Why is tolerance control so important in lithium battery design?
In lithium-ion battery R&D and manufacturing, people tend to focus their attention on the design parameters themselves—such as areal density, thickness, baking time, and so on.
However, any process intended for mass production must include one crucial factor that is often overlooked: tolerance.
Tolerance may seem insignificant, but it is the most important criterion for determining whether process variation is abnormal. Coating areal density has its tolerance, tab-welding position has its tolerance, winding-core dimensional standards have their tolerance… No matter how small a detail may appear, if the tolerance is defined unscientifically, problems will inevitably emerge later in production.
Based on the real design experience of our engineering team at iRay Energy, let’s take a look at the most critical points to pay attention to when defining tolerances.
1. Tolerance Determines Design Margin
There is an old saying in lithium battery design:“Without margin, there is no design.”
Why is this true? Because process variation is inevitable, and the size of the tolerance directly determines how much design margin you need to reserve.
A simple and intuitive example:
If the coating areal density tolerance is ±4%, then the cell capacity design margin must be ≥4%. Otherwise, even if the electrode weight meets the specification, the capacity will still be low during grading, and nothing you do afterward can fix it.
This is why the tolerance of every process parameter must be strictly controlled from the very beginning of battery design, instead of pushing the issue to engineering after problems occur in production. If design engineers do not know the extent of process variation, they cannot know how much margin to reserve. If the margin is insufficient, the product will inevitably become unstable. Therefore, when a process parameter has a large tolerance, that parameter must be assigned a larger design margin—otherwise a quality incident will eventually happen.
2. Tolerance Must Not Be Looser Than Process Capability
“If the process can achieve ±2%, then giving a ±4% tolerance in the specification will be safer, and the CPK will look better.”
This is one of the most common misunderstandings in process engineering—and it is completely wrong.
Setting a tolerance much looser than the actual process capability creates two fatal problems:
① Design margin is forced to increase → Material and cost waste
If you loosen the areal density tolerance, design engineers must increase the capacity margin accordingly.
This is essentially forcing yourself to consume more material unnecessarily.
② It creates the illusion that both the upper and lower limits are acceptable → Process becomes uncontrolled
For example, the calendered thickness specification is 100 ±4 μm, but the process is actually capable of ±2 μm.
As a result, the production line may do:
98 ±2 μm today (near the lower limit)
102 ±2 μm tomorrow (near the upper limit)
Both conditions “meet the specification,” but the results are completely different.
The looser the tolerance, the more scattered the manufacturing behavior becomes, and the worse the product consistency.
In summary:Tolerance must not be so wide that the production line can “do anything it wants.”
The stricter and more reasonable the tolerance, the more stable the manufacturing process will be.
3. Tolerance Must Not Be Tighter Than Process Capability
As mentioned above, tolerance cannot be too loose—but it absolutely cannot be too tight either.
If the tolerance is stricter than what the process can achieve, the outcome is straightforward: a large number of defects will occur.
For example, if the process capability can only reach ±2.5%, but you insist on specifying ±1%, the outcome will be either scrap or a forced expansion of the tolerance later—after which you still need to reassess and release those previously out-of-spec products.
Tolerance must match process capability. It cannot be too loose, and it cannot be too strict. This is why process specifications must be based on real production-line data rather than experience or intuition.
In summary:blindly tightening tolerance = pointless torture for the production line. It will not improve the product; it will only reduce yield unnecessarily.
4. The Same Parameter Must Have Consistent Tolerance Across Different Processes
Some factories have a situation that is “ridiculous but common”:
Winding process: tab center distance ±2.5 mm
Incoming inspection for packaging: tab center distance ±3.0 mm
At first glance, the reasons sound “reasonable”:transfer between processes may cause variation, measurement methods differ, etc.
But if your production flow contains many steps, and each step increases the tolerance a little, then the further the product goes, the looser the requirements become—and product consistency will be completely destroyed.
The correct approach is: if the measurement method is the same → the tolerance must be the same.
Unless the measurement method is fundamentally different, or the process transfer must cause a physical change, the tolerance should not be modified casually.
For example, if the mold used to produce a spacer has a machining accuracy of ±0.05 mm, then the process specification should also define ±0.05 mm—rather than arbitrarily relaxing it to ±0.2 mm.
In summary:the process cannot become “looser and looser” as it proceeds.
Otherwise, it is impossible to expect the production line to execute consistently.
5. Tolerances in a stack-up cannot be added linearly
Many process issues are not determined by a single parameter, but by multiple inputs jointly influencing one output.
A common example: CPP exposure length.
Its influencing factors include:
Tab welding position
Insert position during winding
Position after being placed into the case
CPP shoulder height
Top seal trimming width
Each factor may have a tolerance of ±0.2 mm.
If you simply add them up, it becomes ±1 mm — but the acceptable CPP exposure window might be only 0.2–2.0 mm!
Why can’t we directly add them?
Process data follows a normal distribution
The probability of all inputs hitting their extreme limits simultaneously is extremely low
Six Sigma methodology inherently allows an extremely small probability of nonconformity
At the same time, the output tolerance must be ≥ the tolerance of each individual input, otherwise the downstream process will be impossible to execute.
In summary:tolerance stack-up requires statistical thinking, not simple arithmetic.
6. Conclusion
Process tolerances are not a numbers game—they are a core logic of lithium battery design. Many newcomers think that “tolerance is a minor detail,” but in fact, it is exactly the opposite:
Tolerance determines design margin
Margin determines product performance
Product performance determines whether mass production can be stable
Tolerance is not a number to be written casually; it is the baseline jointly followed by R&D, PE, process engineering, and quality control.
Most importantly: designing scientific tolerances means locking in product consistency and stability before the product is even made.
This alone determines how far a battery manufacturer can ultimately go.
