Tutorial: reflow soldering


The purpose of this tutorial is to give the reader a basic understanding of the steps involved in automated soldering of printed circuit boards (so-called reflow soldering) using the popular infrared reflow oven T-962 from the Chinese manufacturer Puhui Technology Co. Ltd. available online for less than EUR 200,-.

Reflowing printed circuit boards is clever, since much smaller components can be used along with more dense designs. The goal is to have consistent quality with respect to mechanical and electrical properties between printed circuit boards. Electrical connections are established by reflowing solder paste and solidifying the solder components.


There are many potential caveats and plenty of parameters that need some attention, if not optimisation, with the T-962 reflow oven, before a decent looking printed circuit board can be made. However, the modest investment in the oven itself along with about three hours of your spare time will get this oven to produce professionally-looking printed circuit boards with very little effort.

Necessary modifications including firmware upgrade are explained in this excellent video tutorial by Nick Williams. The following provides a few snap-shots from my own modification of a T-962.

The most surprising experience, however, is that the T-962 reflow oven should not be used straight out of the box. Apparently, the low price is also reflected in the making of some of the internal technical solutions.

A factory painted metal enclosure typically has a high risk of not being grounded correctly. This was also true in the case of my T-962. The solution is simple: scrape off same paint and create a firm connection between the different parts of the metal casing and make sure that this ground connection is available on the mains power cord.

Another imperative technical beef is the lack of cold-junction-temperature compensation, which has been hard-coded in the software in stead. In essence this implies that the temperature controller, sitting on top of a hot oven, always assumes that the printed circuit board, where the K-type thermocouples are terminated., has the same temperature. No matter, where the reflow oven is situated or how it is used.

Luckily, the good people at the company Unified Engineering of Sweden AB have released a much improved firmware that greatly optimises the performance of the T-962 reflow oven. Flashing this firmware to the onboard NXP microcontroller is seamless and offers many advantages.



First part of this tutorial will be taking the T-962 reflow in operation straight out of the box. Later in this tutorial, recommendations for reflow profiles and other modifications will be proposed based on my own experiences with reflowing printed circuit boards with the T-962.


Inspecting the T-962 reflow oven right out of the box gives a good first impression. The unit is presentable and appears sturdy enough for hobby use as well as prototype development in smaller businesses. The drawer will accommodate twelve circuit boards with dimensions 50 mm by 50 mm, so this will greatly speed of my production of Arduino Uno R3 compatible shields (for instance the 0x0E). There is no ventilation/exhaust connection, so all obnoxious fumes will probably be emitted from the area around the drawer.

T-962 right out of the box


Removing screws holding the top in place and lifting off the upper part of the chassis enables us to have a look inside (mains power not connected. Yet).

The inside of the T-962 right out of the box


The space is reasonably well-used and the space is tidy without loose wires hanging around or too much glue. One disturbing element, however, is the use of masking tape as thermal insulation. This is asking for a lot of obnoxious fumes, when the oven is in use.

Excessive use of masking tape for heat insulation


Masking tape is replaced by Dupont’s Kapton tape, which is poly-imide based and in essense quite thermo-resistent. Here I used quarter of a roll in total for the entire oven.

Masking tape replaced with Kapton tape 

Removing the ground connection to the chassis reveals that paint is effectively hindering a legal ground connection. Here I scraped off some paint with a screw driver to illustrate a connection to bare metal.

Ensuring proper ground connection to chassis 


Another wire is added and the ground connection is secured tightly.

Adding another ground-connected wire


The other end of the newly added wire is mounted on a terminal on the upper part of the chassis. Here I connected it to one of the mounting screws on the fan. Secure the connection tightly.

Grounding the upper part of the chassis

For cold junction temperature compensation of the K-type thermocouples inside the oven, a digital temperature transmitter type DS18B20 from Maxim Integrated is used. It comes in a standard TO-92 encapsulation and is easy to install. Furthermore, it makes use of the OneWire protocol, which is quite clever here also. The following image shows the transmitter wired for Parasite Power Mode.

Preparing DS18B20 temperature transmitter


For the Danish-reading audience, Poul-Henning Kamp has an interesting blog post about this particular sensor.


The DS18B20 temperature transmitter is glued onto the terminal block, where the K-type thermocouples are terminated. The two shorted pins are soldered onto the printed circuit board (I removed a bit of solder mask with a screw driver to expose bare copper).

Establishment of CJT measurement point on circuit board


The DQ pin on the DS18B20 temperature transmitter is wired to a free solder pad on the circuit board and a 4k7 pullup resistor (0805 or 1206 form factor) is added.

Connecting DS18B20 DQ-pin to 0805 sized 4k7 pullup resistor


The small fan on the side of the T-962 reflow oven is always on per default. It is noisy and annoying. Luckily, this can be fixed easily. Here, an NPN transistor type BC547 in TO-92 encapsulation, along with a 4k7 base resistor, is added to control the fan. Kapton tape is used for electrical insulation. The control signal comes from a vacant pad on the circuit board.

Adding BC547 with base resistor to control internal fan by PWM signal


Five female-female Dupont wires are connected to the ICSP (In Circuit Serial Programming) header on the circuit board. This enables the firmware to be overwritten.

Adding five female-female Dupont wires to microcontroller’s ICSP header


Dupont wires are brought out the chassis, the top is loosely put on, mains cord is connected, and the reflow oven is now ready for a firmware upgrade.

Fixing lid, before powering up the T-962


The programmer interface is a simple USB bridge with the CP2102 controller. Two switches are added to put the NXP ARM7 LPC2134 microcontroller in bootloader mode as well as resetting it after firmware upgrade. I used Flash Magic for flashing the firmware along with this HEX file.

Connecting ICSP connections to CP2102 USB bridge and external bootloader switches


Upon successful firmware upgrade, flipping the reset switch will make the Unified Engineering of Sweden AB splash screen appear on the display. Here the firmware is version 0.5.1.

New splash screen appears after successful firmware upgrade


Now it is time to run some performance tests.  I developed a series of simple test patterns for surface mount components (single boards with 0805, 1206 or SOIC-14 as well as board with both 0805, 1206, SOT-12, and SOIC-14).

In the following image, the same pattern is repeated and the idea is to have a large number of solder joints on a relatively small space in order to document potential temperature gradients and other undesired behaviour in the solder reflow process. Here, the lefthand-side clearly demonstrates a problem. I used lead-containing solder paste and the standard Sn63/Pb37 temperature profile in the firmware from Unified Engineering of Sweden AB.

Example of uneven heat distribution. Note un-molten solder paste on lower/upper lefthandside


Another test pattern gave better results with the same reflow profile.

Example of even temperature disctribution. All solder joints look nice and shiny


Adding enough solder paste on the pads can give the following characteristic solder joints.

Same curvature on all pads documenting enough solder paste has been applied


For small components, the nozzle/needle on the solder paste syringe will dispense quite a lot of solder paste. For SOT-23 components, the following can be the case prior to reflow.

SOT-23 components placed in solder paste


Here, lead-free solder paste is used, so the Amtech Syntech reflow profile is selected. The peak temperature is quite higher than in the previous profile.

Running the pre-loaded Amtech Syntech RoHS-compliant profile


After this solder reflow process, the SOT-23 components have aligned by themselves. This is an interesting feature of reflow processes. Note that it only works for smaller components (the forces doing this magic are relatively weak, but nevertheless they do surprise me from time to time).

SOT-23 components have aligned automatically during the reflow process


A test pattern with SOIC-14 components gave this result. Here, solder paste has been dispensed in a thin line across pads on each side of an IC. This results in solder bridges, but here only one solder bridge had to be removed, which is easy with a holder solder iron and some solder wick.

Solder bridge between pin 3 and 4 on the upper left SOIC-14 component


On the following test pattern, 0805 resistors have been placed in solder paste. Note the distance between pads is 0.5 mm, so solder bridges could be anticipated. Especially, if the solder paste has not been thoroughly mixed prior to application (segregation of metal particles and loss of volatile flux gives some interesting results sometimes).

0805 components placed in solder paste


A similar pattern with 1206 components prior to reflow.

1206 components placed in solder paste


Finally, a test pattern with SOIC-14 components. The solder paste stripes are quite clear.

SOIC-14 components placed in solder paste


The three above patterns side by side.

Three different component types placed in solder paste


The 0805 pattern after reflow. No solder bridges and components are neatly aligned.

0805 components after reflow process


In tilted view, the curvature on the solder joints is easier to see.

0805 components after reflow process, tilted view


Simlar observations can be made on the 1206 test pattern.

1206 components after reflow process


Also here, solder joints are easier to inspect in tilted view.

1206 components after reflow process,tilted view


The SOIC-14 components, however, have not moved much, and they should have been aligned better prior to reflow. Note the large number of solder bridges due to an excessive amount of solder paste applied.

SOIC-14 components after reflow process


Fitting four types of components with a pair of tweezers onto a single test board.


Preparing a batch comprising of 9 pcs. of the 0x0E design for reflow soldering.


Here is an example of a 1206 resistor that has moved during the reflow process. Manual correction with solder iron, solder wick, and fresh solder is easy.


Dispensing lead-free solder paste (Sn42/Bi58) onto a 0x05 circuit board.



SMD solder test pattern 100×50 (0805 1206, SOT-23, and SOIC-14): SMD_solder_ex_100x50_Gerber

SMD solder test pattern 100×50 mm (0805 1206, SOT-23, and SOIC-14): DipTrace_PCB_design_file_v3


SMD solder test pattern 100×100 mm (0402 0603, 0805, 1206, SOT-23, SOIC-14): Solder_ex_100x100_Gerber

SMD solder test pattern 100×100 mm (0402 0603, 0805, 1206, SOT-23, SOIC-14): DipTrace_PCB_design_file_v3



Comments are closed.