Developing a reflow profile for high power printed circuit boards follows the same procedure as for printed circuit boards. The challenge in creating a proper thermal profile for power boards is assessing the thermal density of boards and components and recognixing the thermal density variation on a given assembly. There are several parameters that must be taken into consideration when creating a thermal profile for high power applications. To illustrate this point, let's compare the EMPF-009 board and the high power ETO board, as shown in Figure 2-1. Both surface mount assemblies have been manufactured at the EMPF.

First, let's compare the components. The largest component on the EMPF-009 board is a 169 I/O PBGA. It measures 23mm x 23mm and is 0.54mm thick. The EMPF-009 board contains two 169 I/O PBGAs. On the ETO, one of the largest surface mount components is a power MOSFET, measuring 6.73mm x 6.22mm and is 2.38mm thick. The ETO assembly contains 68 power MOSFETs.

If the entire PBGA were made of plastic, with a thermal mass of 873 e-6 J/mm3 C, as shown in Table 2-1, 49.88J is required to raise its temperature from 20°C to 220°C (T= 200°C).Conversely, if the ETO's power MOSFET is made of ceramic, 59.06 J is required to raise its temperature 200°C.

Next, consider the board characteristics, as defined in Table 2-2. Both boards are made of FR4. However, the ETO occupies over 3.5 times the area of the EMPF-009. The ETO is a six layer board vs. two layers for the EMPF-009 board. The ETO has two copper ground plans with 4oz copper while half the EMPF-009 board contains a single ground plane with 1oz copper. For the ETO to raise its temperature from 20°C to 220°C (T= 200°C), it is estimated that it will require 9,469.97 Joules. The EMPF-009 board will require 556.43 Joules. If the EMPF-009 board was the same size as the ETO Board, it would require 1,969.44 Joules still significantly less than the ETO's heat requirements. This does not include the effect caused by the quantity and types of components on each respective board.

When comparing the soldering process, there are several differences in the equipment set-up. Using the same reflow oven for both the ETO and the EMPF-009, Tables 2-3 and 2-4 show the temperature settings for the oven's heating panels, the Cooling Zone setting, belt speed, convection rate, and nitrogen usage. Both assemblies use nitrogen, but that is where the similarities end. In general, the ETO's panel zone temperatures were set higher than the temperatures for the EMPF-009. When running the ETO, the oven's convection rate was set at high versus medium for the EMPF-009. The belt speed for the ETO was 5 inches/minute slower than for the EMPF-009 board. All of the factors had to be modified to take into account the difference in thermal mass between the two assemblies, as shown in Table 2-2.

Of interest is the zone 7 temperature setting. The ETO's temperature setting in zone 7 is 270°C, which is 10°C less than the maximum panel temperature setting of the oven in question. If the ETO assembly were larger and more densely populated, it is possible that this oven would not have the thermal capacity to solder the unit. Reducing the belt speed may provide more thermal capacity, but it would reduce hardware output.

With the increase in temperature, and temperature exposure time due to the slower belt speed, it is possible that component reliability may be compromised. Components with low thermal masses may reach temperatures that they are not rated for. The higher exposed temperatures can increase the component's moisture sensitivity, based on IPC/JEDEC J-STD-020A, Moisture / Reflow Sensitivity Classification for Plastic Integrated Circuit Surface Mount Devices. Component delamination per Figure 2-2, "popcorning", and electrical failures can occur when components are exposed high temperatures during the reflow soldering process.

If the ETO assembly were to be manufactured with lead-free solders, which typically require a reflow soldering peak temperature 20°C to 40°C higher than for its Tin Lead (SnPb) counterpart, this oven may not be able to support a Lead Free ETO production run.

Developing a thermal profile for power circuits has the same challenges as with circuit boards. Typically, a thermal profile is broken up into 4 components, as shown in Figure 2-3. These components are defined as:

Initial (preheat) Ramp: This zone starts the soldering process. During this time, the hardware's temperature increases from ambient. As the temperature increases, the solder paste loses its volatiles. Too fast a ramp rate will cause the solder paste to splatter resulting in solder balls. Also, a prohibitively high ramp rate could result in the delamination of the board and plastic components (popcorning). An acceptable ramp rate is 2°C/second to 4°C/second. This rate is specified in the solder paste manufacturer's data sheet.

Flux Soak Zone: During this period, the temperature gradient of components with different thermal masses is reduced. Also, within this zone, this allows the solder flux and activators within the solder paste to clean the surfaces to be soldered, promoting good solder wetting.

Reflow Zone: This is also known as the "thermal spike", where the hardware's temperature is driven 20°C to 40°C above the solder's melting point for 30 to 90 seconds. By this time, all the solder joints are in liquid form. It is also during this time period that the solder joint is formed. Depending upon the component’s weight, the solder joint’s surface tension can snap the component back into proper alignment.

Cool Down Ramp: This is where the temperature is lowered below the solder's melting point. As with the Preheat Ramp, an acceptable ramp rate is 2°C/second to 4°C/second This rate is specified in the solder paste manufacturer's data sheet.

Failure to properly create a thermal profile for either of these assemblies could have caused several problems. For instance, soldering may not have occured because the temperature is too low, or inadequate preheating may have led to solder balls being caused by solder paste splatter. Too high a profile could have resulted in solder residues being baked onto the assembly, becoming difficult to remove.

With proper preparation and process development, one can achieve the same quality levels with a high density assembly as you could with a less dense piece of hardware. High power printed circuit boards have a higher thermal mass than their digital circuit counterparts due to their heavier construction. A critical activity when developing the process is creating the thermal profile which supports the higher density electrical hardware.

There is a caveat - too much heat applied results in higher soldering temperatures that can cause hardware delamination and electrical failures. To prevent these types of failures from occurring, component and board preparation, such as baking the hardware prior to soldering to remove moisture, might be required.