On The Forefront: June, 1997
by Phil Zarrow
Fear and Loathing with Double Sided Reflow Soldering
Let’s start off by elaborating on an important facet of reflow profiling that was presented in the April On the Forefront column. When you set out to profile a PCB assembly, each application will have a Reflow Specification. The reflow specification is comprised of two temperatures which represent the minimum and maximum thermal excursions that should be present during the course of reflow soldering of that assembly.
The first temperature, of concern is the Full Liquidus Temperature, also known as the Minimum Reflow Temperature (T1). This is the temperature at which the liquidicity of the molten solder is ideal for flowing across the metal surfaces onto which it will wet and form the solder joint. It is determined primarily by the actual alloy of the solder paste but can be affected by the solder sphere size and other factors of the formulation and may be stated as a range on the data-sheet. For example, for Sn63Pb37, the ranges average from 200 - 225 deg. C. The minimum value of the range given for the specific solder paste being used is becomes the minimum temperature each interconnection being reflow soldered in that cycle must attain - no less. Again, this will typically be around 15 to 20 deg. C above the melting point of the solder. It is a common, erroneous assumption that only the melting point of the solder must be reached. That kind of thinking will get you in trouble.
The second component of the Reflow Specification is the Most Vulnerable Component temperature (T2). As the name implies, this corresponds to the component on that assembly with the lowest threshold of thermal pain. Whatever that component is and whatever temperature it gets into trouble at, subtract a buffer of 5 deg. C from that temperature and the resulting value becomes the MVC. This is the highest temperature you want any part of the assembly to reach. Staying below this will keep you out of trouble. The Most Vulnerable Component may be a connector, a DIPswitch, a LED, or it could even be the substrate material or even the solder paste. It obviously varies from application to application and may require enlisting the assistance of Component Engineering personnel to help research it.
Now that the range of temperatures at the peak of the reflow cycle have been established, note also that the maximum tolerable gradient across the assembly is also determined (T2 - T1). Whether you can stay within that band depends upon such factors as the mass and surface geometry complexity as well as the substrate composition of the assembly and the thermal transfer efficiency of the oven being used Ideally, a minimum gradient is desired with peak temperatures across the board as close to T1 as possible. This helps in reducing the liquidus dwell duration as well as the overall exposure of the assembly to high temperature excursions.
Beyond this, reflow profiling is a matter of minimizing the liquidus dwell duration and matching the time/temperature intervals to what the solder paste manufacturer requires and specifies. Regarding the rate of heating, most practitioners run at 4 degrees C per second or less. A good practice is to keep your rate of cooling the same or less than your rate of heating to avoid thermal shocking the components - the speed limit that applies going up should be the same going downhill.
With that said, let’s get on to the matter at hand - reflow soldering of double-sided assemblies.
One of the early-realized thrills (in a manner of speaking) with surface mount technology, was that components could be mounted on both sides of the substrate. The dilemma soon came up that, assuming that the second side of the assembly is to be reflow soldered, as opposed to wave-soldering, how does one keep the previously reflowed components on the inverted side of the assembly. After all, those solder joints were going to be re-reflowed and won’t there be a problem with the parts falling off when the solder becomes molten again?
A number of approaches have been taken to accomplish double-sided reflow. One method used is to glue the components to the assembly (just as is done with wave-soldering passives, SOTs and SOICs). However, this involves the additional steps and equipment dispensing the adhesive and then curing it. This is great if you happen to be in the business of manufacturing automatic dispensing systems and curing ovens. This, however, is a common method.
What about using two different solder alloys for top and bottom with the second side material having a lower melting point? Yes, that’s done, too. For example the first side might be soldered with Sn63Pb37 (with a melting point of 183 deg. C) and the second side soldered with Sn42Bi58 which melts at 138 deg. C (and is eutectic). While that will work for some applications, many others may find the tin-bismuth alloy restrictive either because the melting point is too low with respect to the service temperature that the product will be residing in (it is definitely a “bad-hair day” when the product reflows during use) or perhaps the alloy is too brittle for the service environment. Using a higher temperature alloy on the first side may be difficult with respect to endangering the components or substrate. Remember, we have to take the board at least 15 deg. C higher than the melting point of the alloy to the Full Liquidus Temperature.
How about turning off the lower bank of emitters on the oven and only heating the top-side of the board? Ask any reflow oven manufacturer and they will tell you that they get at least two requests per week to pull this stunt. You see, it’s not that easy to do[1]. In each vertical heat zone we are trying to present a near-equilibrium environment in terms of heat transfer. To do so selectively as such, is not an easy feat. Think about it -we’re trying to get a low gradient across the width and length of a board yet trying to present a 30 degree C gradient between top and bottom. But believe it or not, it is possible.
I tested the first such oven to manage this, a couple of years ago. It was designed and built in France and not really distributed in North America. It was quite innovative and the way it worked was when a top-to-bottom differential was desired, the operator hit a switch and the bottom bank of emitters lowered hydraulically about a foot (the Citroen of ovens). Cool air would be blown across the bottom side of the board and indeed, an differential was created. Unfortunately, the oven was Convection/IR with a large IR component and the top-side gradient was nothing to write home about. Nevertheless, the concept (of blowing cooler gas across the bottomside of the assembly - not lowering the emitters[2]) has been implemented by a number of oven manufacturers.
Do we really want to do this? What about the stresses we are imparting into the substrate when we create such a gradient in the Z-axis? What is the effect upon vias and internal layers? Not really knowing the answers, the oven manufacturers are merely responding to the demand of the market. In some applications, such stress will likely have nominal effect but one is advised to proceed with caution.
Actually, there is a far more practical solution for many applications than any of the approaches discussed thus far. Don’t underestimate the adhesion of the molten metal. It is far stronger than that of the solder paste. Understanding this, the greater the bonding surface area of a component (lead to pad sum area) the greater the force in place to keep it from falling off.
To determine which components are candidates for bottom side attachment and consequential re-reflow, a ratio that evaluates the mass of the component in relation to the lead/component pad contact area was derived[3]:
Formula for Secondary Side Mounting
Weight of components in grams
Total pad mating area in square inches
Grams per square inch must be 30 for secondary side mounting
Table 1: Lead / Component Pad Contact Area Ratio
Try this one at home, gang. While there are more sophisticated and exacting formulas out there, experiment with this one as it applies to the components on your application. Obviously more massive components will have to be glued or, perhaps, selectively soldered. When a new component is being considered for bottom side attachment, this calculation should be done prior to approving the product design. The formula to complete the ratio calculation is included in the printed circuit board design standard, so the designer of the board can complete the calculation during the initial design layout.
There you go. This is a lot easier than specifying multiple solder paste alloys or trying to do weird things with your reflow oven. It’s also lower risk with respect to your materials and, with elimination of machine and process centers (glue and cure) it’s a great way to increase productivity. And isn’t that what it’s all about?