Do some filaments shrink?

Hi I thought I had my Ormerod set up correctly ( at long last ), however when I printed the replacement gears provided by Matt I discovered the motor shaft hole was far too small.
I did print a set a few weeks ago using the pla supplied with the kit and they fitted well. I just wanted a different color so I am using another pla and it looks as if the items are slightly under size.

As a test, I printed off Matts thin wall test and the height measures 69.93mm, Y width 40.16mm, X 40.17mm - Matt is this correct?

Is there any other prints I could do to ascertain if my set up, including compensation is correct?

I put a lot of my problems down to an unstable bed which I have now fixed by using the heat spreader sheet as the bed base.

So either my setting are wrong or the pla shrinks.

Any thoughts?
Paul

Hi Paul,

I have also noticed alot of sizes come up short, I'm not sure if its caused by shrinkage or by print inaccuracies.

I drew up some simple calibrations rings 30mm, 20mm and 10mm that fit inside each other. the results were as follows.

30mm ring : outside = 29.75mm - inside = 19.69mm

20mm ring : outside = 19.69 - inside = 9.80mm

10mm ring : outside = 9.81mm - inside = 1.76mm

This test was with transparent filament, clearly the sizes are coming up smaller for me, i usually compensate but upscaling in slic3r by 1% if its a part that is printed to fit with others.
i'll do the same with another colour soon.

Matt

---

Limited Edition Red RS Ormerod 1 #144 of 200 - RRP 1.09fw
iamburnys Ormerod Upgrades Github
Follow me on ThingiVerse My Designs

Thank you Matt, I'll print the rings and see how the measurements compare to yours.
Paul

Hi Matt,
Well My measurements are very similar to yours:-

The height is 10.28

30mm ring : outside = 29.72mm - inside = 19.52mm

20mm ring : outside = 19.63 - inside = 9.66mm

10mm ring : outside = 9.82mm - inside = 1.6mm

I am going to try with my original white pla as supplied with the kit and see how that reacts.

I'm a little confused if I upscale in slic3r by 1% will that make the outside and inside diameters larger as well as an increase in height? Or will the inside hole be reduced by 1%?

When I did the Orthogonal Axis Compensation, the last instruction says:-
"Finally, measure the distance between the tip of the projection near the angled corner and the centre of the end of the screw. This should be 78mm."
Over the weeks I have printed 3 of these sets and they all measured 74-76mm, 2-4mm short, this seems excessive and made me think that I was measuring it wrong but the instructions are clear.
Can you remember what your distance was?

This is so frustrating, I just want to print accurately and get on learning openScad
Paul

Quote
appjaws1
Over the weeks I have printed 3 of these sets and they all measured 74-76mm, 2-4mm short, this seems excessive and made me think that I was measuring it wrong but the instructions are clear.
Can you remember what your distance was?

markbee and myself got around 75 as well
http://forums.reprap.org/read.php?340,285560,286668#msg-286668

Thanks for that, perhaps things aren't as bad as I thought. So back to trying to solve the shrinkage problem.
Paul

The firmware values for steps/mm are preset to 87.4890 for X and Y, and 4000.0 for Z. The X and Y values appear to me to be correct if the MXL belt has its nominal tooth pitch of 2.032mm (0.080"). However, I don't know what the tolerance on the MXL belts is, nor at what tension the nominal tooth pitch is quoted and whether that tension is achieved in the Ormerod (especially in the Y axis if you haven't added belt fixing or tensioning components). So I suggest doing a simple check of carriage movement against a ruler. The steps/mm can be adjusted using the M92 command.

It's also likely that the PLA is shrinking a little when the bed cools down. Printing with a lower bed temperature, or on cold blue tape, may be worth trying. Otherwise, adjust for the shrinkage either in slic3r or using the M92 command.

---

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].

My experience is that all my PLA filament supplies shrink. I have found that inner diameter shrink more than outer diameters. I compensate in the CAD files but agrees it is not ideal.
I have had exceptional results fitting large splined shafts together (see post "In defence of Ormerod") with very tight fits but the actual dimensions were never as expected and the reason for the perfect fit was down to the way the fit was CAD drawn in the first place (using affinity rather than relying on absolute sizes).

I suspect the "problem" may be due to extrusion width, default slic3r settings are very fat and it your bed is not perfectly level the extrusion gets squashed too much/too wide. With custom settings (160% for perimeter and 90% for top infill) I get better results, but it takes longer.

Thank DC42 and arnaud31,
I have tried lowering the temperatures but I then get problems with the filament not sticking to the bed. I am currently printing a ring test at 189 hot end and 60 bed. I had to start off with the bed at 65 to get the filament to stick.
I did the ruler check and it looks as if I could be a little short in the Y direction i.e. 0.5mm over 100 mm length.
DC42 how do I use the M92 code to try and correct shrinkage and would I put the command in the config file?
Also what parameters would you change in slic3r to compensate for shrinkage?

I've adjusted perimeter and top fill as suggested by arnaud31 so I'll see if that helps on my next print.
Paul

Hi dc42,
I have just completed a print with 100mm length at right angles. In the x direction it was spot on but in the Y direction it was 99.89mm.
I know this is a small amount but how would I compensate for it.

I have also printed the 3 rings in the original pla and they do not fit together the dimensions are:-

Original PLA
The height is 9.96

30mm ring : outside = 29.76mm - inside = 19.29mm

20mm ring : outside = 19.75 - inside = 9.5mm

10mm ring : outside = 9.79mm

With the new PLA

The height is 10.28

30mm ring : outside = 29.72mm - inside = 19.52mm

20mm ring : outside = 19.63 - inside = 9.66mm

10mm ring : outside = 9.82mm

How do I compensate for this?

Thanks
Paul

If the x-dimension was spot on but the y-dimension was short, this suggest to me that you have the correct tension in the x-belt but that the y-belt is a little slack. Have you printed any of the parts described in other threads for securely fixing and/or tensioning the y-belt?

If you want to correct the y-distance in software, then you can send M92 Y## before printing, where ## is the new number of steps per mm. Try a value of 87.4890 * (100/99.89).

---

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].

Thank you,
I have retightened the belt and will apply the correction and see if that helps.

How do I solve the 3 rings problem?
Paul

I guess the 3 rings problem is due to shrinkage of the PLA. You could upscale the whole part in slic3r (which will increase x, y and z by the same proportion), or use M92 to adjust the steps/mm. The command "M92 X87.4890 Y87.4890 Z4000.0" is equivalent to the default settings. You can increase or decrease the X, Y and Z values in the proportions that you want to grow or shrink the print.

---

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].

Allow me to share my experiences.

I also experienced strange effect when it comes to inner diameters so I decided to consult a friend of mine who is working in plastic moulding and extrusion on professional basis. He informed me with the following info:

Plastic behaves strange unexpected when you move from melted towards solid phase and in the professional process of extrusion there are lots of parameters which we can control only limited with the Omron, and in lots of situations not al all. One parameter is fixed which is the type of material but unfortunately my friend does not use PLA haha.

The way of settling plastic after extrusion has to do with various parameters like
Pressure in which it is extruded
Temperature at which it is extruded
The speed of cooling down
The type of plastic
etc. etc.

I thought I had nice parameters in slice but they appeared to be only for one single item. Other shapes did not fit as I expected so I decided to do intensive testing. I designed a test piece with holes of 20,12,10,8,6 and 4 mm and added various points for measuring. You can see a drawing of the test piece in enclosed pdf fileInnerDiameterTests.pdf, STL is attached

I checked the positioning of Slicr for straight lines and curved lines and according to the G files it positions perfect at a distance of 0.25 mm from the target (using a 0.5 mm nozzle).

During all the tests I did you see that straight lines appear to be pretty accurate, round shapes get bigger errors.
the error for smaller holes becomes bigger.

I tested temperatures of the bed from 20 to 90 degrees in steps of 10 degrees. Found that the advised temperature of 62 degrees is fine. Too cold will force curling of the samples, higher shows no improvement.

I tested single parameters like
fill density: no significant influence
Single vertical shell: no significant influence
Lower extrusion temperature: no significant influence
Low layer height (0.1 in stead of 0.25 mm): slightly worse results and rough top surface

To check relative error I did some tests on bigger prints and found that there is a (limited) relative error which can be compensated with the pulses/mm parameters. But there is an absolute error which is difficult to solve, this absolute error forces bigger difficulties for smaller holes.

My final test for now was to set the extrusion width for perimeters to 0.64mm. This generated an improvement, not perfect but closer towards the targets. Disadvantage was that the top-layer showed a few small gaps.

My conclusions so far:
Temperature of the bed shows no relevant influence, too low temperature forces curling, advised temperature of 62 degrees looks to be OK.
The only parameter which has a positive influence looks to be the extrusion width for perimeters but the small gaps in the top surface are disappointing.

The only real solution so far is as followed:
Correct the relative error (will vary per machine)
Measure the absolute errors as a function of it’s diameter
Adjust your design and correct the inner diameters.
Be aware that this calibration will differ according the used filament and individual printers

No time today for these tests.

What would be helpfull in Slicr would be an option for an absolute correction of inner perimeters

Notes:
Vertical holes show the errors, horizontal holes look to be much better as I noticed after printing a manipulated dice :-)

I printed directly on glass with a PVA glue layer
Bed is mechamical levelled, no auto levelling
Fan is modified so air is taken from above

For measuring I used a digital scale from Mitutoyo. But measuring internal diameters will never be accurate specially for smaller holes.
The samples could show some roughness
Therefore measurements are reported in 1/10 of mm’s, smaller is in my opinion of no sense for inner diameters.

What I did not test was the cooling! Amount of cooling could be of influence so limiting the air flow could be of interest although I have no high expectations.

For measured data and parameters check enclosed pdf file. InnerDiameterTests.pdf

Hope the info helps a bit, and if there are any comments then don’t hesitate to react :-)

Have fun (I have)

That is good information, Marc, thanks for the work. I've noticed that I have to empirically change hole sizes in my designs until the part comes out with the correct size hole if the size is critical. I've found it requires a different correction depending upon whether the hole is in a thin flat plate section of the design or down the centre of a column (i.e. the ID of a tube).

I recall using a hole diameter measuring tool many years ago - it consisted of a conical piece of metal with an engraved spiral scale marked in hole diameter from point to base. You pushed it through the hole as far as it would go, and read off the diameter from the scale. Not having such a tool, I usually use drill bits to find the diameter of very small holes that I cannot easily measure with inside callipers - just find the biggest drill that fits in the hole and estimate how much side-to-side play remains around the shank to add to the drill size.

Dave
(#106)

Real good information indeed, Marc. But can you clarify some points?
Do I understand correctly, that before any adjustments, both your the inner diameters and the outer sizes were too small (what you call the "relative error")? Then, after adjusting the pulses/mm, he outer sizes were fine, but the inner diameters still too small (what you call the absolute error)? I would expect that to be possible to be compensated with the extrusion width.
You say that changing the extrusion width for perimeters to 0.62 (from, I assume, 0.5) improved the situation, that is, it actually made the inner diameter bigger? I don't really understand how that works, I would expect that enlarging the extrusion width would make the inner diameters smaller.
Anyhow, have you tried with overcompensating the pulses/mm, and then changing the perimeter width more, in order to get both correct?

Edited 2 time(s). Last edit at 02/21/2014 02:11PM by Flyskyhy.

Quote
Flyskyhy
You say that changing the extrusion width for perimeters to 0.62 (from, I assume, 0.5) improved the situation, that is, it actually made the inner diameter bigger? I don't really understand how that works, I would expect that enlarging the extrusion width would make the inner diameters smaller.

Not necessarily, because Slic3r will compensate for the increased width by laying down a larger diameter circle. It may indicate that there is less percentage of error in a fatter width (which would make sense if any error is a fairly fixed amount - e.g. a 0.1mm error on a 0.4mm width is 25%, but the same 0.1mm error on a 0.5mm width is only 20%).

Dave
(#106)

@Dave
Differences in a thin hole or a tube is explainable. The first layer always expands a little. It flows more then the next layers causing a wider surface. In my testpiece I forced the first layer 0.5 mm smaller to prevent this effect and nevertheless the inner diameters are smaller compared to higher layers (measured as W3B ). I would advise a deburr tool for thin layers.
I also did a test with a tube resizing in different diameters to see what the influence would be on the inside but I could see no difference (you can see it in the photo with all the other test plastics).

Measuring the inside diameter with the conical tool would nice yes, as long as the inside wall has a nice round surface which unfortunately is not always the case for the smaller diameters.

@Flyskyhy The outside diameters were OK for my specs. You see a small error for larger sizes but I think that error is relative to the dimension. The error in the inner diameters becomes bigger when the size becomes smaller.
During the test I did not do any changes in other parameters then described so I did not change the calibration in pulses/mm during the test.
Concerning the extruder width, like dmound explained. larger extruder size forces the head to make a bigger circle. I don't know what Slicr is doing with the amount of plastic. In theory it could increase the extrusion speed but since you would like to see it as a correction, you hope is doesn't.


I had my thoughts about the problem and came to a relative simple theory.
The extruder head is round with a diameter of 0.5mm. When the head makes a circle it will dose on the inside the same amount of plastic as in the outside. However, the inner circle describes a shorter distance then the outer circle so you get relative more plastic on the inside compared to the outside.
The difference will be larger if the diameter is smaller and that's exactly what we see in the measurement. Smaller holes have bigger errors,

Question remains: how to correct the error? Enlarging the extrusion head will compensate some of the error but in theory it will have a negative effect on the outside diameter. In theory, Slicr should be able to calculate the correction based on the radius of a curve, whether it is an inside or outside perimeter and the size of the extruder.

Any other ideas?

Gr Marc (who is busy designing a 1.75mm filament extruder to regain all his test material)

Quote
dmould

Quote
Flyskyhy
You say that changing the extrusion width for perimeters to 0.62 (from, I assume, 0.5) improved the situation, that is, it actually made the inner diameter bigger? I don't really understand how that works, I would expect that enlarging the extrusion width would make the inner diameters smaller.

Not necessarily, because Slic3r will compensate for the increased width by laying down a larger diameter circle. It may indicate that there is less percentage of error in a fatter width (which would make sense if any error is a fairly fixed amount - e.g. a 0.1mm error on a 0.4mm width is 25%, but the same 0.1mm error on a 0.5mm width is only 20%).

Dave
(#106)

An extrusion width of 0.62 is smaller than the default slic3r produces (0.85mm for a layer of 0.24mm) if you leave the value to 0.
As mentioned earlier, i use 160% (of layer height) because slic3r extrudes very "fat" and get better results.

Hi Arnoud,

thanks for your reply.
Could you explain how Slicr would produce an extrudor width of 0.85 at a layer height of 0.25mm?
I checked the G-files and saw that the extruder (set at 0.5mm) during circles was positioned exactly 0.25mm away from the perimeters.

Other question, with custom setting 160% for perimiters I would expect a higher flow of plastic in stead of less or?

Quote
Marc van Beelen
Hi Arnoud,

thanks for your reply.
Could you explain how Slicr would produce an extrudor width of 0.85 at a layer height of 0.25mm?
I checked the G-files and saw that the extruder (set at 0.5mm) during circles was positioned exactly 0.25mm away from the perimeters.

Other question, with custom setting 160% for perimiters I would expect a higher flow of plastic in stead of less or?

I would presume that a wide extrusion is simply produced by flowing more filament through the nozzle. With a layer height of 0.24mm, there is theoretically 0.24mm gap with previously extruded layer, the more filament you extrude the more gets squashed and squeezed sideways and therefore give you a wide extrusion, this must all be built into slic3r and computed based on nozzle and filament diameters so that slic3r can position the nozzle in the correct place.

You can read the extrusion width of a given g-file (at the start of the file). Using default (set to 0) values gives 0.85mm for a 0.24mm layer or approx. 350%.
The 160% value I set gives me width of 0.36mm (again you can check the g-file but you will also notice it by looking at the print) so in this case this case it would flow at less than half the default flow rate, not a higher flow rate.

I guess what is unclear is how can you extrude 0.36mm through a 0.5mm nozzle? Without knowing much at all about the physics of PLA extrusion, I would expect PLA to behave like any fluid flowing through a restriction, the cross section of the flow can contract due to frictional losses immediately after the restriction (nozzle).

Edit: remember that 160% means 160% of layer height not 160% of default. As pointed above, default seems to be in the region of 350% of layer height.

Edited 1 time(s). Last edit at 02/21/2014 05:27PM by arnaud31.

OK, I understand what you mean since you asume that Slicr adjusts the feed rate to create a bigger extruder hole.

In my test I mention it and I simply don't know what Slicr does with the feedrate when wider extrudor is defined.
What I do know that defining a wider extruder gives better results for the smaller holes
Also I know that with standard extruder width of 0.5, layer height 0.25mm, Slicr positions the head exactly 0.25 mm from the borders (of a round object) so I think that Slicr simply adjusts the extruder speed depending on layer height but positions the head, depending on extruder width (but no idea if it also sets a higher feedrate in case of wider extrudor head).

But to be sure I can do a simple test with the adjusted and tested 0.64 setting and see where Slicr positions it's head and compare the feedrates.

Anyhow, thanks for the info so far, I will keep you updated.

Quote
Marc van Beelen
OK, I understand what you mean since you asume that Slicr adjusts the feed rate to create a bigger extruder hole.

In my test I mention it and I simply don't know what Slicr does with the feedrate when wider extrudor is defined.
What I do know that defining a wider extruder gives better results for the smaller holes
Also I know that with standard extruder width of 0.5, layer height 0.25mm, Slicr positions the head exactly 0.25 mm from the borders (of a round object) so I think that Slicr simply adjusts the extruder speed depending on layer height but positions the head, depending on extruder width (but no idea if it also sets a higher feedrate in case of wider extrudor head).

But to be sure I can do a simple test with the adjusted and tested 0.64 setting and see where Slicr positions it's head and compare the feedrates.

Anyhow, thanks for the info so far, I will keep you updated.

I am away from printer at moment for another few days so can't try. But I would be interested to know if you notice a proportional offset of the extruder after setting a wider extrusion. I cannot see any other way to achieve wide extrusions than for slic3r to feed more wire through. You could slice the same part with 2 different extrusion width but the same fill ratio, then if you look at the very bottom of the g-file it tells you how much filament will be used (in length and volume), there should be a difference reflecting the over or under-extrusion... will check when I have a moment.

I looked into the Slic3r code to see what it actually does. If I understand it correctly, then it takes the width directly from the settings, specifically from the Print Settings->Advanced->Extrusion Width->Perimeter parameter. The line of plastic is deposited half that width from the edge. From the width, the extrusion speed is calculated, according to the following:
- if width > nozzle_diameter + layer_height, then the plastic is assumed to form a rectangle with two half-circles to the sides, squeezed sideway between the nozzle and the lower layer. It extrudes enough plastic to fill such a shape to the specified width.
- otherwise, the cross section is assumed to be a rectangle with the width of the nozzle and height of the layer, with extra material going into two half-ovals to the side. Enough is extruded to create a shape like that with the specified width.

Now, this has interesting consequences:
- The width cannot be smaller than the nozzle diameter, the logic does not account for that situation (I am not sure what would happen in practice anyway; is it possible to lay down lines thinner than the nozzle diameter?)
- If the width is exactly equal to the nozzle diameter, then the logic assumes an exactly rectangle line of plastic is deposited - not very realistic.

This becomes especially strange, because of the way the width is determined from the setting:
- if the setting is a percentage, it is a percentage of the layer height, NOT of the nozzle diameter.
- if the setting is 0 (the default), then it takes the nozzle diameter as width - again, not very realistic.

To give a concrete example: with a nozzle diameeter of 0.5mm, and a layer height of 0.25mm, if you set the perimeter width to 160%, it will actually be 0.4mm, smaller than the nozzle diameter! Therefor, you will not get accurate walls with those settings

In concrete, I would recommend to always set the perimeter extrusion width equal to the nozzle diameter + the layer height. That way, I think you should get a realistic computation and extrusion, and therefore more accurate walls. Slic3r should probably be updated to take that as default.

As soon as I get a chance, I will try to experiment with this a bit, too see what happens in practice with different parameters

Quote
Flyskyhy
I looked into the Slic3r code to see what it actually does. If I understand it correctly, then it takes the width directly from the settings, specifically from the Print Settings->Advanced->Extrusion Width->Perimeter parameter. The line of plastic is deposited half that width from the edge. From the width, the extrusion speed is calculated, according to the following:
- if width > nozzle_diameter + layer_height, then the plastic is assumed to form a rectangle with two half-circles to the sides, squeezed sideway between the nozzle and the lower layer. It extrudes enough plastic to fill such a shape to the specified width.
- otherwise, the cross section is assumed to be a rectangle with the width of the nozzle and height of the layer, with extra material going into two half-ovals to the side. Enough is extruded to create a shape like that with the specified width.

Now, this has interesting consequences:
- The width cannot be smaller than the nozzle diameter, the logic does not account for that situation (I am not sure what would happen in practice anyway; is it possible to lay down lines thinner than the nozzle diameter?)
- If the width is exactly equal to the nozzle diameter, then the logic assumes an exactly rectangle line of plastic is deposited - not very realistic.

This becomes especially strange, because of the way the width is determined from the setting:
- if the setting is a percentage, it is a percentage of the layer height, NOT of the nozzle diameter.
- if the setting is 0 (the default), then it takes the nozzle diameter as width - again, not very realistic.

To give a concrete example: with a nozzle diameeter of 0.5mm, and a layer height of 0.25mm, if you set the perimeter width to 160%, it will actually be 0.4mm, smaller than the nozzle diameter! Therefor, you will not get accurate walls with those settings

In concrete, I would recommend to always set the perimeter extrusion width equal to the nozzle diameter + the layer height. That way, I think you should get a realistic computation and extrusion, and therefore more accurate walls. Slic3r should probably be updated to take that as default.

As soon as I get a chance, I will try to experiment with this a bit, too see what happens in practice with different parameters

I have been printed for a while with a width of 0.36mm (layer height of 0.24mm and 160%) and have found the wall to be better. I do not know however if the wall are 0.36mm wide (should measure it really). Also with default settings (0) I get a width of 0.85mm in the code. It is strange that you get you get nozzle diameter (0.5mm) as width by default when I get 0.85mm by default?

Thanks for the feedback Flyskyhy

You could be in the right direction, i checked produced G code from Sli3r,
Defined two nozzle wdth (1 standard 0.50, the second 1.00)
First layers were the same, second showed the difference. Nozzle was moved 0.25mm more from the sidewalls, flow of extrusion was 89% higher (not 100%). Layerheight was defined as 0.25mm
In both situation for the second layer the nozzle was positions half the nozzle width from the side.

Unfortunatly I have no time for further experiments since I will be busy for a few weeks abroad but if someone able to do futher testing then please do.

good luck!

@Arnaud31, I don't understand either why your perimiter width is set at 0.85. It will be a combination of various parameters. In my situation it took over the value as defined of in case of default the diameter of the nozzle.

Quote
Marc van Beelen
@Arnaud31, I don't understand either why your perimiter width is set at 0.85. It will be a combination of various parameters. In my situation it took over the value as defined of in case of default the diameter of the nozzle.

I found why we have different default width. I just sliced coat hook with old and new version of slic3r (I run both because I find latest version of slic3r to give worse prints than old version) and I get different default width (see attached g-code). Note also different length of filament used (last line of files);

Both slicing based on rectilinear infill at 0.3 density.

After advice of Flyskyhy I set the extrusion width to nozzle diameter + layer height and get much better results for the inner diameters. Outer dimensions are smaller! :
Layer height, 0.25mm,
Nozzle Diameter 0.5mm
Extrusion width perimiters 0.75mm

Measurements on my testpiece:
Target round 20mm inner, measured 19.9mm
Target round 12mm inner, measured 11.9mm
Target round 10mm inner, measured 9.9mm
Target round 8mm inner, measured 7.9mm
Target round 6mm inner, measured 5.9mm
Target round 4mm inner, measured 3.9mm

All gaps less then 0.1mm of target!

But outside dimensions are all measured below spec
Target straight 94mm outside, measured 93.6 mm
Target straight 30mm outside, measured 29.6 mm
Target straight 19mm outside, measured 19.9 mm
Target straight 15mm outside, measured 14.9 mm

The top layer is also not fully closed but that's something that could be fixed I guess.

Update for Slic3r with seperate custom setting for inside/outside perimiter would be nice :-)

This was my last test, sorry I have other priorities now, feel free to test check and update.

So, I guess the only thing to do is to design items with larger holes.
The problem with that is deciding how much bigger the hole should be because the hole error gets larger as the hole size reduces.
Would their be a formula that could be applied?
I only want to affect small hole sizes e.g. bolt holes of 3,4,5 and 6mm diameter. So I don't want to introduce more errors in other parts of the print by altering the steps in X,Y and Z directions or by applying an overall increase in % size of print.

Anyway thank you all for the efforts made and the suggestions so far.

---

appjaws - Core XYUV Duet Ethernet Duex5
firmware 3.1.1 Web Interface 3.1.1
Ormerod 1-converted to laser engraver, Duet wifi
OpenSCAD version 2020.07
slic3r-1.3.0, Simplify3D 4.1.2, Cura-4.4.1