Reflow Conveyor Vibration: Where Components Move
Conveyor vibration is one of the few reflow variables that leaves no trace in the profile. The thermocouple records the peak temperature, the zones hold their setpoints and the boards come out at the right rate, yet a handful of small parts have moved a fraction of a millimetre and a few have rotated. The cause is mechanical, and it acts during the few seconds when the solder is liquid and the component is floating on it.
The forces involved are tiny. A chip resistor weighing a few milligrams, sitting on a film of molten alloy with almost no friction, responds to accelerations that a person cannot feel through a bench. The job of the process engineer is to find where that acceleration comes from and to reduce it before it reaches the board. A component shift of a tenth of a millimetre is enough to fail a placement check, and a rotation of a few degrees is enough to bridge a fine pitch part.
What Actually Moves
The board does not need to jump for a component to shift. A lateral acceleration applied to the assembly for a second is enough to slide a part on a liquid film, and a vertical impulse is enough to bounce it clear of the paste before the alloy solidifies. Once the part has moved, surface tension pulls it back towards the pads, so the final position can be a compromise between where it started and where the wettable area is.
Rotation is a common outcome because a part on two pads has a centre of mass that is rarely exactly between them. A small asymmetry in the paste deposit or in the acceleration is enough to start the rotation, and the wetting force at the ends does not always pull it back. Small two-terminal parts and tall parts with a high centre of mass are the most sensitive.
Where the Vibration Comes From
The largest source of conveyor vibration in most reflow ovens is the drive train. A chain or a mesh belt running over sprockets transmits a periodic impulse at every tooth, and a worn or mis-tensioned chain produces larger impulses at the same frequency. A drive motor with a soft mount, a gearbox with backlash and a belt with a stiff joint all add their own component to the spectrum.
The forced convection fans are the second source. A fan that is out of balance, or a set of fans whose speeds beat against each other, puts energy into the oven structure and from there into the conveyor. A fan bearing that is beginning to fail raises the amplitude gradually, which is why a change in the defect rate can often be traced back to a maintenance event rather than to a material change.
How Much Displacement Matters
Useful numbers come from the board rather than from the oven frame. A displacement of 10 to 20 um at a frequency of a few tens of hertz is normal in a healthy oven, while a spike of 100 um during the time the alloy is liquid is enough to move a chip part. The time above the liquidus is what matters: an alloy that stays molten for 45 seconds gives the vibration 45 seconds to work.
That is why the same oven can run a large board without complaint and a densely populated small board with a persistent shift problem. The large board damps the motion and its components are heavier, while the thin small board is a good transmitter and its parts are light. Any change that shortens the time above liquidus, such as a tighter profile or a lower peak, reduces the exposure as well.
Small Packages and Tall Components
Chip parts below 0402 size and small outline transistors are the usual victims. Their pads are small, so the restoring force of the wetting solder is small, and their mass is low enough that a modest acceleration produces a large response. A component that has been placed with a deliberate offset to compensate for wetting is especially sensitive, because the offset is part of the design and the vibration consumes the margin.
<img src="https://www.gopcba.com/wp-content/uploads/2026/08/1024×1024-9.jpg" alt="Reflow oven conveyor belt with boards entering the heating zones” />
Tall parts behave differently. A connector or an electrolytic capacitor has a high centre of mass, so the same acceleration produces a larger moment at the solder joints, and the part can lean rather than slide. The standard placement discipline that keeps parts centred on their pads is described in the notes on placement offset control, and it is the first defence because a well centred part has further to travel before it fails an inspection limit. The flatness of the board under the same thermal load is covered in the notes on board warpage control.
Measuring Vibration at the Belt
The measurement has to be taken where the boards ride. An accelerometer taped to the oven frame reports the structure, not the belt, and the two can differ by an order of magnitude. A thin instrumented board, or a small logger that fits within the board outline, is placed on the belt and run through the oven at production speed while the acceleration is logged in all three axes.
The record should cover the whole transit and it should be compared against the same measurement taken when the oven was new. Frequency content is more useful than a single peak value, because a tooth frequency that grows over time points to the drive, while broadband energy that appears after a fan service points to the fans. A baseline taken once and never repeated is of little use when a defect appears two years later.
Oven Support, Leveling and Fan Balance
Mechanical fixes start at the floor. An oven on adjustable feet that are not all in contact, or on a frame that has been levelled to the floor rather than to the belt, transmits the floor vibrations of the building into the process. Anchoring the frame, shimming the feet and checking the belt for level across its width are inexpensive and often decisive.
Fan balance is the next item. Fans are balanced at manufacture and lose that balance as they collect flux condensate and dust, so a scheduled cleaning is a vibration control measure as much as a thermal one. Drive chains are tensioned to the oven maker specification, sprockets are checked for wear, and a belt with a repaired joint is replaced rather than run, because the joint produces a periodic impulse at every revolution.
Belt, Chain and Rail Systems Compared
A mesh belt supports the whole board and therefore damps the motion, but it also carries the board on a moving surface that can itself transmit the tooth frequency of the drive. A chain with edge fingers supports the board at its edges and leaves the middle free, which is better for double-sided assemblies and worse for thin boards that can sag between the supports.
A rail system with a pin chain and adjustable rails gives the most stable support in the centre, at the cost of setup time and of fixtures for narrow boards. The choice is a trade against the product mix, and the vibration behaviour should be part of the evaluation. Where a shift problem is confined to thin boards, the support strategy is often the fix rather than the oven.
Corrective Actions and Their Cost
Ordered by cost, the actions run from profile changes to hardware. Shortening the time above liquidus and lowering the peak within the alloy window reduce the exposure at no capital cost. Increasing the paste volume slightly raises the restoring force. Adding support pins or a carrier under the affected area damps the board. Servicing the drive and balancing the fans removes the source.
Only the last group addresses the root cause. A profile change that hides a shift defect also narrows the process window for every other requirement, and support pins have to be placed without touching a component. Where a defect is intermittent, it is worth recording the oven maintenance dates alongside the defect dates before deciding which action to take.
Verification and Records
Verification is a repeat of the vibration measurement plus a first-article inspection of the affected components. The measurement should be taken at the production speed and load, because both change the response. Where a carrier or a support pin has been added, the inspection has to confirm that the added support does not itself disturb the parts and does not shadow a zone.
The records that matter are the baseline vibration signature, the maintenance history of the drive and the fans, and the profile that was running when the defect appeared. The profile record should follow the discipline described in the notes on thermocouple placement, because a measurement taken at the wrong point on the board will not show the peak the component sees. Together those records turn an intermittent placement defect into a maintenance decision.
FAQ
Can vibration move a part before the alloy melts? Not usually. Paste holds the component until the alloy is liquid, so the critical interval is the time above the liquidus and the few seconds after it.
Is a smoother belt always better? No. The belt damps the board, but it also transmits the drive impulses. The measurement at the board is what decides, and different products can prefer different support systems.
Why does the defect appear only on thin boards? A thin board flexes and transmits the motion to the components, while a thick board damps it. Support in the centre of the panel is usually the answer.
