Contact Us

Reflow Profile Verification

Reflow Profile Verification

A reflow oven holds a program. A profile is what the board actually sees, and the two are never the same thing. The gap between them is where most of the arguments about solder quality start, and it is closed by one activity: putting thermocouples on a real board, running it through the oven, and reading the curve.

Reflow profile verification is often treated as a setup task done once when a product is introduced. In practice it is a control that decides whether the joints on panel four thousand are the same as the joints on panel one, and whether that can be shown to a customer two years later.

What a profile really is

The oven display shows zone setpoints and belt speed. The board sees something else: gas temperature above it, convection from the sides, heat conducted up through the support pins, and its own thermal mass pulling the readings down. A heavy aluminium housing and a thin control board running the same program at the same speed come out with different peaks, different time above liquidus and different cooling rates.

Where the thermocouples go

Attachment comes first. High temperature solder on a pad is the most reliable method; adhesive works on a surface that cannot take solder, and tape does not survive the peak. The bead has to touch the joint area, and the wire routed so it does not lift the part it is measuring.

Placement follows the thermal map of the board rather than the convenience of the operator. Five points cover most assemblies well:

One on the largest thermal mass, usually a ground plane area, a shield can or a heavy connector, because that point is the last to reach peak. One on a small chip near a board edge, because that is the first. One under a large body where the joint is hidden. One on the bottom side if the build is double sided. One on the panel frame or a tooling hole, which is the reference point that makes two profiles comparable.

The bands that matter on a SAC305 curve

SegmentBand most SAC305 builds work toWhat drifting out of band looks like
Ramp to soak1 to 3 C per secondFaster ramps spit solder balls, lift small bodies and stress the laminate
Soak, 150 to 180 C60 to 120 secondsShort soak leaves flux unspent and solvent trapped in the paste
Time above liquidus, 217 C60 to 90 secondsShort time gives incomplete wetting and rough fillets
Peak235 to 245 C at the coldest jointLow peak leaves dull, grainy joints that fail pull test
Cooling1 to 4 C per second, even across the boardA slow or uneven cool grows coarse grain and weakens the joint
Oxygen in the zoneBelow 500 ppm where nitrogen is usedRising oxygen dulls the fillet and widens the wetting spread

These are working bands, not a specification for a particular board. A heavy assembly may need the peak moved up and the soak shortened. What does not change is that the bands are stated in writing before the first panel runs, so a deviation is a fact rather than an opinion.

Verifying the profile instead of the display

Verification means measuring twice and comparing. The first run confirms the curve sits inside the bands. The second run, on the same program a few panels later, confirms it lands in the same place. If the two curves separate, something moved: a zone element, a thermocouple, the ambient airflow, or the load on the belt.

The belt is the quiet variable. A heavier load, a panel that bridges the rails differently, or a pallet that absorbs heat all change the dwell, so a palletised build is measured as its own profile.

When to re-verify

A profile is evidence, so it is re-measured whenever the evidence could have changed: a new panel, a change of paste alloy, a stencil change that alters deposit volume, any maintenance on a zone, thermocouple or belt drive, a change from air to nitrogen. One verification per shift on a running build is cheap next to a scrapped lot.

Records that survive an audit

A curve on a screen is not a record. What belongs with the build is the dated chart, the oven identification, the program number, the board part number, the paste alloy and lot, the number of thermocouples and where each one sat, the measured values for each band, and the name of the person who ran it. Medical and automotive customers ask for exactly this set, and asking for it after the lot has shipped is the wrong time to discover that half of it was never written down.

Where profiles go wrong on real builds

Double sided builds are the common case. The second pass reflows the first side, so any heavy part on the bottom has to survive two excursions, and the profile for the second pass is measured with the first side already populated. Moisture sensitivity runs alongside this: a package that absorbed water before the first pass can crack at the second, and no profile change fixes a part that should have been baked.

Nitrogen changes the curve as well as the chemistry. Gas with a different oxygen level transfers heat differently, so a profile approved in air is not automatically valid under nitrogen. The nitrogen reflow practice we follow is measured with the same thermocouple routine.

How SUPERB PCBA verifies a profile

First articles are profiled on the actual panel with five thermocouples, and the curve is stored with the build record. The oven is a 12-zone reflow oven running SAC305 under nitrogen below 500 ppm, and it is re-profiled after any maintenance that touches a zone, a thermocouple or the belt. Assembly runs on seven SMT lines with 3D SPI after print, 3D AOI after reflow and X-ray read to IPC-7095 with voiding under five percent, held to IPC-A-610 Class 2 and Class 3 under ISO 9001, and under ISO 13485 for medical work.

Every deviation is reported with measured values rather than a statement that the build was in control.

Questions buyers ask

Can we use the profile from our previous supplier?

As a starting point, yes. As a released profile, no. The oven, the gas and the panel decide the curve, so it is measured again on our line and compared against your bands.

What happens if a curve drifts mid run?

The build is held at the oven exit, the affected panels are quarantined, and the cause is found before the line restarts. A re-profile is run and compared against the released curve, and the panels held are inspected against the class criteria.

Next step

Send the Gerber set, the BOM, the assembly drawing and any profile bands your quality group already requires. The reply returns the profile plan for the panel, the thermocouple positions, and the values we expect to hold, so the first article is measured against your window and not against ours.