Studio Contrechoc

design & textile & technology entries

Monthly Archives: October 2014

Knitting around a corner

The knitting produces a flat surface in principle, you can make a knitted 3D shape because you can fit shaped parts together in making just like a dress with woven cloth. Knitted surfaces fit more easily because of the stretch property of the knitting.

The 2D knitting surface is just a rectangle of knitting needles and tours, width x height.  Thus 2D property of the surface is just the most basic possibility using B and D positions of the needles. Using the E position of the needle and the H mode of the slider you can vary differentially the number of knitting needles over the surface and create in the surface itself tension. This tension makes 3D shapes inside the knitted surface, which can create pockets or bends. WIth this technique gloves are made, the boxes of the last post.

2014-10-30 10.08.11  2014-10-30 10.08.00

The E position in action on the knitting board:

2014-10-30 21.23.31

The threads comes back behind the slider over all needles, but before returning you have to put the thread under the last needle. Slider at far left, will be coming to the right.

2014-10-30 21.23.15

To diminish the needles in E position you have to put this needle in D position. Slider is at far right will be going to the left and put this D needle back in B position.

2014-10-30 21.23.01

Differentially knitted tours, so you have to keep pulling the knitting down, replacing the weights all the time.

2014-10-30 10.08.25

Two examples where you start knitting tricot normally, then put all the needles in E psoition and work your way back needle by needle to the B position. You see the knitting makes a angle, start of this knitting was right side.

2014-10-31 08.16.12

The same with yellow lines indicating 10 tours, this knitting was started at left, for 20 tours then you can follow from the corner middle left upwards to the right.

2014-10-30 10.08.30

This knitting is made making 10 tours of normal tricot, left side, now deformed into an angle. Then put all the needles in E position, but for the center 2 needles. Then every tour putting the needle oposite to the slider in D position. The extra tours are starting small in the middle and growing wider as more needles are knitting again. This shape as shown  in the picture is totally stretched, normally a 3D shape.



Learning to make a thumb-box: 親指ックス手袋の編み方

This title part in Japanese is no doubt totally wrong 🙂

I saw this youtube movie: and tried it out, only not for my sock, which is indicated in the title ( 足袋ックスつま先の編み方, means the way (方) to knit (編み) toe (つま先) boxes (袋ックス) for foot (足)). I just made this box thing and it turned out to be one third of a glove. I made the Japanse title up for this knitted thumb box.

Although i am interested in, and aiming at datavisualisation in knitting, the way to improve technical skills is to do simple exercises. It is clear to me that this knitting device is far more than a “mirical knitting printer”, with one button and pushing this button makes a perfect copy.

So for the material of this post no Arduino Due was used at all, the tricks were in using hand and braincells, closely observing seemingly unimportant details, which naturally turned out to be very important…

A few lessons were learnt. First that I started with thread which was too thin. It broke several times:



Not even all the disasters are shown. The good thing of knitting is that you can recycle the thread after a knitting disaster. The idea that you only waste (called learning)  time and not material (eg in 3D printing you cannot put the PLA back) is rewarding, even if the thread costs nearly nothing.

Then that in the H mode ( needles totally sticking out are kept on hold) you have to start with putting the needles in the E position at the other side of the slider, then go back (otherwise your thread is around this bunch of E needles)


Then after a few time with the thin yarn in the end knitting with a thicker thread,


all steps were taken, and although this was just a test, it nearly became a glove…


Some holes at the base are still a bit too big….

Practice again…..


  • Set up a normal tricot , 30 needles, 10 tours
  • from -8 to 8, insert thread to keep track of the start of the box
  • from left, put rest of needles at right in E position
  • after that from right the rest of the needles at left in E position
  • knit middle part 20 tours, carefully hanging a weight in this part
  • now from right, one needle -7 to E, knit, from left thread over this needle, at right needle 7 to E, knit
  • repeat until 4 needles at both side are in E
  • then retreat: from right put needle -3 back again (still winding the thread around the needle you come from), the same from left, till all the needles (-8 – +8)are back in B position.
  • Then from left, pick up at the left side of the box knit stitch number 3 (or 4?), put it back on needle +7, knit, the same at the other side.
  • You are working your way back up, closing the box.
  • If you do everything regularly, you won’t end up with the holes like me 🙂
  • Look at the Japanese video for getting the idea.




Practicing my Japanese, knitting katakana

The Hiragana writing is easier for me to read than the Katakana…Hiragana is met more often – as a beginner – and katakana is like isolated islands in the sea of Japanese. Added to that is the transformation of mostly English words into the Japanese syllable format. For me always a bit like a puzzle – in these words English is taken rather literally from the sound, not the writing – it’s punishing English for swallowing letters :-). Beside that, for me also always the r/l uncertainty… But slowly I manage…and than suddenly it helps when you are in a language area where the machines are Japanese, Brother, Silver Reed…..more reasons to practice.

So i am hacking the titles out of Japanese Youtube movies about knitting, scanning this Machinelabo site, making a Katakana list, improving again a little bit my Katakana reading fluency:-) – i’ll never be able to pronounce this in Japanese – the same thing with my pronunciation of Frenchified English words in French, just impossible to get this English right in this other than English language transformation…Isn’t language BIG fun!!!

デザイナ・ニット = Designa Knit. the software program for interaction with the Brother Knitting Machine
ブラザ = Brother – the brother knitting machine : ブラザー機 , (Brother has become something like “BU-RA-ZA”)
シルバー = Silver – Silver Reed knitting machine : シルバー機, (SHI – RI – VA – A, the -r/l thing)
ニット = knit, N(I)-ji -TO
パンチカード = punch card, PA – N – TSJI – KA – A – TO
テンション = tension
リーダー = reader, ニットリーダー = knitreader
パラフィン = parafin
ニットキャンバス = knit canvas
カウンター = counter

プレーン = plain
レース = lace
スレッドレース = thread lace
スカラップ = scallop, knitting in shell shapes, (i have to find this out)
ルームソックス = room socks
プレーンソックス = plain socks
カカト= heel(s)

and then there are some Kanji that stick out:
編み = amu, to knit
機 = ki, machine
編み機 = amiki, knitting machine
ゴム編み = ribber
模様 = mojou – pattern
ス模様インプット= su moyō inputto, – scan input pattern
靴下 = kusjta – sock, or socks

general word list:
サイト = site
マニュアル = manual
ファイル = file
ページ = page
コンプリート = complete
コメント = comment
ツール = tool, but also “tour” – knitting line ?
スタンダード = standard eg standard version.
プロ = pro, eg pro version of software
コンタクト = contact
メールアドレス = mailaddress
メッセージ = message
ドレッシング = drssing
インスタントラーメン = standard ramen

Now this is a nice one:
マイペース = mypace, doing things in my own pace

Words that are in Katakana, but Japanese…
ビックリする, bikuri suru, to be surprised

Captura de pantalla 2014-10-21 a la(s) 09.27.27

Most of these words I found at: and checked if in doubt with Rikachan plugin and in Google Translate.

The mysterious KH-940 weave knit

One of the knitting techniques that was omitted from the list we experimented with uptil now (the skip, patent, nordic, lace and jacquard) was the weave stitch. Experimenting a bit with the weave knit was giving puzzling results. Most of the time this is frustrating, because it means this is going to cost time. On the other hand, if something costs time, it also means that there is something to discover…

I discovered that moving around at either side of the knitting on the knitting machine caused a needle count error, this is fixed now in the software:

(This still is the very basic software. No fancy stuff for an image moving in the Processing sketch when knitting a tour. If the image is too big the underside can even be out of sight. I am wondering about the possibility of having a choice between the different stitches, then give a simple analysis of the image, if this is the right image for this stitch…-possibility to get the inverse -negative – image or the mirror image. Even have an indication if the distribution of the pixels is fit for this stitch.)

The weave stitch is using the entrance of the slider, without being inserted in the first or second thread opening in the middle of the slider:

2014-10-19 09.23.27

The base thread (in the middle opening) has to be very thin and the tension is set to 1-2. The second thread has to be thicker yarn.

Because of the tension setting the produced knitted surface is about half of the image knitted: so starting with the normal dimensions

Captura de pantalla 2014-10-19 a la(s) 10.50.17  The white pixel is showing the thick yarn at the backside.

you get this knitting:

2014-10-19 10.47.11

This is the back of the knitting, that is why the letters in the image are mirrored. For this experiment, we used the MPPT graph again. Apperently, for the weave stitch the pattern of the filling is important. This has to be prepared in Photoshop using the (self defined) preset brushes and patterns.

Then the next step is to enlarge the image height. The graph has to be stretched in the image, to have it normally scaled in the knitting. You cannot use “free transform”, because this would destroy the fill pattern of the surface (although this sometimes gives ideas too). The rescaling has to be done by hand,

Captura de pantalla 2014-10-19 a la(s) 10.48.48

Also we varied the fill pattern of the surface. You already see that it is difficult now to keep the graph in the right shape.

This resulted in knitted weave images were not convincing, the graph appears, but only dim, the weave pattern inside the thin thread is not crisp: See here back (left) and front (right) side:


During knitting the thick yarn was changed three time, to see which would work best. Vertically up the thicker yarn was yellow, middle we used two middle thickness yarns – yellow and light pink- and below the thick plastic pink yarn also used for the pollution scarf.



Here the thicker yellow yarn and the double threads is displayed.

testing further:

Again changing the fill and also making the graph more clear, taking the brush of example 426 from Stitch World Pattern book: x00x00x.

Captura de pantalla 2014-10-19 a la(s) 10.48.59

Also changing the basic thin wire to even thinner (shining blue very thin yarn).

Starting of with the thick plastic pink yarn we wondered about the fact that the knitter was not at all complaining about the thickness of the pink plastic yarn, while the tension was set to 1. Normally for this yarn the tension has to be 9-10. Ok then try even thicker yarn, the bright yellow which we also used for the presentation “Yellow Things”. This was knitted on a knitter with much bigger needles and distance between the needles.

There was no problem knitting this thicker yarn and the image got out very clear now:

2014-10-19 10.33.04  front 2014-10-19 10.28.32 back (on the knitter)

This is the fron side of the image, where the thicker yarn is beautifully overlayered by a mesh of the blue thinny yarn.


2014-10-19 10.33.36 2014-10-19 10.33.41  2014-10-19 10.33.46  2014-10-19 10.33.10,

Details of the fill, the mesh formed by the thin blue yarn, the way the graph is showing in the texture. You can see how the thick thread is catched inside the mesh of the knitting.

This weave pattern begs for more experiments!

Finishing with the cones of the yarn side by side:

2014-10-19 11.31.10

When i looked back i realized the maybe it is not really the thickness but the relative textures of the threads…more experiments are needed.


Weaving the same image with a very thin plastic fiber. The ordinary yellow wool seems to be caught inside an invisible plastic mesh:

2014-10-19 15.43.41

2014-10-19 15.43.47

2014-10-19 15.53.12 2014-10-19 15.53.48

some details where the plastic mesh can be seen, and the side with the arched wires.

2014-10-19 15.44.59  the plastic thread used.






Back to knit stitch basics: Comparing stitches

On the one side we already started experiment with real data on the other hand this is still quite a lot of properties to discover about the different stitches. So we went back to basics with a very simple basic image: (Rectangular image – built from black and white squares)


We  made a test knit in different stitches, Nordic, Skip stitch, Patent stitch, lace stitch and Jacquard. The differences were quite amazing. The basic image helped to filter out many interesting characteristics.

Nordic problem case:

2014-10-15 19.39.15 2014-10-15 19.39.23

2014-10-15 19.39.29 2014-10-15 19.40.01

2014-10-15 19.40.20

The image above done in Nordic double thread technique (on one bed) becomes a mess, because the very vertical seperation of the big squares. When the image is not under tension anymore it seems to have holes in the structure.

bw_nordic_diagonal  2014-10-15 19.37.17#1

Adding the diagonal is a normal trick in Nordic knitting, the purpose is to reduce the long free threads at the back, and it can fix the problem of the structure falling apart. Of course this reduces the bigger shapes, there are always lines in front.

Also the other stitches are asking for a different translations of this basic image, filling in either the white or the black squares. The fill pattern of course can be varied more.


This image above is used for Jacquard stitch, also using two threads:

2014-10-15 10.28.01 2014-10-15 10.28.13

2014-10-15 19.42.55  2014-10-15 19.43.05

The Jacquard knit is dense and feels heavy, like the Middle Ages must have felt 🙂

The patent stitch is working on one wire, with delaying wires to be bound. This can be put to an extreme, where the threads  falls over the needle, which is not done here.

The patent stitch in the square forms a nice hexagon:

2014-10-15 09.48.31 2014-10-15 09.48.43

patent stitch backside: hexagons

2014-10-15 19.41.33 2014-10-15 19.41.56

patent stitch front side: this is purl and knit alternated. The patent is very stretchy and feels light (compared to the Jacquard knit). Remarkable is also the deformation due to different tensions, hich can be seen in the meandering line between the squares.


used for lace and skip stitch.

The skip stitch in squares show very well that the length is reduced, because the squares are under tension and transformed in the tension of the textile at the knitter:

2014-10-15 09.50.16 2014-10-15 09.48.21


2014-10-15 10.16.50 lace under tension

2014-10-15 20.14.53 2014-10-15 20.15.04

without tension, there is a big difference between front and back side, whicj could be enhanced or diminished by the choice of the second – cotrast yarn, depending on color and thickness. (The idea of the lace knit is that this second yarn is very thin.)


In the end all the stitches were done in one longer piece of textile: This has the advantage that at least one of the threads – yellow here –  is kept constant for comparing. The effects are also very dependent on the choice of the yarn. But added to that the way to fill in the pattern parts (either black or white in the pixel image) will further influence the outcome. The fill’s chosen for these basics  are only showing the first step. In the fill’s you can go further in creating relief and structure but this pixels structure has to be done very specific to each stitch to avoid knitting disasters.

2014-10-15 20.31.40


Experimental Astronomical knittings

In experimenting with bigger images and different threads new properties of the different stitches (here Jacquard and lace stitch) become clear. For my way of making images: lace stitch has a white background, Jacquard a black fond. Reverse this, and the threads go wild on the knitter.


First an image of the sun’s disk with the sunspots. It can be found on this marvellous site of the Nasa:

Captura de pantalla 2014-10-14 a la(s) 10.20.59 sunspot_jacquard_100

Captura de pantalla 2014-10-14 a la(s) 08.21.33  2014-10-14 10.57.04

2014-10-14 11.24.32 2014-10-14 11.24.41

2014-10-14 11.24.25  sunspot_jacquard_half

This was knitted in the Jacquard stitch, 100×100. Clearly the Jacquard is knitted very tight and the sun has become an oval. Photoshopping it to a circle is easy, but the knitting stays the same :-). The right image is reality!



Nice is the so called butterfly effect in solar activity, sunspots. Observed from 1874, you can see the suns activity rise till a maximum around 1960, then gradually decline till the present day.

The picture of the sunspots on the suns latitude appeals because it resembles a leaf.

Captura de pantalla 2014-10-14 a la(s) 10.21.57

sunspot                     2014-10-14 09.42.19 2014-10-14 09.42.29

Knitted in the lace stitch, 100 x 200, the purple thread is at the limit of thickness. The small cone is having some bling thread, invisible in the pictures.

2014-10-14 11.23.10 2014-10-14 09.42.38 2014-10-14 11.50.33



Another view of the sun’s acitvity can be found here: it shows the solar wind, which is warped because of the rotation of the sun (like a beacon at see). Added to that you can see a blob of extra particles coming out. The view is from above. (This site is loading slowly….)

This observing the sun has a very practical purpose: when the Earth is inside a very active region computers might fail, airplanes falling from the sky, drones going for outer space….

Captura de pantalla 2014-10-14 a la(s) 08.28.24

Originally this is an animated GIF and you can see the dark blue region spiraling outwards.

sun_eruption_100                      2014-10-14 11.23.30 2014-10-14 11.26.34

Knitted in the lace stitch 100 x 100.


Conclusions about these experiments: with increasing number of needles used, clearly the effect of the open stitches are becoming less significant. Also the different stitches produce different tightness, so the pictures cannot serve without mending size and particularly length (number of tours) for all stitches at the same time.


Towards another Supernova datavisualisation

At the etextile summercamp I presented the t-shirt with the bright led panels performing the explosions of the first six month of Supernova observations – it is the t-shirt with the microwave background radiation image in Mollweide projection. My goal with the knitting was to do this pattern (in the Python image) in one of the knitting techniques.

Image of the positions of the observed supernova’s (star explosions) of the first half year of 2014:


Data from:

My scripts, see python script for getting the data in the Mollweide projection:


This post reports the first step towards that goal of getting this image in thread:

Testing what a pixel is in this lace knit technique and testing how much resolution I can generate.

These are images of the two sides of the same knitting:

mollweide2 mollweide1

The image used for knitting: (70 needles/pixels wide)


Important is the order of the two yarns: thick = base color first thread, thin is contrast thread.

So we can compare the “pixels”:

Which means that we can indeed go for one pixel width.

The knitted oval shape is slightly shorter that the base-image, so the image should be longer to get the same width-height relation.

Of course the knitting has a parallel track in 3D printing:

2014-10-13 12.56.27

With the 3D prints I am prepairing a “print” in the more traditional techniques of blockprint on paper. The inprint on paper without any ink is also a possibility (called blind printing in Dutch).  This block print will be a third parallel track.



KH-940 scripting details

As told in an earlier post,

the software didn’t work with my combination KH-940, Arduino Due and MAC laptop. Their software is probably device dependent, like my code is! The device dependency could be due to the lags in communication between the Arduino, knitter and laptop.

I decided to program my own scripts, thus learning about the sensors and the problems involved.

In this post I publish remarks around the first scripts which can be found at github:

In this post i reproduce the README file, with a few images from the KH-930 manual:

README on Github:

Be careful: the communications are probably device dependent, the knitic code didn’t work for my machines. That’s why i wrote my own. This software could just not work on yours because of the same reason!

The connections/board are in principle the board (but for one PIN):

for needle PIN’s: (my PIN 2 is fried.)
digital: 3,4,5
Solenoids array: digital: 22,24,26,28,30,32,34,36,37,35,33,31,29,27,25,23
right/left solenoid sensors: analog 0,1

Programming concept: all pixels are sent in CHAR’s one pixel is one bit. send as less information as possible beteen the Arduino and Processing



  • step 1: sample image and to be send to Arduino in char code that is 8 bits of information (pixels) in one char. step 1a: add 4 lines in fron of the image – see detail 2
  • step 2: waiting for the arduino signalling start sending image
  • step3: sending image line by line after a signal from the arduino
  • step 4: receiving the control image and displaying this image, waiting for knitting to start


  • step 1: asking for the image after some start up time
  • step 2: asking for img width and height
  • step 3: asking for image line by line (tours) in char code.
  • step 4: sending back a check of the image


Knitting: (starts at right side only at the moment)


  • step 1: reaction on right sensor: send one line to knitter
  • step 2: count stitches from needle sensor information 1,2,3,4 rithm, done by interrupt functions
  • step 3: react om passing the middle
  • step 4: changing direction (deduction from 1,2,3,4 reversal)
  • step 5: send line to processing


  • step 1: receive knitted line, displaying as a transparent overlay in the image, no information of needle count is displayed
  • step 2: sound signal after direction reversal
  • step 3: after the knitting is done: sound signal to indicate end of image

Detail 1:

critical in the Arduino code are the (hard coded delays) eg in the loop to receive the image:

Serial.println("w");///sending code to ask for image width
delay(7);//necessary, otherwise loss of data

change these values and information can be lost

Detail 2:

The first 4 lines transferred contained random errors, from the 4th line on information was processed without errors. This was solved in a rude manner: insert 4 extra lines at the start of the image and ignore these 4 lines in knitting.

Detail 3:

The image is kept in the memory of the Arduino sketch. This was not possible in INTEGER format. In CHAR format it is, but of course the image can in principle have an infinite length (while width has a max of 200). Also the CHAR array has boundaries. At the moment the maximum value is: 200 tours (lines of knitting).

Detail 4:

The KH-940 needle sensor information. With interrupts the needle information is processed. There are two signals alternating for each needle. These were given a code 1,2,3,4 and from this the direction is deduced. When reversing direction counting is done backwards, not losing any needles.

Captura de pantalla 2014-10-11 a la(s) 19.17.21

Detail 5:

The right and left sensor. The right sensor is used not for putting the needle count on 0, but for deducing the distance between the needle 0 and the position wehre the carriage combines with the belt – solenoids offset. By trial and error was deduced that there are 4 significant combining positions, which can be reduced even further, with a special offset. After introducing these belt offsets, the image didn’t shift anymore.

encoder graph

Detail 6:

Does this mean that I exclude knitting direct sensor information? No, because a sensor connected to the Arduino can influence the row to be knitted. But because anyway, you have to correct for direction of knitting when doing eg a graph, you have to program this case individually independent of the background image sent from the laptop.


Doing invisible knitted datavisualisation

With the software i wrote (see last post) and still test for the KH-940 i now explore the possibilities to knit datavisualisations beside the Nordic knitting technique.

The other techniques are: patent stitch, skip stitch (Dutch slip steek), tuck stitch, jacquard, ajour, lace stitch. It involves all kinds of tricks with the needles.You get a relief, or open stitches, or connections between tours which are far apart.

This post presents first experiments, in later posts teh different possibilities will be explored individually and more in detail, what the knit stitch really is, how to code for it etc.

With a GPS track already shown earlier I prepared the images for different knitting techniques. Because you skip needles or keep thread in waiting several tours you cannot do everything you are used to for “simple” Nordic knitting. Also, because you are “torturing” the needles a bit, better start with not too fancy threads. (I always mix colors and thread, but for instance very smooth thread easily slip of needles.)

A few results:

GPS design slip stitch


GPS knit

back of a Nordic knitting

GPS track jacquard prep

Nordic knitting, image prepared for the following skip stitch and patent stitch

slip stitch 4

skip stitch giving a nice relief

2014-10-11 12.50.03

Another skip stitch (two different threads together)

2014-10-11 12.29.22

And a tuck stitch (two different threads together)

2014-10-11 15.35.00

Jacquard, (two different threads in seperate Jacquard piece)

The corresponding image (slip stitch style), where you have to “know” which needle position (B or C) is the black pixel, reverse the needle positions (black and white pixels) and disaster strikes…

Captura de pantalla 2014-10-11 a la(s) 14.29.44

2014-10-10 15.03.29#1

This is the GPS track again, the most interesting part is the PLA stitches around the track!



2014-10-11 19.11.35 2014-10-11 19.10.43

Two sides of the same shape, which is the left side of the next image. This is the needle signal of the knitter, knitted again. The slips are particularly “thick” because some double threads were used.


Image of the sensor code of the KH-940, left the needle sensors, right the solenoid (which should cover 8 needles for half of it.) The next image is from the


which can be downloaded from the internet, explaining how the needle sensors keep track of going left or right. It is a bit like walking with two legs, 1,2,3,4, the long stretch is the solenoid counter.

Captura de pantalla 2014-10-11 a la(s) 19.17.21


The different techniques require different images. In experimenting you can go wrong, but what is wrong?

slip stitch knit slip stitch 2

really very longs slips, indeed this knitting was a disaster – i have to work on the image, get the right number of pixels next to each other.

slip stitch 3

Very long connections because of images that are not really fit for the slip stitch technique.

A view of my floor with all kind of experiments:

2014-10-11 13.59.23

of course if I am knitting, my 3D printer can work as well. With the Blender possibility of making a 3D shape of an image i explore these datavisualisations further, some of the results closely resemble the knittings, here the MPPT graph, in knitting and 3D print:

2014-10-11 12.39.15  2014-10-12 13.23.02

MPPT knit _ nordic (actually mirrored MPPT)

Nordic knitting of the 3D print ( An MPPT graph is the efficiency curve for a solar cell with varying load, for another version:

This is the Jacquard patent knitting for the MPPT graph, from another image:

2014-10-12 13.21.57  Captura de pantalla 2014-10-12 a la(s) 13.25.00

(Since Jacquard is mirroring the image, there should be a button “mirror” depending on the technique used added to the Processing sketch.)

close up:

2014-10-12 14.33.39

The best for the last: (this kind of visualising i was looking for when starting…)

2014-10-12 14.28.20 Captura de pantalla 2014-10-12 a la(s) 14.24.52

lace stitch (not mirrored like Jacquard), which is using two threads, one of which is very thin, creating the open stitches:

2014-10-12 14.33.18

You see that the threads are not hanging free in these area’s like the Nordic knitting. Further experiments, whith different threads and images are very much required!


The next challenge is adding the L carriage. But then a bit of extra programming is required for the sensor value of the L carriage to have effect on the tour count.



Writing my own software for the KH-940

Last year in 2013 I joined the lobotomy workshop of in Bruxelles. We made this knitic board (now called the first version) and did some experiments, which failed using my board on a KH-940. But in the end I was happy anyway, full of plans to make this work! It took a bit of time!

On the school ( we had the KH-940 knitters but I didn’t want to test too much on these machines for the students. After a fail I abandoned the experiments.

Before the summer I bought my own Brother KH-940 which I tested during the etextile summercamp: I discovered much more than the Nordic knitting technique!

For me the knitting of images using the Nordic technique was already “normal”, we could do this using Designaknit at school:

Back in Rotterdam, besides discovering patent, slip, tuck, ajour (knitting a sock too 🙂 ) I could try to re-lobotomize my knitter. Then (after lots of frustration) I analyzed this board and by measuring I discovered that the PIN 2 of my Arduino Due board was always GND. Since this was a PIN used by the needle count it was obvious important signales were missed. This was easily fixed by assigning another PIN.

During the summer I used the software of but i noticed that lot’s of time the needle count was lost and the image shifted out of view:

I decided to program my own software. That took some time, analyzing the sensor information and building up the Arduino software in combination with the Processing software.

The main point discovered was that the sensor information from the knitter was lost if you sent too much information to the laptop. The knitter is sensing a lot of stuff and we need it all. So in the end the information sent to the laptop is minimized as much as possible.

(and what do you do when software is not working? you add checks, in this case sending more and more information to the laptop….resulting in a real knitting disaster, because then even more signals were lost. 🙂

The second point was getting the offset right for the solenoids. This took some figuring out, because the belt and the carriage have several ways to connect. By discerning between four ways to combine carriage and knitter apparently all possibilities were tackled.

My software is tested further, and I suspect that there will always a dependency on the type of laptop, because the Serial Port always needs some time to “digest” between sending and receiving signals. So if either the Arduino or the Processing sketch go too fast between sending and receiving on either side information can be lost. (The laptop itself, when attending other tasks, can also be a cause for losing information as could be seen in my interface, where sometimes the image is truncated.)

The processing sketch has a basic interface, showing the image and the tours you are knitting. A few nice sounds accompany you while knitting and signaling the end of the knitted image.

Hardware board: basically arduino Due + connections to the KH-940 sensors and 2 ULN2803.

See:, although their board is much more nice now then mine, which was their first version, but is works. And they have developed this fantastic double bed version!

2014-10-11 13.10.56 2014-10-11 13.11.02

Basic interface made with Processing, showing the image, and the tour you are knitting.

2014-10-11 10.14.35

Needles are set according to the knitted tour.

2014-10-11 10.14.51

My fascination is to use the “other” possibilities of the KH-940 for datavisualisation: the slip stitch, the patent stitch, the tuck stitch, well that is starting to become possible…..see next post).

Working at a 3D printer at the same time with the same imagery of datavisualisation you notice one thing: the knitter is not a printer! The knitter is much more like a usical instrument. Knitting will always involve handwork, never be a push the print button and wait action…..:-)

The software can be found here, but as mentioned there is a dependency on laptop, so be careful! First version not documented yet. Still testing!!!!

One remark about Open Source. I always like Open Source code, because it allows you to start fast and comfortably. But then you have learnt nothing,  beacuse you are using other people’s code. If something goes wrong….you have to do it yourself anyway. Making it yourself costs a lot of time. But then you have your own code, you know the structure, and you have learnt something…

Open source is a bit like Fast Food! Better also do some Slow Cooking!

Se more scripting details in this post: