The difficulty of packaging rubber parts for humanoid robots lies in the stage from sampling to mass production

The current stage of humanoid robots is very special: the whole machine is still iterating rapidly, and the demand for encapsulated parts is characterized by "urgent need, few orders, frequent updates, but when it is finalized, it may suddenly increase in quantity". This rhythm conflicts with the traditional logic of mass production of silicone parts, and the most concentrated conflict is in the stage from sampling to mass production. This article discusses from the perspective of a contract manufacturer where this hurdle lies.

Level 1: There are actually three methods in the sampling stage, and choosing the wrong one can mislead later on

Soft mold sampling. Make several to dozens of pieces using simple molds (aluminum molds, resin molds, or simple steel molds) to verify the structure and feel. The advantages are fast and cheap; The disadvantage is that the precision and lifespan of the mold are limited, and the produced parts cannot represent the mass production status.

3D printing of elastic resin. Printing directly with elastic photosensitive resin with similar hardness (such as 50A) takes one or two days to produce, suitable for verifying appearance, assembly, and rough feel. But one thing must be clear: the feel, friction, and rebound of resin and silicone are different. Printed parts can verify whether the shape is correct, but cannot verify whether it works when used. Treating the feel of 3D printed parts as the final feel of silicone parts is the most common misjudgment during the sampling stage.

Formal sample mold. Create a simplified mold according to the mass production process (with fewer holes and sufficient processing accuracy), and the resulting parts will be close to the mass production state, which can be used for overall machine validation. Cost and cycle are between the two.

The correct sequence is usually: 3D printing to verify the shape and assembly → soft mold or sample mold to verify the material and feel → and then decide whether to start mass production mold. The time saved by jumping often doubles back later.

Level 2: Sample mold and production mold are not the same thing

Many teams who are making encapsulated parts for the first time will ask: Why does the production mold feel different when the sample is made well?

Because these two molds are already different:

Sample molds are usually single hole or a few holes, with sufficient processing accuracy, relatively low steel requirements, and manual demolding and part extraction. It solves the question of whether it can be done.

Mass production molds are multi cavity, steel and heat treated according to their long-term lifespan, equipped with complete exhaust and ejection systems, and combined with automated component extraction. It solves the question of whether it can be made stably and inexpensively.

The number of holes changed from 1 to 8 or 16, and the flow path, filling time, temperature and pressure uniformity between each hole of the rubber material all changed. Within batch fluctuations in size consistency, appearance, and even hardness can all undergo changes.

Level 3: Process parameters cannot be simply copied

The vulcanization temperature, time, and injection speed that run out of the sample mold usually need to be readjusted when transferred to the mass production mold. The reason is that the heat capacity of the mold is different - the sample mold is thin, small, and heats up quickly; The mass production mold is large, thick, and has a more complex temperature distribution. With the same parameter settings, the actual temperature history experienced by the rubber material may differ by more than ten degrees.

If this step is saved, the typical manifestation is that the mass-produced parts have missing materials, bubbles, or local sulfur deficiency that have not been experienced in the sample stage.

Our experience in small-scale mass production is that during the sample mold stage, it is necessary to record the parameters completely (not only temperature and time, but also injection pressure, back pressure, and holding pressure switching points). When the mass production mold is first put into operation, the starting point should be based on the recorded values, and then adjusted according to the results of the first piece, rather than starting from scratch.

Level 4: Mold modification is more common than mold opening

The rubber parts of humanoid robots are frequently modified - the structure has been adjusted, the assembly relationship has changed, and the hardness needs to be changed. At this point, we are faced with whether to modify the mold or reopen it.

The cost of mold modification (processing on the original mold) is much lower than reopening, but the prerequisite is that the modification cannot touch the basic structure of the mold (parting surface, cavity layout, ejection system). Changing rounded corners, adding grooves, and adjusting local dimensions can usually be done; The overall structure has changed, and the cost of mold modification may be close to reopening, and the reliability of the modified mold is also discounted.

In actual projects, the cost of mold modification ranges from several thousand to tens of thousands, depending on the amount of modification and mold structure. If you have to spend money on some modifications, it's better to directly open a new mold - this judgment should be clarified during the quotation stage, don't wait until the mold arrives at the factory to discuss.

Three suggestions for the robot team

The first step is to make samples according to the mass production process during the prototype stage. Even if the quantity is small, try to use a process route that is close to mass production (the same rubber system, the same vulcanization method), so that the validated data has reference value.

The second point is to inform the supplier in advance about the matter of 'mass production in the future'. Mold design, hole reservation, and material selection will all be different as a result. Wait until mass production before changing the process, and most of the previous validation work needs to be redone.

The third point is that the single item price during the sampling stage will be very high. It is normal for a single item to be expensive during the sample mold stage - if the mold cost is shared among dozens of pieces, the unit price will naturally be higher. The true price will only be revealed after the mass production mold is put into use. Using the sample price to calculate the mass production cost will lead to incorrect conclusions.

A truthful statement

In the field of humanoid robots, the current common state of encapsulated parts is "more samples and less mass production". For factories in this industry, the difficulty in taking on such tasks is not in technology, but in how to maintain efficiency in the rhythm of small batches, multiple varieties, and fast iterations. For the robot team, it is necessary to find a supplier who is willing to accompany the iteration and can catch up with subsequent mass production - these two things must be established simultaneously, and only half of them will cause the project to get stuck at a certain stage.