I. Introduction. Hand soldering of smaller multi-leaded SMA components can be extremely challenging. Hand soldering of ball grid array (BGA) devices is even more problematic. And complex projects involving dozens of SMA components, even though they can be done by hand soldering, can be extremely tedious.

For these reasons, as my projects have become more complex and have involved ever smaller components with more and more leads per component, I have wanted a better solution than hand soldering. After looking at possible alternatives to hand soldering such as hot air guns¹ and hot plates², I decided that a reflow oven³ would be the best choice for these projects with one or more of the following characteristics: very small SMA components, BGA devices, a high device count.

I looked at commercially available reflow ovens, and found that they were priced starting at $300⁴, with most of the units that looked reasonable in terms of quality, reliability, and features being priced at $600⁵ and up. I completed this project in 2013. At that time there had been one KickStarter reflow oven project, called RefloLeo⁶. Since then there have been several more, including Zallus⁷, Reflowster⁸, ControLeo2⁹, and several others¹⁰, some of which were canceled¹¹.

In addition, there were then and now are an even greater number of webpages describing homebrew reflow ovens, all based on simple toaster ovens¹² ¹³ ¹⁴ ¹⁵. I have given references for four such webpages, but there were dozens available. Most of these projects had in common three key features:

In addition, I noted that there were a number of discouraging reports on the less expensive commercial ovens (e.g the T962A), with reported problems including uneven heat distribution, burning of PCBs, etc¹⁶ ¹⁷. Although some solutions to these problems were offered, it seemed to me to make more sense to look for a system that did not have sub-optimal results requiring correction in the first place.

I was not enthusiastic about the KickStarter projects because each of them was in some way limited, and I felt that I could myself assemble a more versatile, more extendable controller / reflow oven.

So I decided to make my own reflow oven, using a PID¹⁸ (Proportional-Integral-Derivative) controller to control the temperature. After reviewing the various projects found on the web I decided to use the osPID¹⁹ as my PID controller because it was very flexible, easy to use, and had a nice software package backing it up²⁰, and because a user reported good results using it for a reflow oven controller²¹. Furthermore, because the osPID is open source, I had confidence that I could write new software for it if necessary (it turned out that was not necessary).

The name of the osPID is derived from “os” for open source and PID for Proportional-Integral-Derivative. The osPID is a collaboration between Rocket Scream Electronics, creators of the Arduino Reflow Controller Shield²², and Brett Beauregard, creator of the Arduino PID Library²³. It consists of an Arduino compatible board, an 8 x 2 alphanumeric character LCD display, a 4 push button navigation interface, a USB converter chip for communication with a PC, and input and output cards including a type K thermocouple input port, all enclosed in a PID form factor case²⁴.

I decided to use a Panasonic NB-G110P Flash Xpress Toaster Oven, which I obtained from Amazon for $101.99. Many of you will have an old toaster oven. If you use that, you will reduce the cost of this project by 50%. I chose this oven because it was used with good results by someone who had built a reflow oven using the osPID as a controller, so I knew that it would do the job.

I got my Solid State Relay (SSR) from Amazon. It was an Amico SSR-40 DA Single Phase Solid State Relay SSR 40A 3-32V DC 24-380V AC w Heat Sink and it cost $11.45. I could have gotten one from eBay for less than $5 without a heatsink²⁶.

The Type K thermocouple was a model SEN-00251 Thermocouple Type-K Glass Braid Insulated from SparkFun, which cost $13.95²⁷.

So my cost of major components, including toaster oven, was $212.39. For this price I got a unit that I believe is better than commercial units costing several times my expenditure, and I have the pride of using a homebrew creation rather than a commercially produced appliance when I reflow solder my homebrew PCB projects.

If you have your own toaster oven, then your costs would be limited to the OSPID, the SSR, and the thermocouple, the cost of which totals $110.40.

From my junkbox I got the cabinet into which I installed the SSR and the connectors I used to connect the OSPID and the toaster oven to the SSR.

II. Theory. Reflow Soldering. Reflow soldering is the process in which solder paste is used to temporarily attach electrical components to contact pads on a circuit board, after which the entire assembly is subjected to controlled heat which melts the solder and permanently connects the component to the board physically, and with an excellent low-resistance electrical connection as well.

There are four consecutive stages to reflow soldering. These phases are generally called preheating (or ramp-to-soak), soak, reflow, and cooling. The picture below shows a reflow profile for Kester Easy Profile 256 No-Clean Solderpaste (EP256), which is what I use.

As you can see in the image on the left, for Kester EP 256 the preheat stage lasts 90 seconds, and takes the temperature from 20 C up to 150 C. During this stage the solvent in the solder paste begins to evaporate. If the rate of temperature rise is too low, evaporation of the volatile components of the flux is incomplete. If the rate of rise is too fast, it can lead to component damage, cracking, and spattering of solder paste.

For EP256, the soak stage runs from 90 seconds through ~190 seconds. During this stage evaporation of the volatile components of the flux is completed and the flux is activated, initiating oxide reduction on component leads and PCB pads. If the temperature during this phase is too high oxidation of the paste, pads and component leads can occur, as well as solder spattering and balling. If the temperature is too low, then the flux may not activate fully.

For this solder the reflow stage extends from ~190 through ~265 seconds. During this stage the solder becomes liquid (melts) and surface tension is reduced, allowing metallurgical bonding and giving a good solder joint. Maximal acceptable peak temperature is determined by the component with the lowest tolerance for high temperatures. Excessive peak temperature, in addition to damaging components, may foster intermetallic growth and brittle solder joints, and cause PC board damage. Insufficient peak temperature may prevent adequate reflow. If the time of this stage is too long, then flux may be prematurely activated or consumed, leading to a dry solder joint. Excessive length of this stage may also cause intermetallic growth and brittle solder joints, just as does excessive peak temperature. If the duration/temperature for this stage are insufficient, then inadequate removal of solvent and flux, as well as inadequate reflow may result, leading to cold, dull, defective solder joints as well as solder voids.

The cooling stage for EP256 begins at ~265 seconds. During this stage the board and components are gradually cooled to ambient temperature. A cooling rate of 4C/sec is often suggested. Proper cooling rates minimize the chances of thermal shock and intermetallic formation.

Oven and Heating Elements. The goal when building a reflow oven is to have an even heat distribution within the oven that follows as closely as possible the recommended reflow profile for the solder paste being used, so that one can avoid the problems described above that occur when the profile is not closely followed. To accurately follow the profile the oven needs to have adequate heat generation to be able to ramp up the temperature rapidly enough to follow the suggested temperature vs time curve during the preheat and reflow stages. Furthermore, the oven needs to have an even distribution of temperature, to avoid the situation where some portions of the board are exposed to excessive temperatures while other portions are inadequately heated. Additionally, the oven must not have excessive thermal inertia, which would prevent an appropriate rate of cooling during the cooling stage.

Size is an important consideration. Too small and you won't be able to fit into the oven your project boards. But if the oven is too large, there is more potential for uneven heat distribution, and for inadequate temperature ramp-up. Most toaster ovens use infrared heating elements, and these are the preferred heating elements. Single-element ovens are problematic, as they tend to lead to very uneven temperature distribution, leading to component damage and inadequate solder joints. Dual element ovens, with one element at the top of the oven and the second element below, provide more even temperature distribution. The toaster oven that I chose to use for this project is a dual element oven. Quad element ovens potentially provide even more uniform temperature distribution within the oven. In general quad element ovens will be more expensive than dual element ovens, and I have had good results with the dual element oven I chose.

The next issue is how much heating power is needed to achieve adequate results. Based on what I have seen on the web, it appears that 1300-1500 watts is sufficient to achieve a good temperature profile. My oven is rated at 1300 watts and provides excellent performance.

The final considerations are whether or not to get an oven with a fan, and whether or not to add a fan to an oven without a fan. Having a fan might reduce temperature gradients within the oven cavity. But if it causeed cooler outside air to be sucked into the oven cavity, it might exacerbate temperature gradients rather than reducing them. Additionally, if the fan produces too much air movement, it might cause components or boards to shift position, which would be undesirable. Finally, having to control both a fan and the heating elements might complicate the control circuits needed for this project, because one might want separate control over the fan and the heating elements. I chose to use an oven without a fan, and to not add a fan.

Controller. The PID²⁸ (Proportional-Integral-Derivative) controller is the heart of this project. A PID controller is a control loop feedback mechanism that attempts to achieve the desired output (in this case temperature) based on three factors: (1) the difference of the current variable value (in this case temperature) from the desired result (the proportional term), the integral of past errors (differences of the measured temperature from the desired temperature) over time (the integral term), and a prediction of future error, using the rate of change of the error (the derivative term).

Commercial/industrial PIDs are easily obtainable. I chose to use the osPID for this project rather than a commercial/industrial unit because of the osPID's versatility. The osPID uses an Arduino Uno or Duemilanove compatible device along with Brett Beauregard's PID Library (referenced above) to give a fully functional PID. Because the osPID is based on the Arduino and open source software, it is extendible and modifiable in a way that a commercial/industrial PID cannot be.

During usual operation of the osPID with the provided software, the user has full control over the three typical PID tuning parameters Kp, Ki, and Kd which respectively control the magnitude of the effect of the proportional, integral, and derivative error values on the output of the PID.

The osPID consists of a main board which contains an Auduino-compatible CPU as well as the LCD display, USB jack, and four input buttons on its front side and two edge connectors for input and output cards on its back side. The input card will receive temperature input from either a thermistor or a type K thermocouple sensor. The output card has two relay outputs and one digital output for connecting an external solid state relay (SSR).

Pictures of the osPID with and without its covers are below.