Welcome: ADD Components Limited
Language:
∷
∷
∷
Getting the temperature profile wrong is the fastest way to kill a semiconductor. A diode that survives soldering can still be dead because the junction got too hot for too long. A transistor that looks perfectly soldered can have shifted gain because the thermal cycle was aggressive. The temperature curve is not just about melting solder. It is about managing how fast heat enters the component, how long it stays there, and how fast it leaves. Every semiconductor package has a thermal limit, and exceeding it even briefly creates damage you cannot see with any inspection tool.
Resistors and capacitors can take a lot of abuse. A thick-film resistor survives 260 degrees Celsius for ten seconds without blinking. A semiconductor cannot. The difference is the junction.
Inside every diode, transistor, or IC, there is a tiny junction of doped silicon. That junction defines the electrical characteristics of the part. If the junction temperature exceeds its rated maximum, even for a fraction of a second, the dopant profile shifts. The part still works. It still passes a continuity test. But its forward voltage, its gain, its leakage current, all of it drifts. You will not catch this with a multimeter. You will catch it six months later when the circuit behaves differently than the simulation predicted.
For silicon diodes, the maximum junction temperature is typically 150 to 175 degrees Celsius. For germanium devices, it is lower, around 100 to 125 degrees Celsius. For power transistors in TO-220 packages, the junction can handle more, but the wire bonds inside are the real limit. Those bonds fail at around 150 degrees Celsius if held for more than a few seconds.
A 0402 resistor has almost no thermal mass. It heats and cools in under two seconds. A TO-220 transistor has a massive metal tab and a big silicon die. It takes five to ten seconds to reach soldering temperature, and it holds that heat for a long time after you remove it from the oven. If you use the same profile for both, you will underheat the transistor and overheat the resistor. Every package needs its own curve, or at least a curve tuned to the largest component on the board.
Most people start with the solder melting point and work backward. That is backwards. Start with the component limit and work forward.
Check the datasheet for the maximum junction temperature, labeled as Tj max. This is your ceiling. You never want the junction to reach this temperature. Your target peak junction temperature should be at least 20 degrees below Tj max to give yourself a safety margin. For a silicon diode with a Tj max of 175 degrees Celsius, your target peak is 155 degrees Celsius or lower.
The junction temperature is not the same as the board temperature. The junction is inside the package, and it is always hotter than the board surface. The difference depends on the package thermal resistance, labeled as Rth j-c or Rth j-b in the datasheet. A small SOT-23 package might have a thermal resistance of 200 degrees Celsius per watt. A large TO-220 might be 2 to 5 degrees per watt. Multiply the thermal resistance by the power dissipated during soldering to estimate how much hotter the junction is than the board. This number tells you how much headroom you need in your profile.
Once you know your junction limit, work backward to find the maximum board temperature you can use. For leaded solder, the liquidus is around 183 degrees Celsius. For lead-free SAC305, it is around 217 degrees Celsius. Your peak board temperature needs to be 20 to 30 degrees above the liquidus to ensure full melting and good wetting. For leaded, aim for 205 to 215 degrees Celsius. For lead-free, aim for 235 to 245 degrees Celsius.
But here is the catch: that peak temperature is what the thermocouple reads on the board surface. The junction inside the semiconductor is hotter. If your thermocouple reads 220 degrees Celsius and the package thermal resistance is 50 degrees per watt, and the component is dissipating even 0.5 watts during heating, the junction is at 245 degrees Celsius. That might be over the limit. You have to account for this in your profile setup.
Every reflow profile has four zones. Each one matters for semiconductors in a different way.
The preheat zone ramps the board from ambient to around 100 to 150 degrees Celsius. For semiconductors, this zone needs to be slow. A ramp rate of 1 to 3 degrees Celsius per second is ideal. Going faster than 4 degrees per second creates thermal shock inside the package. The silicon die, the wire bonds, and the lead frame all expand at different rates. A fast ramp creates stress that can crack the die or break a wire bond before the solder even melts.
The soak within the preheat zone is just as important. Hold the board at 100 to 150 degrees Celsius for 60 to 120 seconds. This allows the entire board to reach thermal equilibrium. Every component, every pad, every trace gets to the same temperature. Without soak, the large components are still cold while the small ones are already hot. That temperature difference is what kills semiconductors.
The soak zone sits between 150 and 200 degrees Celsius. Hold for 60 to 90 seconds. This is where the flux activates and removes oxides from the pads and leads. For semiconductors, this zone also gives the junction time to catch up with the board temperature. If you skip soak or make it too short, the junction lags behind the board. When you hit the peak, the board is at 240 degrees but the junction is still at 180. The sudden jump when the junction finally catches up is what causes damage. A proper soak eliminates that lag.
The reflow zone ramps from the soak temperature to the peak in 30 to 60 seconds. The ramp rate here can be faster than preheat, around 2 to 4 degrees per second. Once you hit the peak, do not dwell. The time above liquidus should be 30 to 60 seconds for leaded solder and 45 to 90 seconds for lead-free. Every second above the peak temperature is another second of stress on the junction. A long dwell does not improve the joint. It only increases the risk of junction damage.
For semiconductors, keep the peak dwell at the minimum that still gives you good wetting. If your solder wets properly in 30 seconds, do not hold for 60 just because the profile says so.
The cooling zone is the most critical part of the profile for semiconductors, and most people get it wrong. The board must cool from the peak down to 100 degrees Celsius at a rate of 2 to 4 degrees Celsius per second. Faster than 6 degrees per second creates thermal shock. The silicon die contracts faster than the lead frame, and the stress cracks the die or breaks wire bonds. Slower than 1 degree per second is wasteful and can cause excessive intermetallic growth in the solder joint.
For power semiconductors in TO-220 or D2PAK packages, the cooling rate needs to be even slower, around 1 to 2 degrees per second. The large metal tab acts as a heat sink that keeps the junction hot long after the board has cooled. If you cool the board fast, the junction is still at 150 degrees while the board is at 80. That 70-degree difference across the package creates massive stress. Slow the cooling down and let the junction follow the board temperature gradually.
One profile does not fit all. Here is how to tune it for the most common discrete packages.
These are tiny. They have low thermal mass but high thermal resistance. The junction heats up fast relative to the board. Use a peak board temperature of 220 to 230 degrees Celsius for lead-free solder. Keep the total time above 200 degrees Celsius under 45 seconds. The preheat ramp should be 2 degrees per second. The cooling rate should be 3 to 4 degrees per second. These parts are fragile. A profile that works for a resistor will destroy a SOT-23 transistor.
TO-92 packages are bigger than SOT-23 but still sensitive. The glass or plastic body has low thermal conductivity, which means the junction lags behind the leads during heating. Extend the soak zone to 90 seconds to let the junction catch up. Peak temperature can be 230 to 240 degrees Celsius for lead-free. Cooling rate should be 2 to 3 degrees per second. The longer body creates more thermal mass, so the cooling needs to be gentler than for SOT-23.
These are the toughest packages but also the most dangerous to get wrong. The large metal tab conducts heat efficiently, which means the junction follows the board temperature closely. That sounds good, but it also means the junction stays hot for a long time during cooling. Use a peak of 235 to 245 degrees Celsius. Extend the soak to 120 seconds. The cooling rate must be 1 to 2 degrees per second. If you cool a TO-220 at 4 degrees per second, the junction will crack within a few hundred cycles. For power devices, slow cooling is not optional. It is the single most important parameter in the entire profile.
These are similar to SOT-23 in thermal behavior but the junction is more sensitive because diodes have a lower thermal budget than transistors. Keep the peak board temperature at 215 to 225 degrees Celsius for lead-free. Total time above liquidus should not exceed 40 seconds. The cooling rate of 3 degrees per second is fine for these packages.
You cannot measure junction temperature directly without embedding a thermocouple inside the package, which is impractical for production. But there are ways to verify.
Place a semiconductor with a known thermal resistance on a test board. Attach a thermocouple to the pad nearest the component using high-temperature tape. Run your profile and record the thermocouple reading. Compare it to the datasheet thermal resistance to estimate the junction temperature. If the estimated junction temperature is within 20 degrees of Tj max, your profile is too aggressive. Back off the peak or slow the ramp.
Solder a batch of semiconductors using your profile. After cooling, measure the forward voltage of diodes, the gain of transistors, and the leakage current of all devices. Compare these values to the datasheet specifications. If the forward voltage of a silicon diode has shifted by more than 5 percent, the junction was overheated. If the gain of a transistor has dropped by more than 10 percent, the profile was too hot or the dwell was too long. Parametric drift is the most reliable indicator that your profile is damaging the junctions even when the joints look perfect.
Take a sample of boards soldered with your profile and run them through thermal cycling, typically from -40 to 125 degrees Celsius for 500 to 1000 cycles. After cycling, re-measure the semiconductor parameters. If the values have drifted significantly, the soldering process weakened the junctions. If they are stable, your profile is within the safe window. This test takes time but it catches problems that no thermocouple can find.
This is the most common mistake. The profile that works perfectly for 0603 resistors and ceramic capacitors will overheat a SOT-23 transistor. The peak is too high, the ramp is too fast, and the cooling is too aggressive. Semiconductors need their own profile. If your board has mixed components, tune the profile to the most sensitive semiconductor on the board, not to the passive components.
Most people look at Tj max and stop there. They do not check the thermal resistance values. Without thermal resistance, you cannot convert board temperature to junction temperature. You are flying blind. Always check Rth j-c and Rth j-b in the datasheet. These numbers tell you how much hotter the junction is than the board, and that difference is what determines whether your profile is safe or destructive.
Lead-free solder requires a higher peak temperature, typically 20 to 30 degrees higher than leaded. If you use a leaded profile with lead-free solder, the solder will not fully melt, and you will get cold joints. If you use a lead-free profile with leaded solder, the higher temperature will overheat the semiconductor junctions. Always match the profile to the solder alloy, not the other way around.
Soak is the zone that lets the junction temperature equalize with the board temperature. Skipping it saves 60 to 90 seconds per board, which feels like a lot in high-volume production. But without soak, the junction lags, and when the reflow zone hits, the sudden temperature jump cracks the die. The 90 seconds you save costs you ten times more in field failures. Do not skip soak. Ever.
Even with a good starting point, you will need to adjust based on real results.
When a joint does not wet properly, the instinct is to crank up the peak temperature. For semiconductors, this is the wrong move. A cold joint on a semiconductor is almost always a flux or pad contamination issue, not a temperature issue. Clean the pads, check the flux activity, verify the solder paste volume. Raising the peak will damage the junction before it fixes the joint.
If your parametric tests show drift, drop the peak temperature by 5 to 10 degrees Celsius and re-run. If the drift goes away, you found the problem. If it does not, slow the ramp rate in the preheat zone. A slow ramp gives the junction more time to follow the board temperature and reduces the thermal gradient across the package.
Tombstoning on SOT-23 or SOD-123 packages usually means one end wetted before the other. The first end to wet anchors the part, and surface tension pulls the other end up. Slow the preheat ramp to 1 to 2 degrees per second. This gives both ends time to reach soldering temperature at the same time. Also check your solder paste volume. Too much paste on one pad causes that pad to wet first, which triggers tombstoning.
One detail that often gets overlooked: the conveyor speed in a reflow oven affects the profile more than people realize. A faster belt speed shortens the time in each zone. If you speed up the line to increase throughput, the soak time drops and the cooling rate spikes. The profile that worked at 1 meter per minute might destroy semiconductors at 1.5 meters per minute. Every time you change the line speed, re-validate the profile with a thermocouple and a parametric test. Do not assume the profile still works just because the oven settings did not change.
Contact: Joanna
Phone: Info@addcomponents.hk
Tel: 852 5334 3091
Email: info@addcomponents.hk
Add: FLAT/RM C -13/F HARVARD ,COMMERCIAL BUILDING 105-111 THOMSON ROAD,WAN CHAI HK