TGV alignment packaging is a different kind of engineering problem. On a glass panel that measures 191.59 mm by 172.52 mm, four separate PI film pieces are placed on the same substrate, and each placement has to hold a bonding position of ±0.03 mm. At that scale, an alignment error is not a cosmetic issue; it turns into trapped air, and trapped air ends a batch. This is the specific difficulty we are usually asked to solve when an enquiry arrives as "we have an idea and we need a machine for it."
Start from the alignment problem, not from the machine
The first working session is rarely about mechanics. It is about the process. Before any structure is proposed, we need the panel and film dimensions, the target bonding accuracy, how many placements one unit requires, and whether those placements are sequential or repeated on the same part. Those four answers determine whether the design needs a vacuum plate, a roller, an XYZ or an X-Y-θ stage, and how the part is held between steps.
For the TGV alignment packaging machine we build around this process, glass is fed by hand and located by camera, and the PI film is fed the same way. Linear motor modules move the tooling, and the part sits on an electromagnetic adsorption platform or an electrostatic ceramic adsorption platform depending on what the material tolerates. Alignment correction happens on X, Y and θ, because a small angular error across a panel that size shows up at the far edge as a visible defect.
Why the sequence matters more than the force
Bubbles are the failure mode to design against. Our approach is a defined order of operations: the adsorption platform presses the part, pre-bonding sets the position, the chamber is pumped down, and bonding happens after the negative pressure reaches the set value. Alignment is only released once that value is reached. Doing it in this order is what removes the air that a simple press would trap between the glass and the film.
The equipment also carries an FFU ion cleaning module. Particles on a glass surface of this size behave like spacers, and one particle under a PI film is enough to make a bond line fail. Cleaning is therefore part of the machine, not an extra station in front of it.
What "prototype" has to carry that a lab demo does not
A prototype that only works once is not a prototype. Two things separate a demonstration cell from a machine you can put next to a line:
- Measurement that survives an audit. Our CCD system carries a measurement function, and the data can generate local marks or be pushed out over the network, so a batch can be traced back to the placement that produced it.
- Data that fits the factory. The machine ships with an industrial computer, connects to MES, and stores the run data locally. A buyer's first question about a prototype is usually whether the numbers can leave the equipment, and the answer here is yes.
- A throughput figure you can plan against. For glass vacuum bonding at this format, capacity is roughly 8–10 pieces per hour, counted with one unit requiring four bonding operations.
- An operator who is not a machine builder. A touch screen and vision interface, plus selectable program formulas for different product processes, is what makes the machine usable on a shift rather than only in the hands of the engineer who built it.
The footprint is fixed early for the same reason. The TGV machine is held to L2260 × W1700 × H2300 mm including the vacuum pump and the FFU height. That is decided in the design stage, because a prototype that will not fit the customer's floor plan cannot be validated at the customer's floor plan.
The stages a company actually moves through
In practice the path from an idea to a running prototype has four steps, and most of them are documentary rather than mechanical:
- Process definition. Sample parts are supplied, dimensions and tolerance targets are agreed, and the bonding sequence is written down. This is where an unrealistic ±0.01 mm request becomes a workable number, or where it becomes clear that the request is genuinely possible.
- Design review. Feeding method, holding method, alignment axes, cleaning, safety (safety gratings on this machine) and footprint are decided together with the customer's process owner.
- Build and in-house trial. The machine runs on supplied parts until the alignment accuracy and the bubble requirement are both met, with the recipe formulas saved against each product variant.
- Transfer. Recipes, MES interface, training, and the measurement data format move to the customer's line, so the prototype is judged on the customer's product, not on ours.
None of these stages is a purchase of spare capacity. Because we work as an integrated engineering team, the same group that defines the process builds and commissions the machine, which is why a change discovered at trial stage does not restart the design.
What a buyer should ask before signing off a prototype
Three questions decide most of the cost and schedule, and they are worth settling in writing:
- Which accuracy is measured, and where? A bonding accuracy claim has to state the panel format and the position on that panel, otherwise two suppliers are quoting different things.
- Is throughput stated per placement or per finished unit? For a part needing four bonding operations, that single distinction changes the capacity number by a factor of four.
- Can the recipes be reloaded for a new product revision? Multiple program formulas per process mean a revision is a file change, not a mechanical rebuild.
If you are at the idea stage, send the panel and film drawings together with your target accuracy and your expected volume. That is enough to start a design review, and it is usually enough to tell you within the first round whether a product R&D accelerator route or a full machine build is the cheaper way to prove the concept. The equipment behind the figures in this article is documented on the TGV alignment packaging machine page, and our engineering services cover the stages in front of it. Enquiries can be sent to sales@powerde.net or +86 13023198376.
