Edinburgh Hacklab

The knitting machine is a Singer Magic Memory somethingorother. Do not be put off by the weird pink colour… this machine is fun and a bit complicated!

It has a standard needle bed with 200 needles. 2-colour patterns etc can be obtained using the punched cards or using the buttons below where the punched cards are fed in. Patterns can be a maximum of a 12-stitch wide repeat.

The machine is not yet working perfectly OK. It is working well enough to knit something in a plain colour, or stripes (so long as you have the right tension settings for your yarn). It is not yet working properly when knitting patterns using the punched cards or buttons.

The following yarn/tension combinations should produce plain stocking stitch with no faults. All yarns are in the drawer of knitting machine accessories (3rd drawer down under the Wera screwdrivers):

• Red yarn (DK weight I think, from West Yorkshire Spinners '100% Wensleydale Gems'): Set tension dial to max (above 10) and yarn feed dial to 3. This produced a gauge of 22sts and 32 rows over 10cm.
• White brushed thermal (slightly fluffy) DK weight: Same settings as red yarn above. Gauge not measured.
• Self striping rainbow yarn (this is West Yorkshire Spinners 4-ply): Cast on successfully with tension dial on 5, but became a bit tight 12 rows later, so changed up to 8, which was fine. Yarn feed dial was set to 5 the whole time. This produced a gauge of about 28sts and 37 rows over 10cm.

So far the following errors have been observed when trying 2-colour patterns…

• The same errors tend to occur whether a punched card or selection via buttons is used. When moving pins by hand on the bed, errors do not occur (i.e. it's not the carriage at fault).
• Errors tend to consist of the contrast colour being knit when it should be the main colour instead. This is presumably because the pins have been pushed forward into position D instead of remaining at position B.
• Errors tend to occur at the same place in each repeat of the pattern across the bed (it has only been tested over 3 or 4 repeats of the pattern across the bed though).
• It looks like sometimes a pin to the left of a correctly selected pin comes along for the ride, but this could just be a co-incidence.

Look inside the machine as the pattern errors are happening to see what might be causing them e.g. look to see that solenoids are correctly selected, and if so, see which tabs on the 12 plates that run along the back are pushing pins forward? There (still) might be a tiny amount of misalignment that is causing tabs to lift up and push pins forward when they shouldn't.

It would be nice to be rid of the 12 stitch repeat constraint and instead have every needle individually controllable. This would require hacking, perhaps (servos, cams?). Ideas are welcome. It is possible to buy electronic knitting machines that bypass this constraint and are known to be hackable, but these are 100s of £££ (although if people fancied a pledge…).