How short is too short?

Under 0.3 mm the needle re-enters the hole it just made. Most digitising software removes stitches under 0.3 to 0.4 mm automatically on export; converters and scalers do not.

Does removing short stitches change the design?

Not visibly. A stitch shorter than the thread's own width adds nothing to the surface; dropping the point lets the thread span 0.3 mm instead of 0.1 mm.

My thread breaks everywhere, not in one spot.

That is the machine: needle, tension, thread quality, a burr on the plate. A file problem breaks at the same stitch on every piece. The pre-flight check tells you whether the file has anything to answer for.

Do tie-in stitches count as short stitches?

They are short on purpose, two or three stitches at the start of a column, and the machine handles them. The check counts stitches under 0.3 mm anywhere; a few dozen in a 10,000-stitch design are ties, a few hundred are a problem.

Short stitches and thread breaks: why a file breaks thread in the same place every time

Thread that breaks at the same point on every piece is almost always the file. Thread that breaks at random is almost always the machine. The file kind has one cause more often than any other: stitches so short that the needle comes back down into the hole it just left.

What happens at 0.3 mm

A 40-weight embroidery thread is about 0.35 mm across and the needle hole it makes is a little larger. A stitch of 0.2 mm asks the needle to go down 0.2 mm from where it just went down, which is inside the previous hole. The thread from the previous stitch is still in that hole. The needle's point catches it, the fibres shred, and three or four such stitches in a row cut through it. The machine stops, or worse, keeps going with the thread wrapped around the needle.

At 0.3 mm the needle just clears its own hole. That is the threshold the pre-flight check counts under, and the threshold most digitising software removes under when it exports.

The same physics is why very dense areas break thread: the penetrations are 0.2 mm apart in two dimensions instead of one. Density explained covers that side.

Where short stitches come from

Scaling down. The commonest source. A design with 0.5 mm stitches around its curves, scaled to 60 %, has 0.3 mm stitches around its curves. Software that scales without re-generating stitches does exactly this; the resize tool re-splits runs instead.

Auto-digitising from images. Tracing a bitmap produces a stitch at every pixel corner. The results are full of 0.1 and 0.2 mm stitches on every edge, which is why auto-digitised files break thread and why their pre-flight reports are red.

Tight curves in running stitch. A digitiser who sets a 1 mm stitch length on a small circle gets stitches shorter than 1 mm at the tighter parts. Usually harmless above 0.3 mm.

Format conversion between resolutions. Rare with modern formats, which all use 0.1 mm units, but old files converted through software that rounded differently can have doubled points.

Tie-ins and tie-offs. Deliberate: two or three tiny stitches at the start and end of every column or run so the thread does not pull out. A few per colour block is normal and the machine copes because the previous hole is not in the same place.

Finding them

Drop the file on the pre-flight check. The stitch-length histogram shows how many stitches fall in each range, with the under-0.3 mm bin in red, and the findings list gives the count and the share of the design. A hundred short stitches in a 10,000-stitch design are ties; a thousand are a problem; the report says which.

The viewer's sew-out slider lets you step to a spot that breaks and look at the stitches there; with show jumps on, a cluster of tiny stitches in a curve is easy to see.

Fixing them

Cleanup removes stitches under 0.3 mm by dropping the point and letting the thread span to the next one. Nothing visible changes, because a stitch shorter than the thread's own width contributes nothing to the surface. The report says how many were removed; the pre-flight check on the result should show the red bin empty or nearly so.

Cleanup will not fix a design whose density is the problem. If the histogram is fine but the heatmap is red, the file needs the digitiser.

Ruling out the machine

Before blaming the file, check that the break moves with the file. Sew a design that has always worked: if it breaks too, the machine is the cause. The usual suspects, in order:

Needle: worn, wrong size, or in backwards. A 75/11 embroidery needle for 40-weight thread; change it every 8 hours of stitching or at the first burr.
Top tension too tight. Loosen a step and test.
Thread path: a snag on the spool, thread caught under the spool, a rough thread guide.
A burr on the needle plate or the hook from a previous needle strike.
Old or cheap thread that has dried out; rayon that has gone brittle.

If the working design sews clean and the problem design breaks at the same stitch, it is the file, and the histogram will show it.

Preventing them

Do not scale stitch files without density correction. Check auto-digitised files before paying for them. When you letter, use a tool that ties in with a short stitch rather than a stack of them: the lettering tool uses two 0.7 mm stitches for each tie and keeps every stitch it generates above 0.4 mm except those.