Return Zero

I make bags. Webbing, foam-stabilised panels, layers of interfacing, and straight stitch through all of it. Last year I killed a domestic sewing machine on a set of interlined curtains, spent three months working out what to replace it with, and published Torque, Not Speed, which runs to nine thousand words and ends with a list of machines I would recommend.

Then I bought a machine that is not on that list, from a company most people in Europe have never heard of, for €178.

It is a Feiyue FYe355. This is how I got there, what the research turned up that the guide missed, and the two things I had wrong.

What I was buying on

Heavy sewing happens slowly. You approach the thick part, you slow right down, you walk the machine over it one stitch at a time, and for the worst of it you turn the handwheel by hand. What decides whether the machine gets through is whether the needle develops enough force to pierce a compressed stack at almost zero rpm. Stitch count, wattage and top speed, which is what the category is sold on, tell you nothing about that. The guide argues all of this at length; here it is the assumption everything else rests on.

It left me with three things worth paying for:

• A DC motor. It keeps most of its pulling force down to a crawl. A cheap AC machine slows down by dropping the voltage at the pedal, and the torque drops with it, so you get a lurch over the seam rather than a push through it.
• A speed limiter slider. It caps the top speed so the full travel of the pedal maps onto a slow range. Without one you are stuck with whatever curve the control board applies, and on cheap machines that curve is tuned to feel lively in a showroom.
• A presser foot lift published in millimetres. Not because two millimetres either way changes much, but because a seller who gives you a figure is telling you something, and a seller who says only “two step” or “extra high” is telling you nothing.

Nothing in the guide had all three, which was the annoying part of writing it.

The Janome HD3000 has the most metal in it by a distance and an AC motor with no speed control. The Necchi Q132A is a very good chassis with the same problem. The Veritas 21 Denim has the cleverest drive train on the page, DC through two belts with a reduction stage, and no slider. The Minerva HS1000 was the only machine in my whole table publishing a foot lift, and it has no slider either.

So one of the three had to go, and for a few weeks I could not decide which. What broke the deadlock was working out that most of these are not really different machines.

They are mostly the same machine

Feiyue is a sewing machine manufacturer in Taizhou, China, and one of the largest in the world, with over three hundred products across thirty-one series. Almost nobody in Europe buys a machine with their name on it. Plenty of people buy their machines.

The Texi Force, sold by a Polish industrial distributor. The Jaguar SJ-23 and SJ-34E, sold as Japanese. The Minerva HS1000. Fabric Mouse’s own heavy duty models in the UK. The W6 N 1235 Pro in Germany. Those are Feiyue castings with European badges on them, and once you know that, the price spread stops looking like a quality ladder.

The Texi Force and the Jaguar SJ-23 agree on every specification either company publishes, and there is about ninety-five euros between them. On a shared platform you are not choosing a machine. You are choosing an importer, a warranty and a phone number.

Those are worth real money. A local distributor with a service department is the difference between a repair and a landfill decision. But there is a point where the markup stops buying support and starts buying the badge.

The manual that gave it away

Feiyue’s Brazilian distributor publishes the full manual for each model as a PDF. Put the manual for the FYe390/391 next to the one for the FYe392/393 and they are the same document. Same illustrations, same numbered parts list, items one to twenty-nine identical. The 392/393 adds exactly three entries: 30. Speed adjustment bar. 31. Up/down needle position. 32. Start/stop button.

That is not a different machine. It is the same casting with an electronics module fitted, and the manufacturer documented the module by appending three numbers to a parts list. Which changes what you are shopping for. If the electronics are a module and the casting is the platform, you are picking a casting, a module, and whatever markup the badge on the front adds.

The specifications nobody in Europe publishes

That Brazilian distributor turned out to be the best source in the whole project, and I found it late. European sellers of these machines publish almost nothing. Behind a download link on each Brazilian product page is a full parameter table: foot lift, needle system, frame material, net weight, the lot.

Same manufacturer, same models, same tooling, different market and different disclosure norms. The technique generalises: when a manufacturer sells the same product into several markets, read the market with the strictest labelling culture rather than the one nearest to you.

Feiyue’s domestic machines fall into two families, and almost everything above is one of them. The FY3 series is the older, heavier, mechanical-selection platform. The FY4 series is the newer computerised one. Here they are side by side, in Feiyue’s own numbers.

Four things came out of that table, and one of them cost me a recommendation.

The Minerva’s ten millimetres is not real

My own article says of the Minerva HS1000: the only machine publishing a foot lift in millimetres, and ten is a real number. I put it in the conclusion as the pick for anyone whose priority was the lift.

The Minerva HS1000 is the Feiyue FYe381. Feiyue’s sheet for that casting gives the presser foot lift as 5.5 mm. The ten has no manufacturer behind it. It is the retailer’s number, and I singled it out for praise and then built a recommendation on it.

The same sheet took something else with it. The FYe380/381 table has no fields for speed adjustment, start/stop or needle up/down, because that casting ships without the module. The guide already said the Minerva has no slider. Now I know why.

What the 5.5 mm actually refers to

Then I over-corrected. The English sheets label that field Extra-high presser foot lift, so I read 5.5 mm as the maximum, which would be a startlingly low ceiling.

It is not a plausible maximum. An ordinary domestic lifts about six millimetres at the normal raised position, and a second step typically reaches nine to twelve. The Juki TL-2010Q is nine millimetres by lever. The Janome HD9 is 12.7.

The bilingual sheet gives it away. The Portuguese column for that field reads Calçador altura de elevação, presser foot lifting height, flat, with no “extra” in it. Only the English adds the superlative. So 5.5 mm is the standard lift, and what the second step reaches is unpublished on both platforms, by everybody, in every market.

The cheaper machine is the heavier one

The FY3 machines are heavier, have a bigger motor, and cost less. The FY4 machines are lighter, have a smaller motor, and cost half as much again. That is backwards from how everything else in the category is priced.

It is not the mechanical cam stack, which was my first guess. It is the frame, and Feiyue say so directly: the FY3 platform is built around an integrated metal frame, the FY4 platform is aluminium and plastic. The extra money on the FY4 buys a stepper for needle positioning, a stepper for feed, a control board, an LCD panel, and a trimmer with its own motor, cam, hookers and blade. Those are real components, and they are also the parts a buyer can operate in a thirty-second demonstration on a shop counter. A heavier casting is not demonstrable that way, and nobody has ever been sold a sewing machine by being invited to lift it.

Eight layers of khaki

The FY4 table carries the only published thick-fabric capacity I found anywhere in this category: 8 layers khaki.

Khaki twill runs roughly seven to nine ounces, call it half a millimetre a layer, so eight layers is about four millimetres compressed. Against a 5.5 mm lift that fits with room to spare, which makes it a torque and feed number rather than a lift-limited one. For scale, it is roughly four to five layers of fourteen-ounce denim, or one ordinary jeans hem crossing a flat-felled side seam. Useful domestic capacity. Not bag hardware, webbing sandwiches or six layers of upholstery leather. Not nothing, and not heavy duty.

It is also the only number of its kind on the page. Everyone else leaves the column empty.

The needle ceiling

Both platforms publish a needle system of Nm75 to Nm100, a size 16 at the top. The standard domestic range runs to Nm110/18, and Nm120/19 exists for leather and heavy topstitch. Feiyue cap one full size below where the needle system itself tops out.

What that costs you is thread, not fabric. A 100/16 comfortably carries about Tex 40 to 50 topstitch cord. Then I checked what I actually use, which I should have done first. Gutermann Mara 80 is Tex 35, and its recommended needle is 80/12 to 90/14, so I sit in the middle of the published range with room above me. The ceiling I had spent an evening treating as a serious limitation is not one for my work, though it would be if I moved to Mara 30 for structural seams, which is Tex 100 and outside the range entirely.

Check the specification against what is already on your shelf before you weigh it. I nearly talked myself out of a machine over a thread I do not use.

What the sheets could not tell me

No parameter table answers the question I actually had, which was whether the extra electronics on the newer machines cost you anything in durability. So I asked two sellers, and spent a genuinely silly amount of effort on the wording.

W6 are a German manufacturer who sell their own machines direct and publish more detail than anyone else in the category, down to net weight to the gram. Fabric Mouse are a UK retailer who sell Feiyue machines under both their own name and Feiyue’s. I asked each of them the same thing about their cheap machine and their expensive one.

Four things I learned about asking:

• Do not offer the easy answer. My first draft ended “or are they similar?” That is the answer that costs nothing and commits to nothing, and handing it over is asking them to take it.
• Remove their incentive to shade it. Saying you are happy to pay the higher price regardless does more work than any amount of asking for an honest opinion. The reason to be vague is the sale, so take the sale off the table first.
• Ask about abuse, not quality. “Which is better” asks someone to rank their own flagship second. “Which tolerates more abuse” is not a quality ranking at all, and there is a face-saving answer available: the older one is the simpler, tougher one, the new one is the refined one.
• Give them an exit that is not a climbdown. Ending with “or would you say stick with the older model?” makes recommending the cheaper machine a reasonable answer rather than an admission.

Both messages ended up in the same shape. Here is what I make, straight stitch through thick layers is all I care about, the stitch count and the extra features are no use to me, it would be doing that for years. Happy to pay more for the expensive one, but only if it holds up to that as well as the cheap one does. Does it, or would you say stick with the cheap one?

Both of them talked me down.

W6 recommended the mechanical N 1235 Pro over the computerised N 2800 Pro for exactly the work I described, and added, unprompted, that these are household machines which can reach their limits under daily intensive use over many years. That is a seller declining fifty-four euros and volunteering a caveat about their own category.

Fabric Mouse replied that the computerised machine, with many more moving parts, may fail earlier in the long run, and that their 3 series would suit me better for straight stitch work. The machine they talked down had launched the month before.

Two sellers in two countries, selling different product lines, with no coordination, gave the same answer for the same reason: fewer parts fail. Neither was led to it. It is also the opposite of what the category’s own marketing says. Every heavy duty range in Europe is sold upward. Both of the sellers I asked pointed down.

What I bought

An FY3 casting with the electronics module fitted, under Feiyue’s own name rather than a badge. That machine is the FYe355.

A hundred watt DC motor, twenty-three stitches on a dial, a speed slider, a start/stop button and needle up/down, so I can sew without touching the pedal and set the crawl before I reach the thick part. Six millimetre zigzag, three needle positions, adjustable presser foot pressure, integrated aluminium alloy frame, 6.4 kg.

The twenty-three stitches are not a compromise. For bag work it is straight stitch, zigzag, maybe a triple-stretch for stressed seams and a bar tack. Nobody in the history of sewing has run out of stitches.

What I gave up is the auto trimmer and the auto-lock, and they are not nothing. Auto-lock plus auto-trim is a real per-seam time saving and it compounds over a project. Two things pushed against them for my work. Auto-lock is usually a fixed pattern, and on webbing and strap anchors I want a bar tack or a box-X, which I am stitching manually regardless. And the trimmer cuts short, where a seam end under load wants a longer tail. If I sewed most days it would be defensible. For project work, the machine plus a walking foot is the better use of the same money, and the price gap paid for the walking foot with change left over.

The lift, measured

Which leaves the one specification I had made a condition of the purchase, and then bought without.

Nobody publishes what the second step reaches. Not Feiyue, not Jaguar, not W6, not Minerva, in any market, in any language. What you can find is 5.5 mm for the standard position on Feiyue’s sheets, and a quarter of an inch, 6.35 mm, quoted on a Jaguar listing for the same casting. A two-step mechanism adds a couple of millimetres over the first position, so I bought it expecting about eight, on an estimate, which is exactly what my own article tells people not to do.

I have since put a ruler against it. Eight millimetres.

To make that mean something, measure it against the machine most people in this category already own. The Singer Heavy Duty range is the best known name on the shelf, and its second step is eight millimetres too. So whatever else a FYe355 gives up to a Singer HD, it is level with it on the one dimension that decides whether the work fits under the foot at all. If you have hemmed curtains on a Singer HD and know where it starts to complain, you know where this machine starts to complain. Neither manufacturer will print the figure.

The same number arrives from a second direction. There is a set of teardown and stress-test videos for the Jaguar SJ-23, one of the badged versions of this platform, made by a technician who also sells the machine. He runs it through thirty layers of linen, twelve layers of folded denim hem and six layers of upholstery leather, breaking a needle on the way, and adjusts the hook timing on camera when it starts to skip.

The useful part is where it stops. He reports that the limit is not torque, because the motor keeps driving. What stops it is the needle bar and clamp colliding with the top of the presser foot, at roughly eight to nine millimetres of compressed stack. So the height the foot rises to and the height at which the mechanism runs out of room above it agree, at eight.

Treat those videos carefully. The technician is also a dealer, the demonstrations are promotional, and the machine is set up perfectly by the person selling it. The thirty layers of linen headline is worth less than it sounds, because linen compresses. I am not linking them, because the summary I was working from cited three separate videos that all resolved to the same one, so I cannot say which clip shows what. The failure mode still survives all of that, because a geometric limit is a property of the casting rather than of the demonstration.

So I ended up with a DC motor, the better kind of speed control, and a foot lift I can put a figure to. The figure is mine rather than theirs. That is a smaller compromise than I expected and a worse state of affairs, because the number exists, the manufacturer knows it, and it took a ruler in my kitchen to find out.

What it will not do

It is still a domestic machine. Bottom feed, class 15 bobbin, flat shank needles, grease-packed bearings, a hundred watts, and an eight millimetre ceiling on what can physically sit under the foot. Thick leather is outside what it will do. So is bonded nylon, and so is anything wanting a needle above a 16.

The tier above it is closer than it looks. If I were sewing full days I would be looking at an industrial head with a servo motor for roughly the same money, which the guide argues at length and I still believe. The catch is the same as ever: an industrial stays where you put it, and I do not have a corner of a room to give it permanently.

For bags, curtains and heavy garment work it does what I need, and I am not paying for a thread trimmer and eighty stitch patterns to get there.

Two things I had wrong

One of these reached the published guide and has now been corrected there. The other never made it in, which was luck rather than judgement.

• The Minerva HS1000’s ten millimetre lift, and the recommendation built on it. The manufacturer’s sheet for the same casting says 5.5 mm. Both are withdrawn from the guide.
• My reasoning about why a speed slider and a metal frame never appear together. I had decided a slider needs a control board that a plain mechanical machine does not have, so the two sat on opposite sides of a cost boundary. The Jaguar SJ-34E has a speed dial, mechanical stitch selection and no screen anywhere, and Feiyue’s own manuals show the electronics being fitted as a module to the same casting. There is a whole middle tier I had missed.

Both came from the same mistake. I trusted a European retailer’s figure over a manufacturer’s, because the retailer’s was the only one I could find in a language I was reading. Feiyue’s numbers were public the whole time, published on the other side of the world.

If you own anything on this platform, I would like a number from you. Raise the presser foot to its second step, put a ruler against the gap, and tell me what you get. Mine is eight. The manufacturers will not print it, so we may as well build the list ourselves.

The buyer’s guide is here: Torque, Not Speed. It has the motor physics, the pedal circuitry, the clone tables and the checklist to take to a dealer.