50 g a Piece, ±5 g Tolerance. Three Things I Told the Mantou Baker Before He Flew In
50 g.
One round mantou. It has to land within 5 g.
The buyer runs a traditional mantou shop in northern China. His dough sits at about 50% hydration, which is stiff. He divides before proofing: 50 g a piece, then rounds, then proofs, then steams.
Next week he flies to Guangzhou.
Before he booked the ticket, he told me his terms. Round mantou only, no square and nothing knife-cut. The two jobs that eat labour, portioning and rounding, have to go. And he buys nothing but equipment that holds up.
I answered with three things the machine cannot do.
1. He is not shopping for a machine. He wants two hands back.
Start with the business.
Round mantou is staple food in the north, and it is also stock on a shelf. Round and knife-cut differ by more than shape. Knife-cut runs on a blade: fast, cheap, every piece the same. Round runs on hands: portion one piece at a time, roll one piece at a time, lay them out tray by tray.
Why not switch to knife-cut? He did not explain much. He said the market he sells to is a different market.
So the market he sells to is built out of labour.
Portioning goes by feel, some fast some slow, some big some small. Rounding is all hand: roll too light and the seam will not close, roll too hard and the surface creases. Everyone in this trade knows how many pieces a skilled hand turns out in an hour. The whole business has been propped up by headcount.
Hiring gets harder every year, and skilled hands are the hardest of all. So how does a shop like that keep going?
That is the real reason he called us. He is not short of a machine. He is short of people.
What he wants is a machine that takes the portioning and the rounding off his hands.
2. The first question was not about price
Can you hold 5 g?
A professional question. A mantou maker does not start with how fast, or how many kilowatts. He starts with the error, because 5 g apart at the scale is a tray of different sizes on the shelf, and half the appeal is gone.
I could not answer that on the phone.
Not because I would not. Because an answer down a phone line does not count.
Whether a machine holds 5 g is not on a spec sheet. It is in his dough. Fifty grams a piece at 50% hydration and fifty grams a piece at 60% are two different exams. And nobody had even settled what the 5 g means: one-sided, two-sided, or the spread across a whole batch? Until that is written down, a promise is air.
So I said: come. Bring your dough and your recipe, and you run it here.
3. The three things the machine cannot do
In pre-sales the easy part is making the numbers sound big. The hard part is drawing the line, and telling a buyer where his own risk begins.
I drew three lines.
One: the machine's tolerance band is not a promise about your pieces.
This is our Smart Dough Divider & Rounder (Vertical), model BK-Smart. Its piece-weight tolerance is 0~3g. On steady conditions it holds 0~1g. In this industry 5g is good. 3g is top tier.
That figure describes what the machine can do. It does not describe what your dough will do on it.
There is something more important, and I say it first: fermented dough cannot go on a divider.
Fermented dough is full of gas. The bubbles are different sizes and the structure has gone slack. Push a blade through and it gives way at the big bubbles, so that piece comes out light; where the dough is dense, that piece comes out heavy.
So there is only one order. Mix the dough, divide, round, and only then proof. Dividing happens before fermentation, always.
That is already how his line runs. He has not fallen into this one. But it has to be said out loud: divide after proofing and no machine in the world is accurate. That is not the machine failing, it is the dough no longer being a repeatable part. Anyone who tells you fermented dough weighs out accurately does not understand machines, or does not understand dough.
Two: this machine is at its best between 55% and 65% hydration.
That range is measured, not assumed. Dough outside it still runs. The results just cost you something: the cut edge tears, and the round comes out flatter.
His dough is 50%.
Stiff, and outside the range. He hears it from me now, not from a scale standing next to the machine.
Three: the machine gives you a flat round, not a tall one.
The rounding plate rolls each piece into a ball, and it comes out flat. The diameter is there, the height is not. Northern-style tall round mantou needs a hand to stand the piece up and close the seam, one or two touches.
I am not going to hide this. Machines at this level, the leading brands included, also produce flat rounds. A tall round always takes a hand.
After the three, he relaxed.
When a supplier tells you what he cannot do before you have paid him, you can at least judge that the things he then says he can do are probably not invented.
4. Why this batch of dough is the hardest exam the machine sits
The stiffness is the easy part, one sentence: low hydration means stiff dough, and the blade does not so much cut it as tear it, and a rough edge shows up on the scale. The hard part is the next two layers.
First layer: dough is not a repeatable part
Mixing folds air into the dough. Within one batch, the outside is not the inside. Within one lump, density shifts with how long it has rested and how much it has relaxed.
Dividing means cutting equal weights out of something that is not equal.
It is like sawing equal-weight lengths out of a plank with voids in it. The saw is not wrong. The wood is uneven.
So dividing error is physical. It can be squeezed down. It cannot be zeroed. That is also why fermented dough cannot go on a divider: more gas, looser structure.
Every mantou plant knows this better than I do. What they want is a machine steadier than a hand.
A machine does not know dough better than a hand. It just does not get tired.
Second layer: flat is what rolling gives you
Rounding is rolling. The piece is rolled between the rounding plate and the head, again and again, until its own tack and stretch close it into a ball.
At low hydration the surface is drier. The piece will not turn over, the seam will not close, and you get a flat round.
The standard answer in this trade is to ask the customer for more water. More water, rounder pieces.
But his recipe has been worked out in his own market over years. Change the hydration and the chew, the bite, the proofing time, the steaming time all move with it. Every batch has to be tested again.
Who pays for that, the customer or the machine?
The recipe is his. The machine is the one that adapts.
5. Three things we prepare before the trial
One: pick the disc. 50 g should not sit on a disc's floor.
Same frame, different disc, different weight band: 16-piece 100~200g, 26-piece 50~140g, 30-piece 30~100g, 36-piece 10~70g.
At 50 g, two discs reach: the 26-piece at its floor, the 36-piece through its middle.
We recommend the 36-piece.
The reason is simple. At 50 g the 26-piece disc has no tolerance left below it; in the middle of the 36-piece band there is room on both sides. The same dough behaves differently in the middle of a band and at its edge. This is the point.
The smart model goes down to 10 g, so a 36-piece disc is not a stretch for this machine. It is the machine's normal range.
We also do something in advance. We mix a batch in the workshop at the hydration he described. An engineer sets the rounding scale and the rounding time to a first pass, we keep a second batch of standard dough as a reference, and both batches run at 50 g.
He will not start from zero when he arrives. Two sets of results will already be on the bench, and he can look at the gap between them.
One thing worth stating plainly, because it is easy to mix up. The Semi-Automatic Dough Divider & Rounder (Vertical) and the Smart Dough Divider & Rounder (Vertical) are two different machines, not two names for one. The semi-automatic one keeps both manual and automatic operation and starts at 20 g. The smart model is one-button automatic, lifts by electric actuator, and goes down to 10 g. This article is about the smart one, BK-Smart, and that is the one he will run.
Two: two hands become one button
Press once. The electric actuator lifts the head into position, the position signal comes back, and the rounding countdown starts. At zero the motor brakes itself, the head returns, the buzzer sounds.
Nothing in the middle needs a hand.
Why raise it here? Because his dough is stiff.
The old rhythm runs one hand holding the head down and the other counting seconds, and nobody checks whether the head went all the way down or whether the dough got enough time. Two feelings drift together, and stiff dough drifts further.
Take one hand out, and you take one variable out.
Three: settle what the 5 g means before the machine starts
I put this on the table first: write the acceptance rule down, then switch on.
Is the 5 g a two-sided ±5 g, a 5 g spread across a whole batch, or a 5 g deviation in the average? Three readings, one set of results, and the verdict can differ by a factor of two.
How we usually write it: three consecutive runs, take 10 random pieces from each run, weigh every piece, and read the worst one. Not the average.
Because what he needs is every piece fit to sell, not an average that looks fine.
6. On the day, four things to look at
| What | How | Pass line |
|---|---|---|
| Piece weight holds | three runs, 10 random pieces each, every piece weighed | read the worst piece, not the average; 50g±5g |
| Shape holds | lay one tray, steam one tray, check height and seam | no collapse, no lean; count the touches a tall round needs |
| Another operator can set it | have the operator change the rounding scale and set the rounding time | no veteran's feel needed; the numbers are on the screen |
| A breakdown can be fixed | how the control board swaps, how parts ship, who answers at 24 hours | one board in, production back — not waiting for a visit |
Two of those four belong to the machine. Two belong to the supplier.
Most buyers check the first two, take the machine home, and find out the second two are the expensive ones.
7. The hand work drops from two hands to one touch
Now the money.
Before: portion and round. Two motions, both by hand. One worker turns out only so many pieces an hour, and the work has always been carried by headcount.
After: the machine lays down flat rounds, and a hand stands each one up and closes the seam.
Where is the difference?
Tens of thousands of pieces a day. One second a piece is a shift that no longer needs those people.
I will not draw that conclusion for him. How round his tall round has to be is his call alone. Some plants need one touch. Some want every piece the same height, and then it is three touches, not one.
How much labour it saves is his to time with a stopwatch. The machine only takes the first nine tenths.
8. "Nothing but equipment that holds up" is settled after year one
Buyers compare three things: does it divide true, does it round, and what does it cost.
Get all three right and the machine can still make his life hard at month six, at year one.
So stable and reliable has to be taken apart: into parts, into process steps, into whether a person answers the phone.
- Cast-iron frame: ten years and more in normal service. Not about looks. The frame does not deform, so the clearances stay where they belong.
- Oil-free drive: no oil seal means nothing to leak. A leak is not a maintenance issue, it is a food safety issue.
- Chrome-plated contact surfaces: smooth, rust-proof, washable, and it meets HACCP.
- Four discs, interchangeable, 30 seconds to swap: round mantou at 50 g this year, another weight next year, same machine.
- Control board swapped whole: no tracing wires, no waiting. One board in and production is back.
- 95% of parts made from our own tooling: a cutter disc or a rounding plate bought in three years still fits.
- Every machine runs dough before it ships. Not a sample check. One at a time, piece by piece on the scale.
- One-year warranty, 24-hour engineer response, and trial runs at our place.
The most expensive day in a machine's life is not the day you pay for it. It is the day it sits dead. Have you priced that day?
So do not judge on three lines of a brochure. Look at the parts, look at the steps, look at who picks up the phone at 24 hours. Can you see all three in one day at a trade fair? No. You have to walk into the workshop.
One more thing, said early. A customised machine has to be scheduled. Voltage, disc, colour: if any one of them is made to his requirement, it is not a unit lifted off a warehouse shelf. Time runs forward from the booking. Better now than on the morning the line is supposed to start.
9. The answer: yes, but it gets said at the trial bench
So, can it hold ±5g?
Our read: this machine will basically meet his requirement. Piece weight, yes. Shape, the machine gives nine tenths and a hand gives the last one.
But "basically" belongs on the trial bench. It does not belong on social media, and it does not belong on a brochure.
He arrives next week. Bring the dough, bring the recipe, bring a photo of the mantou he wants, and measure every dimension.
When it is over, he is the one who says whether the machine passes.
WhatsApp Robert — send me your recipe, hydration, piece weight and tolerance, and I will tell you whether this trip is worth your time.