Pull Strength: Design Rules and Process Limits

Pull strength testing measures the force needed to separate a component lead or a wire from its solder joint, and it is the most direct check a shop has on the mechanical quality of that joint. The number it produces is only useful when the geometry, the test speed and the failure mode are all recorded with it, because the same joint can report very different values under different setups.

This guide covers six points for running pull strength testing on solder joints, the acceptance limits that follow from the data, and the failure modes that tell you whether the joint or the test was the weak link. It applies to through-hole leads, wire terminations and surface mount parts tested with a hook or a grip.

What Pull Strength Testing Measures

A pull test loads the joint in tension, which is not the loading a board sees in service. Its value is comparative: it detects a change in process, a contaminated surface or an intermetallic problem before those defects reach the field. A single reading tells very little; a trend across batches tells a great deal.

The measured figure combines the strength of the solder, the strength of the interface at the pad and at the lead, and the stiffness of everything between the grip and the joint. A change in fixture stiffness or in wire length is therefore enough to shift the result without any change in the joint itself.

Point 1: Choose the Right Pull Geometry

A lead should be pulled along its axis, so that the load passes through the joint rather than peeling it away. A wire should be pulled perpendicular to the board with a defined free length, because a longer free length absorbs more energy and lowers the peak force the joint sees.

The geometry has to be identical across every sample in a comparison. Where a hook is used, the hook diameter and the point of contact should be written down, because a hook that bites into the wire creates a stress raiser and the sample then fails at the hook instead of the joint.

Point 2: Fix the Test Speed and Hold Time

Test speed changes the result. Solder creeps under load, so a slow pull gives the material time to deform and reports a lower peak force than a fast one. A speed between about 1 and 5 mm per minute is the usual range for joints of this size, and it must be held constant across the comparison.

Hold time matters where the test is run to a defined load rather than to failure. A joint held at a load for a fixed period either survives or does not, which makes the result a pass or fail rather than a number. Both approaches are valid, but the acceptance criteria have to match the method in use.

Point 3: Prepare the Sample Without Changing It

Preparation is where most pull strength data goes wrong. Cutting a lead flush with the board removes the material the grip needs. Heating the board to remove a conformal coating anneals the joint. Cleaning with a solvent that leaves residue changes the surface and can mask or create a failure.

Where the sample must be prepared, the preparation should be identical for all of them and it should be described in the test record. Where a control batch of known good joints is available, it should be prepared the same way, because an unexplained shift in the control usually points to the preparation rather than to production.

Point 4: Read the Failure Mode, Not Only the Number

The failure mode decides what the number means. A ductile failure inside the solder with a torn surface indicates a joint that yielded. A brittle failure at the pad interface indicates an intermetallic or a contamination problem. A failure at the grip indicates that the test, not the joint, was the limitation.

A pull strength result without a failure mode is not an inspection result. Each sample should be classified and, where possible, photographed, because the classification is what tells the process engineer where to look. Our notes on ball shear and pull testing cover the classification scheme in more detail.

Point 5: Set the Acceptance Limit From Data

The acceptance limit should come from a distribution of results on joints that are known to be good, not from a figure copied out of a standard. A common working rule is a limit at three standard deviations below the mean of the qualified process, which flags a shift without rejecting normal variation.

A minimum requirement from a customer standard overrides the shop limit where the two differ, and the shop limit should then be tightened so that it is met with margin. The basis for both figures should be written into the test method so that a change of operator does not change the limit.

Point 6: Record and Trend the Results

Pull strength data is most useful as a chart. Plotting the mean and the spread by week, by shift and by paste lot shows a slow drift that no individual reading would reveal. Where the mean falls while the spread stays constant, the process has shifted; where the spread grows, the process has lost control.

The record should carry the sample count, the test speed, the failure modes and the operator, alongside the force values. The sampling and reporting practice that applies to any destructive check is described in our notes on PCBA inspection standards. Where a batch fails, the samples should be retained until the cause is found, because a retest of the same parts rarely reproduces the original result.

Comparing Pull Strength With Shear and Bend Tests

Shear loads the joint sideways and is the standard test for area array packages. Bend testing loads the whole assembly and reflects the board level behaviour. Pull strength loads a single termination in tension. The three answer different questions and are not interchangeable.

Where a shop can only run one, pull testing gives the best early warning on through-hole and wire joints, and shear gives the best warning on area array packages. Our guide to shear and pull testing compares the setups, and the general acceptance practice for solder joints is published by IPC.

Verification and Records

Sampling is a compromise between cost and confidence, and the sample size should follow from the failure rate that has to be detected. Where a decision is made by judgement, a boundary sample makes the judgement repeatable between operators and between shifts.

FAQ

How many samples are needed for a pull strength check? Enough to show the spread, which in practice means at least ten per batch and more where the process is new or a change has been made.

Can a pull test be run on a finished assembly? Usually not, because the test destroys the joint. It is normally run on test coupons or on a sample board built alongside the production batch.

Why does the same joint give a different result on two machines? Because the fixture stiffness, the grip geometry and the test speed differ. The two setups have to be correlated on the same samples before their results can be compared.