 3 inner and center) I can get the deviation down to around .35. Does that mean the printer is calibrated better? Think about it and lets discuss your answers!

Code: Select all

Run 1
0.64	0.64	0.67	0.70	0.77	0.86	0.96	1.06	1.15	1.22	1.92	4.06	1.34	0.80	2.37	1.44	0.32	1.22	1.63	0.86	0.96	1.47	1.34	1.12	1.25	1.18	1.15	1.22	0.83	0.80	1.06	1.22	
1.25	1.22	1.18	1.09	1.02	0.96	0.93	0.93	0.93	0.93	1.70	4.06	1.50	0.93	2.24	1.66	0.45	1.82	2.11	1.41	1.38	1.28	0.93	0.54	0.58	1.34	1.15	1.02	0.70	0.80	1.22	1.47	
1.25	1.12	1.02	0.99	0.83	0.70	0.70	0.70	0.74	0.80	2.46	4.06	0.90	1.34	2.40	1.15	0.42	1.82	2.02	1.25	0.80	1.15	0.93	0.74	1.54	1.76	1.41	1.09	0.58	0.67	1.06	1.38	
0.83	0.86	0.90	0.96	1.06	1.12	1.25	1.28	1.34	1.41	2.75	4.06	1.02	0.83	2.21	0.77	0.10	1.41	1.41	1.06	1.12	1.28	0.99	0.83	0.86	1.44	1.31	0.86	0.83	1.02	1.09	0.96	
0.58	0.61	0.70	0.77	0.96	1.09	1.15	1.18	1.18	1.22	1.57	4.06	1.82	8.13	2.18	1.76	0.03	1.06	1.76	0.90	1.18	1.25	1.25	1.06	0.93	0.86	1.15	1.18	0.96	0.96	1.06	0.93	
1.22	1.18	1.12	1.06	0.90	0.86	0.86	0.86	0.83	0.86	1.60	4.06	1.02	0.48	2.43	1.44	0.54	1.38	1.47	0.96	1.47	1.25	0.96	1.06	0.48	0.90	0.99	0.67	0.77	0.83	1.02	1.18	
0.80	0.67	0.61	0.54	0.58	0.64	0.80	0.93	1.06	1.22	1.44	4.06	2.02	0.90	2.05	1.60	8.03	0.42	1.82	0.96	0.83	1.60	1.70	1.02	0.77	0.93	0.93	0.83	0.93	1.25	1.44	1.12	
1.12	1.18	1.22	1.22	1.22	1.18	1.18	1.09	1.02	1.02	1.57	4.06	1.34	1.02	1.79	1.25	0.16	0.90	1.70	1.15	1.22	1.73	1.50	0.93	0.64	1.02	0.93	0.93	1.12	1.25	1.28	1.12	
0.90	0.74	0.64	0.61	0.54	0.51	0.58	0.64	0.74	0.90	2.75	4.06	0.96	1.89	1.98	0.70	0.10	1.15	1.73	1.02	1.12	2.02	1.47	0.61	0.64	1.25	0.86	1.06	1.12	1.31	1.02	1.09	
1.09	1.06	0.96	0.83	0.70	0.70	0.74	0.80	0.90	0.96	1.50	4.06	1.41	1.15	2.02	1.38	8.03	0.58	1.60	0.67	0.86	1.98	1.66	0.86	0.77	1.09	0.90	0.67	1.22	1.12	1.06	1.06	
0.83	0.90	0.99	1.06	1.18	1.31	1.38	1.38	1.41	1.38	1.60	4.06	1.57	0.67	1.60	0.99	8.16	0.48	1.34	0.74	0.90	1.44	1.60	1.12	0.86	1.15	1.12	0.99	1.06	1.06	1.34	1.12	
0.96	0.99	1.02	1.02	1.06	1.06	1.06	1.06	0.99	0.99	2.08	4.06	0.64	1.22	1.41	0.83	0.00	1.06	1.34	0.77	1.60	2.18	1.60	1.15	0.77	0.90	0.77	0.58	0.70	1.02	1.28	1.09	
0.83	0.86	0.86	0.90	0.99	1.09	1.18	1.25	1.25	1.28	1.60	4.06	2.18	0.74	2.27	1.63	8.00	0.10	1.06	0.96	0.54	1.22	1.92	1.60	0.96	1.25	1.25	0.90	0.90	1.15	1.25	1.02

Run 2
0.74	0.64	0.64	0.64	0.74	0.80	0.93	0.99	1.12	1.22	2.08	4.06	1.18	0.86	2.18	1.25	0.32	1.28	1.57	0.90	0.99	1.54	1.34	1.06	1.09	1.12	1.18	1.18	0.86	0.86	1.28	1.70	
1.25	1.25	1.18	1.12	1.02	0.96	0.93	0.93	0.90	0.96	1.25	4.06	2.14	0.74	2.34	1.86	0.45	1.44	2.24	1.41	1.44	1.25	1.09	0.70	0.51	1.28	1.15	0.99	0.64	0.70	1.15	1.44	
1.25	1.15	1.12	1.02	0.83	0.77	0.70	0.70	0.74	0.77	1.98	4.06	0.99	1.25	2.46	1.34	0.26	1.79	2.18	1.18	0.77	1.15	0.96	0.70	1.09	1.89	1.47	1.02	0.58	0.64	0.99	1.31	
0.86	0.83	0.90	0.96	1.06	1.12	1.22	1.25	1.28	1.31	1.98	4.06	1.44	0.74	2.30	1.31	0.03	1.15	1.60	0.93	1.06	1.25	1.02	0.96	0.80	1.28	1.41	0.86	0.83	0.99	1.12	0.99	
0.67	0.70	0.70	0.77	0.86	0.96	1.02	1.12	1.22	1.28	1.73	4.06	1.82	8.16	2.08	1.63	8.13	1.06	1.79	0.93	1.28	1.31	1.28	1.12	1.06	0.80	1.12	1.12	0.93	1.02	1.12	0.96	
1.25	1.15	1.12	1.06	0.90	0.83	0.86	0.86	0.83	0.86	1.22	4.06	1.28	0.06	2.46	1.57	0.51	1.31	1.63	0.93	1.44	1.22	0.96	1.18	0.45	0.86	0.99	0.61	0.45	1.22	1.22	0.86	
0.80	0.70	0.61	0.54	0.58	0.67	0.77	0.90	1.09	1.22	1.82	4.06	1.31	1.15	1.95	1.25	8.03	0.70	1.82	0.93	0.93	1.73	1.50	0.93	0.74	0.93	0.90	0.83	0.90	1.31	1.44	1.18	
1.12	1.18	1.22	1.22	1.15	1.15	1.09	1.09	1.09	1.09	2.21	3.87	1.02	1.31	1.66	0.93	0.16	1.18	1.60	1.15	1.31	1.82	1.44	0.77	0.77	0.99	0.90	0.96	1.18	1.25	1.28	1.15	
0.83	0.74	0.67	0.61	0.54	0.51	0.64	0.64	0.74	0.90	2.14	4.06	1.02	1.76	2.08	0.99	0.00	0.93	1.82	1.06	0.93	1.95	1.60	0.74	0.83	1.25	0.99	0.99	1.28	1.22	1.06	1.06	
1.09	0.99	0.93	0.80	0.70	0.70	0.77	0.83	0.93	0.99	2.88	3.94	0.77	1.57	2.18	0.74	0.06	1.06	1.22	0.61	1.34	2.08	1.34	0.64	0.70	1.06	0.70	0.83	1.22	1.12	1.09	1.09	
0.86	0.96	1.02	1.09	1.18	1.25	1.31	1.34	1.34	1.34	2.72	3.68	0.67	1.31	1.34	0.45	8.16	1.09	1.06	0.77	1.25	1.79	1.38	0.83	0.90	1.06	1.02	1.02	1.12	1.34	1.34	1.09	
0.96	0.99	1.02	1.12	1.12	1.09	1.09	1.09	0.99	0.93	2.72	3.74	0.42	1.60	1.44	0.67	0.03	1.28	1.06	0.74	1.73	2.24	1.63	1.12	0.93	0.93	0.67	0.54	0.67	1.06	1.22	1.06	
0.83	0.86	0.86	0.96	1.06	1.09	1.18	1.22	1.22	1.25	2.82	3.58	1.47	0.83	2.40	0.64	7.97	0.64	1.06	0.70	0.67	1.57	1.89	1.18	1.06	1.38	1.15	0.90	0.96	1.22	1.18	0.96

Run 3:
0.70	0.64	0.64	0.67	0.70	0.80	0.93	1.09	1.22	1.31	1.76	4.06	1.47	0.70	2.27	1.44	0.29	1.22	1.73	0.93	0.96	1.50	1.38	1.09	1.12	1.18	1.15	1.25	0.83	0.83	1.02	1.15	
1.22	1.25	1.18	1.15	1.06	0.99	0.93	0.90	0.90	0.90	1.18	4.06	2.27	0.74	2.30	1.92	0.51	1.41	2.30	1.38	1.41	1.25	1.06	0.67	0.45	1.22	1.25	1.02	0.74	0.77	1.12	1.44	
1.18	1.15	1.09	0.99	0.80	0.77	0.70	0.74	0.70	0.80	2.24	4.06	0.96	1.31	2.43	1.25	0.35	1.79	2.08	1.22	0.77	1.15	0.96	0.74	1.12	1.70	1.44	1.02	0.58	0.64	1.06	1.31	
0.83	0.86	0.93	0.96	1.06	1.09	1.22	1.28	1.31	1.34	2.21	4.06	1.28	0.74	2.21	1.09	0.00	1.22	1.54	0.96	1.09	1.28	1.12	0.99	0.83	1.31	1.38	0.90	0.80	0.96	1.06	0.96	
0.70	0.77	0.80	0.80	0.90	0.96	1.02	1.12	1.15	1.25	2.69	4.06	0.67	0.58	2.24	0.93	8.16	1.76	1.34	0.99	1.41	1.50	1.02	0.96	0.67	0.86	1.34	1.09	0.83	0.93	0.96	0.90	
1.25	1.22	1.15	0.99	0.86	0.80	0.80	0.80	0.83	0.93	2.14	4.06	0.26	0.93	2.53	1.12	0.54	1.57	1.31	0.99	1.60	1.25	0.93	1.12	0.54	0.86	0.93	0.58	0.45	1.31	1.15	0.90	
0.83	0.67	0.54	0.48	0.58	0.67	0.80	0.96	1.12	1.31	2.46	4.06	0.93	1.41	1.89	0.96	8.10	0.99	1.66	0.83	1.09	1.82	1.34	0.74	0.74	0.90	0.90	0.90	0.93	1.31	1.31	1.25	
1.09	1.15	1.18	1.22	1.22	1.22	1.15	1.09	1.06	1.02	1.82	4.06	1.15	1.22	1.70	1.12	0.13	1.06	1.66	1.15	1.22	1.79	1.47	0.90	0.67	0.96	0.93	1.02	1.15	1.28	1.31	1.09	
0.80	0.74	0.67	0.64	0.58	0.51	0.61	0.67	0.74	0.83	1.12	4.06	1.63	1.41	2.30	1.50	8.00	0.58	1.95	1.06	0.70	1.76	1.82	1.12	0.51	1.22	1.15	0.86	1.22	1.18	1.18	1.02	
1.12	1.02	0.96	0.90	0.70	0.64	0.70	0.80	0.86	0.96	1.28	4.06	1.76	1.06	2.11	1.60	8.06	0.38	1.57	0.67	0.74	1.92	1.82	0.99	0.74	1.09	0.90	0.61	1.12	1.12	1.15	1.12	
0.93	0.99	1.06	1.06	1.18	1.28	1.28	1.28	1.31	1.31	1.66	4.06	1.47	0.74	1.60	0.93	8.10	0.48	1.31	0.80	0.99	1.54	1.63	1.15	0.90	1.12	1.06	0.96	1.06	1.22	1.31	1.15	
0.96	0.99	1.02	1.09	1.15	1.12	1.12	1.09	1.06	0.96	1.89	4.06	0.70	1.12	1.41	0.86	0.03	0.99	1.31	0.70	1.54	2.08	1.70	1.31	0.93	0.99	0.74	0.51	0.64	0.99	1.22	1.06	
0.86	0.86	0.86	0.93	1.02	1.09	1.22	1.25	1.25	1.28	3.01	3.49	1.44	0.90	2.37	0.51	7.97	0.64	1.02	0.64	0.77	1.63	1.92	1.25	1.09	1.34	1.09	0.90	0.96	1.22	1.15	0.96

Run 1: Bed probe heights: 0.542 0.550 0.750 0.482 0.396 0.326 0.522 0.568 0.635 0.595 0.502 0.481 0.551, mean 0.531, deviation from mean 0.101
Run 2: Bed probe heights: 0.547 0.511 0.761 0.479 0.408 0.299 0.512 0.575 0.629 0.589 0.505 0.474 0.508, mean 0.523, deviation from mean 0.105
Run 3: Bed probe heights: 0.527 0.553 0.745 0.502 0.441 0.316 0.512 0.581 0.618 0.609 0.532 0.474 0.518, mean 0.533, deviation from mean 0.097

[bruceg]

Indeed, getting as close to the bottom of things as possible rather than skating blithely on top of the ice is the only way to make sustainable progress, particularly in the world of machines were most matters are pretty deterministic.

That accelerometer data is interesting. First a likely superfluous (for my competency) question. Is it being collected motion by the motion control board (presume Duet Wifi on your MAX V3) and reported to a terminal session as a matter of course or via custom application/separate device?

More pertinent to your question, i am a little confused about the 0.6/0.35 numbers. Before i fire up MatLab to chase the data around, are they deviations of the corresponding temporal acceleration values across each of the thirteen probe points (i.e. deviation of the thirteen [or 13 x3runs] maximum acceleration values) or the probe height values (in which case if i am reading correctly, you are getting a deviation of ~0.1 within each run) or another metric?

In any event, to take the bait, i would say the best looking statistical results would be achieved by probing once in the centre of the bed but that the best print results would be achieved by a more comprehensive survey across the bed which should...

(given an uneven bed in terms of height and stiffness, not to mention the absolute stiffness of the effector/arm system in different orientations [i.e. is the system stiffer probing at the centre vs directly adjacent to a tower where one of the arm pairs is taking most of the load? {wonder if there is a paper looking at that in a delta/parallel machine...} ], not to mention the effect, relative to location, of uneven belt tension between the towers)

...produce increasing deviation (to a point) with increasing number of locations sampled.

Aside from the mechanical/kinematic aspects, i presume one also has to consider what happens to the information produced by the calibration process. How is this processed by the motion control? Given that i first need to properly understand what transpires when one happily "skates" through the Delta Calibration Wizard (i.e. even the difference and appropriate application of G29 vs the probing G-code produced by the Wizard), i have a long way to go. I have identified a few threads to follow up on for this and certainly would appreciate any pointers on where to best start digging. And to be sure, as i gain experience with the auto calibration, the hotend will be OFF!

[mhackney]

In any event, to take the bait, i would say the best looking statistical results would be achieved by probing once in the centre of the bed but that the best print results would be achieved by a more comprehensive survey across the bed which should...

Very good Grasshopper! You get it.

As for the setup to collect this data - trash80 published his code for Teensy 3.2 a few weeks ago that allows interfacing the HE280 accelerometer to Duet. I published a version for Trinket shortly after. Bit the Teensy has some great features including the ability to actively monitor what's going over the serial (USB) bus. The reason we need an intermediary/adapter like this is because SeeMeCNC choose to do all of the accelerometer configuration and processing in the Repetier firmware. In order to use the HE280 with other firmware/controllers one either needs to implement the code in the firmware or use a more creative solution - this adapter - that talks to the HE280 on one side and then looks like a standard endstop switch to the Duet. So that's what we did.

With the Teensy 3.2 version, I collect the acceleration data for each probe point and can stream it to the console (my computer) to analyze and dynamically see the effects of changes to both the probing parameters (probe speed, acceleration, starting height, etc) and accelerometer configuration - of which there is a lot of stuff in the accelerometer that is not used. For instance, it has the ability to change the full measurement range. The Repetier implementation uses the maximum 16g full scale. But in general it would yield more precise data to reduce the range to the minimum workable range. I am currently using 4g full scale and the accelerometer vendor's application engineer recommended reducing to 2g if possible. But I'm getting false triggering at 2g that might be fixable using high pass filtering and tweaking the probing parameters. So that's what I'm currently working on.

[nebbian]

MHackney,

That's really good data. Good luck with your continued analysis.

Out of interest, what sort of repeatability are you getting when using the nozzle probing? ie if you probe the one point 10 times, what is the sort of deviation you're getting on the measured Z height?

[bruceg]

ah ha, thanks for the info on the collection setup, always good to have an excuse to get another MCU. I stumbled on the Trinket topic but didn't follow it while getting lost in rabbit holes looking into (honest) calibration. i think it was the Smoothie/Duet discussion re calibration strategies leading to getting interested in the Duet's motion output...

I have a challenging (for me) print to produce an artery phantom consisting of a 35mm dia single wall (0.35mm) x 220mm tall cylindrical tube in Ninjaflex that has to be watertight under modest pressure. Results have been tenuously successful with a continuous spiral approach. However under the microscope a distinct segmentation to the arc of the bead laid down is noticeable. I have wondered if this segmentation might impair the continuity of the amazingly thin web that actually joins the "coils" of the spiral. Now, i have to confess i don't know the resolution of the stl model (123D Design doesn't have a parametric setting as far as i can tell) which i suppose is the first place to start but the more "continuous" tool motion of the Duet is certainly appealing.

But first things first, once i have some experience and consistent calibration results with the HE280 and then Prometheus, would like to see if the accelerometer could be applied to the hover type extruder i use for Ninjaflex. To that end, the accelerometer data via the Teensy application is likely to shed some light on the viability of and approach to that particular windmill, so thank you again for the information and all the work that underlies it!!

[cyber.shifter]

My apologies, I didn't mean to suggest erroneous information. I had no idea people had already talked about using ferrite cores, I just know I used to use them a lot in comms work.

[mhackney]

They absolutely have their applications but they did nothing for the original probing issues.

[Saber7]

I am looking at making a similar upgrade to add the accelerometer though I have an E3D Cyclops currently installed and working well But sometimes I want to just be able to switch back to my E3D V6 which is mounted to a second effector plate and calibrate quickly. It seem like that hole in the center of the accelerometer doesn't look like it would fit the required tubes and wiring on my cyclops effector. Does anyone know if the accelerometer orientation matters? Basically could I design a custom mount and set the board vertically next to the hot end heat sink or is the probe orientation defined in the software or specs of the accelerometer on the board?

If orientation does matter and needs to be horizontal is it worth building my own circuit board with a custom hole? Or is the probing more of a gimmick, and most of you still end up manually calibrating anyway?

[mhackney]

I have a cyclops and can verify that the hole in the PCB will not work with it. Yes, the orientation is critical. The PCB must be as near planar with the earth's surface as possible. Ironically, I was just in the process of calculating effector/printer tilt with the accelerometer today since I was seeing at rest values significantly below 1g.

Probing IS NOT a gimmick and if done well results in much better and consistent prints. At this point, I have done as much accelerometer testing as I can - I was obsessed with it. My conclusion - and this is strictly for my work and quality standards as it can produce acceptable results for most users - is that FSR probing is much more reliable and precise when implemented properly (some folks have difficulty with that too). I'm converting my V3 to FSRs as I type. I will leave the accelerometer probe connected so I can actually do back-to-back probing comparisons to soothe any naysayers. But I have data from 100s of accelerometer probe runs along with manual calibration data during the process and that has convinced me that for my very persnickety work, FSRs are the way to go. I should go on to say, and I'm talking explicitly about probing using the Duet and RepRapFirmware which has the new grid compensation. To get the first layer control I need to print over the entire build plate with +/- 0.05mm precision requires this level. Most things "normal" people print do not!

[mhackney]

@Saber7 - re the Cyclops, it is a very difficult hot end to use. I spent many months learning how to use it. In the end, I found that the Bondtech QR extruder is the only extruder (that I've tested/used) that is capable of pushing filament though that torture chamber filament path that has two 90°turns per filament. http://sublimelayers.blogspot.com/2015/ ... imate.html scroll down to Bondtech QR Extruder.

[Saber7]

Thanks @mhackney I was having issues with my cyclops and I actually got it working decently due to some of your posts awhile back!

So with the cyclops did you manage to figure out how to mount the accelerometer and route everything around it? Basically trying to figure out if its worth spending $30 x 2 for both effectors and then figuring out how to jam it in there somehow on the cyclops. I still switch back ocasionally to my V6 as it has the volcano on it for some of the larger prints I do, but I have been doing it less and less because the calibration sucks and I am lazy sometimes. I might look into the FSR's though if it doesn't make sense to do the accelerometer board.

[mhackney]

No, I have many printers and all of them except the V3 have FSR probing systems. I use the Cyclops on one of those printers.

[dlbyers]

I just followed your process and I'm getting the best first layer I've ever seen my V2 produce. I used the V3 option in the connfiguration.h.

Only thing I had to change is the Z-MAX... Initially it was set at 385. This resulted in a probe followed by dragging across the bed to the X tower, and it appeared to be trying to correct sensitivity in the process, until finally it crashed with belt grinding noise. I backed Z-MAX back to 350 per another post and it calibrated correctly and I'm off & running.

Thank you for putting this list together!

[adarcher]

[img]http://forum.seemecnc.com/download/file ... w&id=14232[/img]

This looks nice and colorful! Mind mentioning the models if they are available?

[bruceg]

I used the V3 option in the connfiguration.h. Only thing I had to change is the Z-MAX... Initially it was set at 385

Indeed, a number of folks have reported a crash with a v2 if they started out with 385 instead of 350...

Glad it is working for you.

[bruceg]

@ adarcher,

i am just finishing up a suite of standoffs for different nozzle sizes/setups. Will try and get them posted on thingiverse in the next couple of days. Hardware wise the system uses 6 of M4x16mm Button Head socket head cap screws

[bruceg]

[img]http://forum.seemecnc.com/download/file ... w&id=14232[/img]

This looks nice and colorful! Mind mentioning the models if they are available?

and here they are in all their multihued splendor....

[adarcher]

[img]http://forum.seemecnc.com/download/file ... w&id=14232[/img]

This looks nice and colorful! Mind mentioning the models if they are available?

and here they are in all their multihued splendor....

Thank you!

[nrbelk]

Anyone know where to go for calibration issues? I upgraded my v2 all the way to v3, but I use the e3d v6 hot end.

I used their new whip to simplify my wiring.

But when I run G29, my hotend goes down the proper distance, but then goes too far and hits the X tower, goes down to the bed, and continues to try to move (grinding the belts) until I turn off the power.

[dlbyers]

Anyone know where to go for calibration issues? I upgraded my v2 all the way to v3, but I use the e3d v6 hot end.

I used their new whip to simplify my wiring.

But when I run G29, my hotend goes down the proper distance, but then goes too far and hits the Z tower, goes down to the bed, and continues to try to move (grinding the belts) until I turn off the power.

I had the same problem and had to change the Z-MAX in configuration.h from 385 to 350.