Key takeaways
- A pull test is destructive, so it qualifies the weld process and the lot, not every part.
- The joint should be stronger than the braid: a good result fails in the braid away from the weld.
- Grip the pad and the braid without cutting strands, and pull along the braid axis.
- A load figure without the failure location is not a qualification record.
What a pull test proves
A braid weld pull test loads the joint in tension until something breaks. It proves that the weld can carry a stated load, and it shows where the assembly prefers to fail. That second piece of information is often more useful than the peak load.
The test does not prove fatigue life and it does not prove joint resistance. A weld can be strong in tension and still have voids that raise the resistance, and a weld with low resistance can crack after a few thousand flex cycles. The pull test is one of three checks, alongside millivolt drop and bend life.
The result is also a check on the process, not on the part. Because the test destroys the sample, it can only qualify a weld set-up, a material lot and a production period. A pass on one sample says the process was capable at that moment, not that every joint in the shipment is sound.
Sample size and frequency
A pull test is destructive, so the sample has to come out of the lot and be replaced. The frequency should be set from the risk: every weld set-up change, every new material lot, the first article, and then a stated number per production period.
The sample should be built with the production weld parameters and the production pad. A test coupon welded on a different machine or with a different electrode gives a number for that coupon, not for the parts. If the weld is made on a fixture, the sample should come off the same fixture.
For a new process, test enough samples to see the spread. Three samples give a range; one sample gives a hope. A qualification that runs five or ten samples, all built to the same procedure, gives a much better picture of the minimum the process can deliver.
Specimen preparation and gripping
The specimen is the terminated braid, with the pad at one end and enough free braid at the other to grip. The free length should be long enough that the grips do not interfere with the weld or the heat affected zone. If the braid is too short, the test measures the grip, not the joint.
Gripping a braid is awkward because the strands are loose. A clamp that crushes the braid will cut strands and start the failure at the grip. A grip that relies on friction may slip. A potted end, a wrapped and clamped end or a purpose-made braid grip spreads the load across the strands and keeps the failure in the test section.
The pad end is usually held through the hole with a pin or a shouldered fixture. The pin should match the production bolt diameter, because a small pin concentrates the load at the hole edge and can pull through a pad that would be sound in service.
A grip failure is an invalid test
If the braid breaks at the grip or slips out of it, the result says nothing about the weld. Record it as invalid and retest with a better grip rather than reporting a low number.
Pull direction and rate
Pull along the axis of the braid, in line with the pad hole and the weld. An off-axis pull adds a peel component and will fail the joint at a lower load than a pure tensile pull. If the service load is off-axis, test that direction as a separate case and say so.
Use a constant crosshead rate and record it. A fast pull gives a higher peak on some materials because the joint does not have time to redistribute load. A slow pull is closer to a static service load. The rate should be stated so the result can be repeated.
Record the peak load and the load at first drop. For a braid, the first strand break may occur well before the peak, and the peak may be carried by the remaining strands. Both numbers are useful, and the failure criterion should say which one is the acceptance value.
The failure modes to expect
The best result is a braid rupture away from the weld. The strands break in the free length, which means the joint carried more than the conductor. The load should be close to the calculated tensile of the braid, allowing for the strand arrangement and the grip.
A weld interface failure pulls the braid out of the pad with little or no copper attached. That points to insufficient heat, pressure or time, or to surface contamination. A heat affected zone failure breaks just beside the weld, where the copper has annealed. That points to too much heat or a weld that is too close to the free length.
A ferrule slip, if the design uses a crimp as well as a weld, shows that the crimp did not grip. A pad tear at the hole shows that the pad is too thin or the pin too small for the load. Each mode has a different cause and a different corrective action, which is why the location has to be recorded.
| Failure location | What it looks like | Likely cause | Action |
|---|---|---|---|
| Braid, away from weld | Strands break in the free length | Sound joint, load above braid strength | Accept |
| Weld interface | Braid pulls out with little copper | Low heat, pressure, time or contamination | Review weld parameters and cleaning |
| Heat affected zone | Break just beside the pad | Excess heat or a short free length | Reduce heat, move the first bend |
| Pad at the hole | Pad tears or pulls through | Thin pad or undersized pin | Increase pad or pin size |
| Grip | Break or slip at the clamp | Grip crushes or does not hold | Improve grip, retest |
The failure location is the most important line in the record. A load without a location cannot be acted on.
Setting the acceptance load
The acceptance load should be derived from the braid, not from the weld. Calculate the tensile of the copper: the total wire area multiplied by the tensile strength of the wire. For illustration, 6.0 mm² of annealed copper at 220 MPa gives about 1.3 kN. A joint that fails below that figure is weaker than the conductor.
A common rule is that the joint must carry a stated percentage of the braid tensile, with the failure in the braid. The percentage should allow for the strand arrangement and the test method. It should not be set so high that only a perfect weld passes, because then the test rejects sound parts.
State the acceptance in the drawing or the control plan: the sample size, the frequency, the pull rate, the minimum load and the required failure location. A supplier who is told only a load will optimise the weld for that load and may not report where it broke.
Recording the result
A pull test record needs the peak load, the load at first drop, the failure location, the specimen identification and a photograph of the failed joint. Add the weld parameters, the material lot and the date. Without the parameters, a failure cannot be traced back to a change.
Keep the failed samples or photographs with the record. A description of a failure mode is less useful than an image, especially when the mode is a mixed one such as a partial weld tear followed by a braid break.
Review the results over time. A drift in the average load, even within the limit, is an early sign of electrode wear, a material change or a fixture problem. The pull test is a process monitor as well as an acceptance check.
Frequently asked questions
How many braid samples should be pull tested?
Enough to see the spread, not just one. Test at every weld set-up change, every new material lot and at a stated frequency in production. A qualification on five or ten samples gives a much better picture than a single pass.
Should the braid or the weld fail first?
The braid should fail first, away from the weld. That shows the joint is stronger than the conductor. A weld interface failure means the joint is the weak point and the process needs review.
Does a pull test prove the joint resistance?
No. A weld can be strong in tension and still have voids or unfused strands that raise the resistance. Use a millivolt drop test for the electrical check and a bend test for the fatigue check.
Why does the pull rate matter?
A faster pull can give a higher peak load because the joint has less time to redistribute strain. A stated constant rate makes results comparable. Record the rate with the load so a later test can repeat the method.
Sources and further reading
- ISO 6892-1, Metallic materials, Tensile testing, Part 1: Method of test at room temperature
- IEC 61238-1, Compression and mechanical connectors for power cables
- Qumek weld qualification and pull test notes
Qumek Engineering Desk
Engineering team
The engineering team behind Qumek drawings, DFM notes, tolerances and inspection standards. Every guide is reviewed against the drawings and processes used on live programmes.
Qualifying a braid weld?
Send the joint drawing, weld method and load case. The engineering desk will review the sample plan, grip and acceptance record.
Send a drawing