EndStop Accuracy Test

deckingman

For info, the repeatability of a cheap mechanical end stop is about +_0.2mm and one micro-step is about 0.0125mm (at 80 steps per mm). So if you want to check for missed steps using an end stop, you need one with a repeatability better than 0.0125mm.

But take a look at the macros that @jay_s_uk posted in this thread https://forum.duet3d.com/topic/32433/determining-max-speed-and-acceleration?_=1695128541482. They might give you some ideas.

Essentially that thread was about trying to find the maximum speeds and accelerations that could be used before missed steps were encountered by homing the printer, doing a series of high speed moves, then using the end stops to check the positional accuracy. So there are similarities with what you are trying to do and the macros might be useful (assuming you can find a suitable end-stop switch with the required repeatability).

dc42

@Nazar I think G1 H4 will do exactly what you need.

Nazar

@dc42 Thank you, I will test it tomorrow
@deckingman Thank you for the reference, I use high quality end stops, I want to look for missed steps

deckingman

@Nazar said in EndStop Accuracy Test:

@deckingman ................... I use high quality end stops, I want to look for missed steps

You don't say what the repeatability of your end stop switches are though. Just because they are high quality, doesn't mean the repeatability will be good enough when looking at differences of 0.0125mm. You certainly won't do it with a conventional micro-switch. I can recommend Metrol Ultra Precision switches which have a claimed repeatability of 0.005 mm (so roughly half of one micro-step). I used to use one as my Z "probe". Here in the UK they cost about £90 GBP plus taxes and shipping.

jens55

@deckingman, what made you go with the Metrol switch when a BLTouch has the same repeatability/accuracy?

deckingman

@jens55 said in EndStop Accuracy Test:

@deckingman, what made you go with the Metrol switch when a BLTouch has the same repeatability/accuracy?

Two reasons. Firstly it was before the BL touch came into existence. Secondly the application was a bit different in that I use the actual nozzle as a probe. My hot ends are mounted in such a way that they have zero movement in X and Y but are able to move in Z a few mm (against a spring). I wanted to be able to detect that vertical hot end movement. Nowadays I use two solid brass contacts which form the physical hot end stop but also make a switch with excellent repeatability and zero histeresis (and it's a fraction of the cost).

Nazar

@dc42 Thank you so much for your advice, G1 H4 works for my application

@deckingman Thank you for bringing a point and end stop recommendation, I am using Omron D2F micro switches (with no arm for increased accuracy). The result I have are controversial, I am getting standard deviation the data set approximately 0.1129 mm, which is much higher that I was expecting, the test code is attached bellow:

echo >"0:/macros/accuracy.g" "EndStop Test Start"
echo >>"0:/macros/accuracy.g" " "

G28 X
G90
G1 Z50 F18000


while iterations <100
	M400
	G91
	G1 X2 F600
	G1 H4 X-10 F240
	var dd = move.axes[0].min  - move.axes[0].machinePosition
	echo "X "^{var.dd}
	echo >>"0:/macros/accuracy.g" ""^{var.dd}

My machine is able to conductively probe the aluminum build plate with a nozzle, I have run the script attached bellow an got similarly large standard divination of 0.0967mm

echo >"0:/macros/accuracy.g" "Nozzle Test Start"
echo >>"0:/macros/accuracy.g" " "
M42 P4 S1        ; Turn on relay, engage probing (ESD Warning)
M208 Z-3.5 S1    ; set axis minima

G1 X-999 U999 Y0 Z5 F18000
G1 Z-99

G91
G38.4 K0 X10
M400
G92 X0

while iterations <100
	M400
	G91
	G1 X-1 F600
	G38.4 K0 X2
	var dd = move.axes[0].machinePosition
	echo "X "^{var.dd}
	echo >>"0:/macros/accuracy.g" ""^{var.dd}

G1 Z10 F18000
M208 Z0 S1    ; set axis minima
G90
G28 X

I have rigidity mounted the same end stop to the toolhead, run the script below, and got deviation from mean 0.002mm, again this is with same end stop, just mounted to toolhead and measured with Z - Axis

G30 P0 X0 Y0 Z-99999
G30 P1 X0 Y0 Z-99999
G30 P2 X0 Y0 Z-99999
G30 P3 X0 Y0 Z-99999
G30 P4 X0 Y0 Z-99999
G30 P5 X0 Y0 Z-99999
G30 P6 X0 Y0 Z-99999
G30 P7 X0 Y0 Z-99999
G30 P8 X0 Y0 Z-99999
G30 P9 X0 Y0 Z-99999
G30 P10 X0 Y0 Z-99999
G30 P11 X0 Y0 Z-99999
G30 P12 X0 Y0 Z-99999
G30 P13 X0 Y0 Z-99999
G30 P14 X0 Y0 Z-99999
G30 P15 X0 Y0 Z-99999
G30 P16 X0 Y0 Z-99999
G30 P17 X0 Y0 Z-99999
G30 P18 X0 Y0 Z-99999
G30 P19 X0 Y0 Z-99999
G30 P20 X0 Y0 Z-99999
G30 P21 X0 Y0 Z-99999
G30 P22 X0 Y0 Z-99999
G30 P23 X0 Y0 Z-99999
G30 P24 X0 Y0 Z-99999
G30 P25 X0 Y0 Z-99999
G30 P26 X0 Y0 Z-99999
G30 P27 X0 Y0 Z-99999
G30 P28 X0 Y0 Z-99999
G30 P29 X0 Y0 Z-99999
G30 P30 X0 Y0 Z-99999 S-1

Results:

G32 bed probe heights: 0.020 0.020 0.016 0.020 0.016 0.015 0.016 0.020 0.016 0.019 0.015 0.015 0.016 0.016 0.021 0.017 0.016 0.016 0.016 0.016 0.015 0.020 0.020 0.015 0.016 0.015 0.015 0.016 0.016 0.015 0.015, mean 0.017, deviation from mean 0.002

When I run test using G38.2 with the same end stop attached to the toolhead I get much larger divinations of 0.2657mm, code is attached below:

echo >"0:/macros/accuracy.txt" "Z probe with G38 Test Start"
echo >>"0:/macros/accuracy.txt" " "



G1 X0 U999 Y0 Z20 F18000
M42 P4 S1        ; Turn on relay, engage probing (ESD Warning)

G91
G38.2 K0 Z-30
M400
G92 Z0

while iterations <100
	M400
	G91
	G1 Z2 F600
	G38.2 K0 Z-3
	var dd = move.axes[2].machinePosition
	echo "X "^{var.dd}
	echo >>"0:/macros/accuracy.txt" ""^{var.dd}

G1 Z50 F18000
G90

@deckingman what do you think?

I have a theory that large deviations can be caused by reading them with G38 or G1 H2 followed by move.axes[2].machinePosition variable, meaning we measure not a triggered position but the position after it was triggered and stopped somewhere @dc42 what do you think?

dc42

@Nazar in RRF 3.5.x the machine position is read during moves instead of only at the end of a move, and at most every 250ms. So if you read the machine position immediately after the G1 H4 or G38.2 move has completed, you may gat a value that is up to 250ms out of date.

I suggest you insert command G4 P260 after the G1 or G38.2 command, before you read the machine position.

mrehorstdmd

@Nazar I ran a couple test prints after I swapped out microswitches for opto endstops in my printer. The precision of the opto endstops was very high. I didn't make measurements, but I compared prints, one made with homing only at the start of the print and the other with homing at each layer change. See: https://drmrehorst.blogspot.com/2020/03/testing-ummds-xy-optical-endstops.html

The endstops I used were purchased on Amazon for 3 for $10 and included an LM393 comparator chip. They claim 0.002mm precision. I haven't seen the same endstops for sale there for a while, but there are probably others like them available elsewhere for similar price.

I ran into one problem with opto endstops in my printer- the work room where the printer sits has a large, east-facing window and sunlight falling directly on the printer interfered with them. It took a while to figure out what was going on, but closing the curtains on the window solved the problem. I've never had any other problems with them.

deckingman

@mrehorstdmd said in EndStop Accuracy Test:

................ I ran into one problem with opto endstops in my printer- the work room where the printer sits has a large, east-facing window and sunlight falling directly on the printer interfered with them. It took a while to figure out what was going on, but closing the curtains on the window solved the problem. I've never had any other problems with them.

That reminds me of a similar issue I came across. I use a slotted opto switch to get my bed somewhere close at highish speed before homing "proper" using the nozzle. When I first tried to test the switch, it didn't seem to be triggering until I realised that it was seeing straight through the thin piece of paper I was using as a "flag". (It doesn't see though a piece of 2mm thick aluminium though).