Anyone interested in a differential IR Z probe for deltas?

Quote
Koenig
Have you made any more of these?

I have the 1.0 version of this already but I just bought a E3D v6 genuine and I don't think the earlier version will work with that and your fanshroud.

So I'm interested in a new one!

If you check DC$"'s Thingiverse link there are 2 STL's in there one for the V1.0 board and one for the 1.1 so you should be ok.

(I also have 2 x 1.0 boards but have yet to print the shroud as I have not quite got the 2nd Delta ready for it).

Doug

Quote
dougal1957

Quote
Koenig
Have you made any more of these?

I have the 1.0 version of this already but I just bought a E3D v6 genuine and I don't think the earlier version will work with that and your fanshroud.

So I'm interested in a new one!

If you check DC$"'s Thingiverse link there are 2 STL's in there one for the V1.0 board and one for the 1.1 so you should be ok.

(I also have 2 x 1.0 boards but have yet to print the shroud as I have not quite got the 2nd Delta ready for it).

Doug

The only links I could find here fitting the 1.0 version was for the J-head and v6, but on his blog there was a link to a remix for the 1.0 version and E3D v6, thank you for the tip.

But I cannot get it to trigger against glass......

Also I would still be interested in getting another, I have some plans for a larger delta. If only I could get it to work......

Quote
Koenig
I have it wired to the duet, it flashes four times when starting up, but it doesn't trigger when going down.
If I put a piece of white paper on the glass I get it to trigger at ~Z=1.50, but if I remove the paper it doesn't trigger at all and the reading in the web-interface doesn't change either, the sensor reading stays at 0 all the way down until the nozzle touches the bed at Z=-0.1.

Sorry to hear you are having problems. Two questions:

1. How thick is the glass, and what sort is it?

2. When the nozzle is touching the glass, how high above the glass is the edge of the board?

I have about 4 version 1.1 boards in stock and another batch being manufactured next week.

Edited 1 time(s). Last edit at 09/05/2015 04:01AM 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].

Quote
dc42

Quote
Koenig
I have it wired to the duet, it flashes four times when starting up, but it doesn't trigger when going down.
If I put a piece of white paper on the glass I get it to trigger at ~Z=1.50, but if I remove the paper it doesn't trigger at all and the reading in the web-interface doesn't change either, the sensor reading stays at 0 all the way down until the nozzle touches the bed at Z=-0.1.

Sorry to hear you are having problems. Two questions:

1. How thick is the glass, and what sort is it?

3mm, borosilicate

Quote
dc42
2. When the nozzle is touching the glass, how high above the glass is the edge of the board?

According to my calculations when I made the fanshroud to fit my E3D v5 clone it should be 1.5mm, but when I tried to measure it in reality I could fit an 2mm allen key underneath with one of the flat sides down (2.01mm) but I cannot get it underneath if I turn it so that one of the edges is against the bed (2.25mm) so somewhere in between that.

Is that to much?

Quote
dc42
I have about 4 version 1.1 boards in stock and another batch being manufactured next week.

I have plans of building a larger delta using these vertices: [groups.google.com] and some 20*60 extrusions so I'm gonna need another one of your sensors anyway, if I get it to work.
That project is going to take some time though.

Edited 2 time(s). Last edit at 09/05/2015 06:29AM by Koenig.

I'm gonna try redesigning my fanshroud so the board gets 0.7mm closer to the bed to see if that solves it.

Quote
Koenig
I'm gonna try redesigning my fanshroud so the board gets 0.7mm closer to the bed to see if that solves it.

Have you tried putting some Black Paper/Card underneath the Glass (This is what Dave recommends and I use successfully with the probe approx. 2.5 mm above the Nozzle) this is a much quicker test If you don't have any to hand just print a page of all Black (Use a paint package or something to generate a all black page)

Doug

Quote
dougal1957

Quote
Koenig
I'm gonna try redesigning my fanshroud so the board gets 0.7mm closer to the bed to see if that solves it.

Have you tried putting some Black Paper/Card underneath the Glass (This is what Dave recommends and I use successfully with the probe approx. 2.5 mm above the Nozzle) this is a much quicker test If you don't have any to hand just print a page of all Black (Use a paint package or something to generate a all black page)

Doug

Painted (sort of, black permanent marker) my glass black underneath on his recommendation.

His recommendation is to get the edge of the board 1.5mm above the bed when the nozzle touches, my rather rough calculations/guesstimates when I designed my fanshroud for my clone seems to be 0.5 to 0.7mm off.

Printing a new now!

Edited 1 time(s). Last edit at 09/05/2015 07:28AM by Koenig.

I had same problem with not triggering on glass.
I would try a flat black sheet of paper like construction paper. Permanent marker on glass is very shiny. He also recommended painting the aluminum plate, not the glass if I recall correctly. That should optically decouple the backside better.
But get the height correct first . Maybe make the mount adjustable? I also had to bend the IR component slightly as shown in his Z-probe blog.
BTW, I used a drill bit to physically measure the gap, better precision if you have a full set of bits.

When testing the sensors, I measure the trigger height against 3mm float glass with black paper underneath, and I reject or adjust any with trigger heights below 2.5mm. I haven't tried black marker pen on the back of the glass. It's possible that the trigger height may be a little different using borosilicate glass because of the different refractive index, and hence the different contribution from the reflection from the back surface. My recommendation of mounting the edge of the sensor board no more than 2mm higher than the nozzle is intended to ensure at least 0.5mm of leeway.

---

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
shadowphile
I had same problem with not triggering on glass.
I would try a flat black sheet of paper like construction paper. Permanent marker on glass is very shiny. He also recommended painting the aluminum plate, not the glass if I recall correctly. That should optically decouple the backside better.
But get the height correct first . Maybe make the mount adjustable? I also had to bend the IR component slightly as shown in his Z-probe blog.
BTW, I used a drill bit to physically measure the gap, better precision if you have a full set of bits.

I did OK with the allen keys, they range from 1 - 1.5 - 2 - 2.5 - .... and so on + they have 2 measurements, one with the flat side down and one with one of the "corners" down.

And I wouldn't know what a full set of drillbits mean, one for each 10th of a mil or maybe a 5th of a mil? There are so many all depending of the precision you want to achieve.

Making it adjustable would be a good idea, just that trying to get it as close to the nozzle as possible makes very little room for inserting any kind of tool, + I had the design already in Solidworks so changing a measurement wasn't to hard + I'm going to change hotend very soon so I don't want to spend to much time on a design I'm not gonna use anyway.

For some reason I don't feel like coloring my hotbed, just OCD or something but anyway.....

Now I'm having other issues, the fanshroud I printed took a chunk out of my glassbed.... (more about it)

OK, so a new fanshroud is printed and the IR-sensor seems to be working, however I seem to have a very "low" offset, it triggers when my head is 0.45mm above the bed, and the edge of the board is now 1.4mm above the bed.

But a new issue appeared, when I issue a "G32" the printer acts up totally, it homes first and then the axis start to go down, only in different speeds so one carriage pushes past the effector down real quickly....

Can't figure out what does this.

EDIT: It seems the command G30 does it.
If I issue a "G30 P0 X-73.6 Y-42.5 Z-99999" directly to the printer it behaves this way as well.

Edited 2 time(s). Last edit at 09/06/2015 10:56AM by Koenig.

G32 and G30 use the M203 maximum speeds as the movement speed to the probe position, so you probably have the M203 values too high for your motors and/or the firmware. Try lower M203 values. Upgrading to my latest firmware (1.09j) may also help achieve high speeds.

---

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
G32 and G30 use the M203 maximum speeds as the movement speed to the probe position, so you probably have the M203 values too high for your motors and/or the firmware. Try lower M203 values. Upgrading to my latest firmware (1.09j) may also help achieve high speeds.

Had already updated to the latest before I did my last edit of the previous post.

However, it was the MAX speed that did it.

But now I'm a bit perplexed, I have run the G32 4 times with this result:

Quote
autocal log
17:57:47M666
Endstop adjustments X-0.22 Y0.18 Z0.03
17:54:07M665
Diagonal 250.42, delta radius 121.02, homed height 463.72, bed radius 105.0, X -0.09°, Y -0.02°, Z 0.00°
17:50:59G32
Calibrated 6 factors using 10 points, deviation before 0.118 after 0.118
17:49:53G32
Calibrated 6 factors using 10 points, deviation before 0.120 after 0.118
17:39:56G32
Calibrated 6 factors using 10 points, deviation before 0.119 after 0.116
17:38:45G32
Calibrated 6 factors using 10 points, deviation before 0.156 after 0.115
17:37:48Connection established!
17:37:48Page Load complete!

Quote
manual cal
M665 R121.1 L250.42 B105.0 H463.66
M666 X-0.67 Y0.07 Z-0.35

But I got flatter movement with my manual calibration then I get with the autocal.

Now In the middle the head touches the bed thus not extruding very much, this behaviour also happens spotvise on other places of the bed, no discernible pattern, except that it seems to slowly go into the bed and then at the next turn it seems to lift and overextrude...

EDIT: Tried to take a photo of a prints first layer, but the camera didn't like black plastic on a black glassbed......

Edited 2 time(s). Last edit at 09/06/2015 04:35PM by Koenig.

0.118 deviation after autocal isn't very good, you should be aiming for 0.05mm or less. The deviation could be caused by the trigger height varying with XY position. This is how I check mine:

1. Heat the bed and hot end to operating temperature.
2. Move the effector to one of the XY coordinates used for bed probing.
3. Lower the nozzle slowly, in 0.025mm steps to finish with, until the nozzle just grips a sheet of paper.
4. Send G92 Z0 to define that point as Z=0, and remove the paper. Optionally, use feeler gauges to measure the height of the edge of the sensor board above the bed.
5. Raise the nozzle to about 5mm, then send G30 S-1 to measure the trigger height. Repeat a few times to get an average. On my delta, the trigger height is reproducible to within about 0.01mm at a given point.
6. Repeat steps 2 to 5 to establish the trigger heights at the other probe points.

If the bed surface is reasonably consistent and the effector does not tilt with varying XY position, then the trigger heights should be the same everywhere. If they are not, there are a couple of possibilities:

1. The effector tilt is varying with XY position. This can be caused by the diagonal rods in a pair not being quite the same length, or the spacing between the bearings at the upper and lower ends of the rods being not quite the same. On my machine, there is a 0.25mm change in trigger height close to the Z tower, caused by a combination of a loose Traxxas joint and the sideways force of the Bowden tube acting in different directions at different XY positions.

2. Bed surface changing a lot between probe points. This could be because you have coated the centre e.g. with glue, but the edges where some of the probe point are have not been coated. Or one of the probe points places the sensor too close to the edge of the bed. Some glues and hairsprays affect the trigger height by around 0.2mm while other brands have little effect.

It's best to get rid of any trigger height differences if you can. For any that remain, you can use the H parameter in the G30 commands in bed.g to make a correction at each probe point.

HTH David

---

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
0.118 deviation after autocal isn't very good, you should be aiming for 0.05mm or less. The deviation could be caused by the trigger height varying with XY position. This is how I check mine:

1. Heat the bed and hot end to operating temperature.
2. Move the effector to one of the XY coordinates used for bed probing.
3. Lower the nozzle slowly, in 0.025mm steps to finish with, until the nozzle just grips a sheet of paper.
4. Send G92 Z0 to define that point as Z=0, and remove the paper. Optionally, use feeler gauges to measure the height of the edge of the sensor board above the bed.
5. Raise the nozzle to about 5mm, then send G30 S-1 to measure the trigger height. Repeat a few times to get an average. On my delta, the trigger height is reproducible to within about 0.01mm at a given point.
6. Repeat steps 2 to 5 to establish the trigger heights at the other probe points.

If the bed surface is reasonably consistent and the effector does not tilt with varying XY position, then the trigger heights should be the same everywhere. If they are not, there are a couple of possibilities:

1. The effector tilt is varying with XY position. This can be caused by the diagonal rods in a pair not being quite the same length, or the spacing between the bearings at the upper and lower ends of the rods being not quite the same. On my machine, there is a 0.25mm change in trigger height close to the Z tower, caused by a combination of a loose Traxxas joint and the sideways force of the Bowden tube acting in different directions at different XY positions.

2. Bed surface changing a lot between probe points. This could be because you have coated the centre e.g. with glue, but the edges where some of the probe point are have not been coated. Or one of the probe points places the sensor too close to the edge of the bed. Some glues and hairsprays affect the trigger height by around 0.2mm while other brands have little effect.

It's best to get rid of any trigger height differences if you can. For any that remain, you can use the H parameter in the G30 commands in bed.g to make a correction at each probe point.

HTH David

All the rods are within 0.02mm.

Going to check how parallel my rods are, just that this driving the hotend into the bed seems to have ripped one of my belt-fasteners right of the carriage.... Must see to that first.

Quote
dc42
0.118 deviation after autocal isn't very good, you should be aiming for 0.05mm or less. The deviation could be caused by the trigger height varying with XY position. This is how I check mine:

1. Heat the bed and hot end to operating temperature.
2. Move the effector to one of the XY coordinates used for bed probing.
3. Lower the nozzle slowly, in 0.025mm steps to finish with, until the nozzle just grips a sheet of paper.
4. Send G92 Z0 to define that point as Z=0, and remove the paper. Optionally, use feeler gauges to measure the height of the edge of the sensor board above the bed.
5. Raise the nozzle to about 5mm, then send G30 S-1 to measure the trigger height. Repeat a few times to get an average. On my delta, the trigger height is reproducible to within about 0.01mm at a given point.
6. Repeat steps 2 to 5 to establish the trigger heights at the other probe points.

If the bed surface is reasonably consistent and the effector does not tilt with varying XY position, then the trigger heights should be the same everywhere. If they are not, there are a couple of possibilities:

1. The effector tilt is varying with XY position. This can be caused by the diagonal rods in a pair not being quite the same length, or the spacing between the bearings at the upper and lower ends of the rods being not quite the same. On my machine, there is a 0.25mm change in trigger height close to the Z tower, caused by a combination of a loose Traxxas joint and the sideways force of the Bowden tube acting in different directions at different XY positions.

2. Bed surface changing a lot between probe points. This could be because you have coated the centre e.g. with glue, but the edges where some of the probe point are have not been coated. Or one of the probe points places the sensor too close to the edge of the bed. Some glues and hairsprays affect the trigger height by around 0.2mm while other brands have little effect.

It's best to get rid of any trigger height differences if you can. For any that remain, you can use the H parameter in the G30 commands in bed.g to make a correction at each probe point.

HTH David

OK, checked how parallel my rods are, and X and Y are within 0.07mm difference at carriage or effector (don't know really how accurate that is though, the carbon fiber rods have a little rough surface, I actually get ~0.07mm difference just measuring at difference locations on one rod, wich then make a much bigger fault marginal, like 0.21mm or so...) and my Z rods are within 0.17mm (which would make it like possibly 0.31mm), this is the carriage that broke.
Now I don't know how much that translates in the XY-plane.

I have started measuring the trigger-heights at the various points, and the value I get, should I add/subtract that from my trigger-height set with G31 and add the difference as the H value in the G30 or should I just put measured value as the H parameter?
This was sort of unclear reading the wiki.

Edited 1 time(s). Last edit at 09/07/2015 11:23AM by Koenig.

The H value in the G30 command gets added to the G31 Z parameter to determine the trigger height at that point. Set the G31 Z parameter to the trigger height at the centre of the bed. At the other points, set the H parameter to the actual trigger height at that point minus the trigger height at the centre.

---

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].

Now something else happened, I cannot connect to the web interface anymore, but if I connect with S3D first and then try with the web interface I get this "RECEIVED: Webserver: rejecting message with: too few command words" in the log window.....
Using Chrome.

EDIT: Solved, I think, had to remove a cookie.

Edited 2 time(s). Last edit at 09/07/2015 01:50PM by Koenig.

OK, so I have gotten the deviation below 0.05:

Quote
web log
21:57:09M666
Endstop adjustments X-0.16 Y0.66 Z-0.50
21:57:03M665
Diagonal 250.42, delta radius 120.78, homed height 463.68, bed radius 105.0, X 0.43°, Y 0.01°, Z 0.00°
21:56:53G32
Calibrated 6 factors using 10 points, deviation before 0.019 after 0.011
21:56:08G32
Calibrated 6 factors using 10 points, deviation before 0.017 after 0.012
21:55:24G32
Calibrated 6 factors using 10 points, deviation before 0.015 after 0.009
21:54:41G32
Calibrated 6 factors using 10 points, deviation before 0.015 after 0.012
21:53:56G32
Calibrated 6 factors using 10 points, deviation before 0.019 after 0.014
21:53:13G32
Calibrated 6 factors using 10 points, deviation before 0.013 after 0.009
21:52:28G32
Calibrated 6 factors using 10 points, deviation before 0.017 after 0.005
21:51:43G32
Calibrated 6 factors using 10 points, deviation before 0.039 after 0.013
21:50:48G32
Calibrated 6 factors using 10 points, deviation before 0.244 after 0.033
21:49:56Connection established!
21:49:56Page Load complete!

The radius is a lot closer to what I had when I made a manual calibration, so I hope/suppose it won't hit the bed in the center anymore.
There's a lot of manual labor to getting the autocalibration to work...

We'll see how it prints tomorrow.

Would adding more probe-points decrease the deviation further?
I only have a 220mm bed so perhaps adding another 3 points would be meaningless?

Quote
Koenig
Would adding more probe-points decrease the deviation further?
I only have a 220mm bed so perhaps adding another 3 points would be meaningless?

Adding more probe points will quite likely increase the reported deviation, because the firmware will then have to seek a compromise that minimises the sum of the squares at more probe points. Whereas if you use 6-factor auto calibration and only 6 probe points, you will get down to near zero deviation, because the firmware can solve 6 unknowns exactly from 6 equations - but that tells you nothing about how flat the printing plane is. For a 220mm bed, I think 10 points should be enough. If in doubt, print a calibration spiral and see how uniform the thickness is, especially in the areas farthest away from the probe points.

---

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].

Can anybody tell me what the average deviation before/after is obtained using IR sensor calibration for delta-printers?

I'm just talking about this line from logs:

G32 Calibrated 6 factors using 16 points, deviation before 0.027 after 0.027

David, what is your deviation for your printer with 0.9-degree steppers?

Quote
roboprint
David, what is your deviation for your printer with 0.9-degree steppers?

Currently it is about 0.08mm using 13 probe points and 6 factor calibration. This is on a 300mm diameter bed. There is play in some of the Traxxas joints that is causing effector tilt in the +Y direction, and probably other issues too. AFAIR the best I ever had was 0.03mm, but that was probably with 7 factors and 10 points instead of 6/13.

Of course, if I choose 6 factors and only 6 points, I will get zero deviation! But that's cheating.

Edited 3 time(s). Last edit at 10/06/2015 04:38AM 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].

David, doesn't the calibration fit a flat plane to least squares of the probe points? I got my large delta finished but my cal results won't go below about .13 St Dev. It seems like the results won't be good unless the physical accuracy is right on. I am in the process of redesigning my stage carriages because they aren't square and stiff enough (and of course I reused the mini Kossel parts so now I have no way to print a redesign).

oh..that is with 15 points btw in a roughly even spread.

Quote
shadowphile
David, doesn't the calibration fit a flat plane to least squares of the probe points? I got my large delta finished but my cal results won't go below about .13 St Dev. It seems like the results won't be good unless the physical accuracy is right on.

Yes, that's right. The larger your printable area is, the more accurate your build needs to be to get a low deviation over the whole of the bed. I have at various times had the following issues with my printer:

- Glass bed not flat (I was assuming that float glass is always very flat, but one of the 3 build plates I had cut is distinctly bent)
- Towers not perpendicular to the bed (much better now that I have metal corners
- Slop in Traxxas joints (I need to either make up some more rods, and either be more careful how I assemble them or switch to ball-and-socket joints)
- Bearing separation not quite the same at the upper and lower ends of the diagonal rods

---

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].

Hi Dave, I have been experimenting with this concept on my cartesian printer and wondering if you had any insight. My printer is a CoreXY (C-Bot) with a MIC-6 aluminum bed, with a sheet of PEI stuck to it (dull side up) with a solid sheet of 468MP transfer tape. My results were highly repeatable measurement of probe distance at the same point, but I was getting vastly different values depending on where I probed. What was happening was that the "far" LED was saturating before the "near" LED would pass it. The yellow pulse here is the "far" LED turn on so the two humps are far response, near response respectively.


As the bed continues to get closer, the far LED will stay saturated, then the near LED will saturate, and finally the far LED will start to drop off. Unfortunately, at some points along the surface, the distance at which the far LED saturates varies. There are some bubbles in the transfer tape so maybe that's scattering light, or there is an area of higher dullness in the PEI. I lowered the value of R1 to 1k to decrease the photodiode's sensitivity (tried down to 680 ohm and up to 47k) and that didn't seen to help enough, so I increased the 150 ohm resistor on D2 (the far LED) to 330 ohm. The sensor stopped working very well on matte surfaces but no longer saturated before the "differential" part kicked in which gave me usable readings.

Jogging the head around a 1mm square I still see a pretty large variation in relative hump sizes, which leads to Z probe results like this (0.05mm difference across a 0.7mm line):

X100 Y100 Z:0.6525 C:522
Y100.1 Z:0.6413 C:513
Y100.2 Z:0.6338 C:507
Y100.3 Z:0.6237 C:499
Y100.4 Z:0.6175 C:494
Y100.5 Z:0.6112 C:489
Y100.6 Z:0.6062 C:485
Y100.7 Z:0.6037 C:483

I will note that if I G1 Y100 and probe there again I get the exact same Z/C (or off by one C occasionally) so the results are consistent for the same point. 0.05mm accuracy is good enough, but it is difficult to gauge if that accuracy is across the entire 300x300 bed and know what the relative min and max error is caused by surface reflectivity or refraction.

So all this typing was to just get to this point where I ask if you had any idea on how to mitigate these differences in hardware, which maybe are caused by things under the surface of the PEI sheet? Or maybe the surface reflection? I didn't know if you had been doing more research or had gone through this in your design process. I can't really change my bed surface because it is all pretty much glued together.

First, from the wave shape I deduce that the scope probe you are using has a high capacitance. The waveform across R1 should be much sharper than that. Try using a x10 probe.

The sensor is not intended to work with the phototransistor saturated. If it does saturate, the output will be set to 'triggered' to protect against driving the head into the bed. The point at which it saturates will vary greatly with the reflectivity of the bed. This is probably why you are seeing such a large variation in trigger height.

The only time I see the sensor saturate is when the target is bright aluminium. Unfortunately I don't have any PEI sheet, so I have no idea how transparent it is to IR. From your scope trace, I guess the PEI is highly transparent and you are seeing the reflection from the bright aluminium underneath. So it looks as if it is behaving like glass. With glass on aluminium, I strongly recommend painting the aluminium black to avoid the strong reflection.

If you can't do that, then my advice is to restore the LED series resistors to 130 ohms (far) and 150 ohms (near) [these values assume that the IR LEDs on your board are pink, not clear], and reduce R1 below its nominal value of 6.8K so that the phototransistor doesn't saturate before it triggers. If you are using a 5V supply, this means that the voltage across R1 must not exceed about 3.9V. Try 1K, or even lower if necessary. Ideally, the signal from both LEDs should be about 0.5V to 2V at the crossover point.

Depending on how thick the PEI sheet is, you may then find that the sensor doesn't trigger even with the nozzle touching the bed. If that happens, try reducing the height of the bottom edge of the board to 1mm above the tip of the nozzle. If that still isn't enough, you could try increasing the trigger height in the manner I describe in the Modifications section of the fitting instructions.

My only other suggestion is to sand or scour the surface of the PEI to reflect more of the IR from the top surface. But I don't know how that would affect the adhesion properties, and it would need to be done uniformly to give a consistent trigger height.

HTH David

Edited 1 time(s). Last edit at 10/19/2015 05:00PM 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].

You're right David, the shapes on the scope probably do look different than you're used to, good eye! I was using the probe at 1X and also I had a 10k resistor on R1 so it definitely would be be as sharp as you're used to seeing. I've already reduced R1 down to 1K and was still getting saturation. It works great on all sorts of different surfaces but boy the aluminum underneath must really reflect some serious IR. Before I swapped out the 6.8k R1 for 1k, it would trigger at a very far distance (6-7cm) as the reflection hit some sort of resonance point, then would turn off, and finally trigger again as the bed got close (and was saturated).

It's good to know that the crossover point should be 0.5V-2V, I can use that to maybe tune the responsiveness and IR output to get a perfect result. As it is currently with the low IR output and 1k resistor, there's very little response in the curves when in proximity to any sort of "normal" surface but it works much better on the aluminum bed. I can't paint the aluminum unfortunately because it's covered with the PEI and a 300mm square of transfer tape which would be near impossible to remove. I also have an second sensor that I'm going to try on a second printer which is aluminum bed + glass + hairspray so I am interested to see how that reacts, but I need to design a mount for the sensor first.

Thanks for your insight!

If it still saturates before reaching crossover with a 1K resistor, try an even lower value for R1. The collector current rating of the phototransistor is 80mA, so even as low as 100 ohms would still be safe. Alternatively, change the LED series resistors to 1.3K and 1.5K.

---

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].