SolderlessPCB: Reusing Electronic Components in PCB Prototyping through Detachable 3D Printed Housings

While all interviewees prototype circuit boards for various objectives, including research, robotic product development, and personal projects, they reported an inevitable circuit validation phase to gain confidence in their circuit diagram. According to the interviewees, breadboarding is the most commonly used technique to conduct circuit validation, as it allows for easy modifications when necessary. By the end of this phase, over 60% of the components are stored for future use. This includes larger components that are placed on the breadboard with DIP packages, such as breakout modules, transducers, and microphones, as well as through-hole components like transistors and LEDs. One participant stated, “The nature of my projects is more or less related, so a lot of components will be reused between projects for breadboards.” Additionally, apart from small components like resistors and capacitors, which are difficult to have their pins straightened after use, people agree that it is easy and efficient to recycle these components from the breadboard for future use.

After the circuit diagram has been verified, PCBs are made and assembled for long-term use. According to the interviewees, PCBs are either outsourced or made in-house using CNC machines, laser cutters, or chemical etching processes. All of them revealed that more than 80% of the components used in these PCB assemblies are SMD components. Two of the interviewees had experience ordering PCB assemblies with components pre-installed by the manufacturer. However, due to differences in cost-effectiveness and lead time between ordering pre-assembled boards and bare PCBs, they ultimately converged their decision with the rest of the interviewees. They chose to make or purchase bare PCB boards and components separately and conduct PCB assembly themselves using soldering irons, heat guns, or reflow ovens.

Despite their experience working with SMD components, none of the interviewees chose to desolder them for reuse between iterations or projects, except in the case of expensive or hard-to-find components. Interviewees highlighted several challenges in reusing SMD components. For example, one interviewee mentioned, “Many components have plastic housing or plastic parts in their packaging. When desoldering them using a heat gun, it is easy to mess them up, and the components end up not being usable anyway.” Besides the desoldering techniques, cleaning the leftover solder also presents challenges. “Sometimes, it’s hard to make sure all pins are properly cleaned after desoldering, so when using components like that, it is hard to solder them again,” said a circuit designer with eight years of PCB prototyping experience. Reportedly, less than 15% of components in PCB assemblies get desoldered and reused.

Snap-fit anchoring consists of four or more custom clips located at the four corners and along the edges of a housing, as shown in Figure 3. These clips are designed with groove heights that exactly match the thickness of an FR-4 baseboard. As a result, the housing can achieve a tight fit with a baseboard without requiring any additional manual process beyond simply pressing them together.

While snap-fit is the easiest assembly method, our experiments have shown that this anchoring mechanism does not provide uniform electrical connections across the entire area of a PCB. Because the downward force is primarily generated through clipping, components situated closer to the edges receive greater force than those nearer to the center of the baseboard. Although allocating more slots for snap-fit mechanisms would improve the overall pressure generated, it’s important to note that these slots take up a considerable amount of circuit board area to accommodate the compliant snap-fit tabs. Each slot occupies around 9  m² m of area. As a result, snap-fit housing will either severely interfere with PCB design and routing, or not provide reliable electrical connections across the board unless the prototyping board is sufficiently small. Due to this major reliability issue, we will not recommend this design overall.

Our second approach uses bolts and nuts with diameters as small as 1 mm to serve as the primary hardware for attaching the housing to the PCB baseboard, as shown in Figure 4. While bolting requires additional manual assembly effort, it allows us to distribute fixture hardware, and thus the downward force, evenly across the entire area of a PCB, regardless of their size. Since each bolt occupies only 0.8  m² m of area per hole on the board, its footprint does not significantly interfere with circuit trace routing. To ensure optimal pressure generation without damaging the housing, we apply a torque of 0.01 Nm to each of the nuts. Unless otherwise specified, all the examples in the paper are based on the screw bolting mechanism.

Two-terminal components are among the most commonly seen SMD components. They come in a series of numerically ordered package designs, with their dimensions in length and width measured in hundredths of an inch. Packages sized 0603, 0805, and 1206 are commonly used for PCB prototyping.

Due to their small size, these two-terminal components often have a degree of height variance. While this tolerance does not affect their performance, as they are designed to be soldered onto PCBs, it can pose a challenge for solderless assembly. Inconsistent height tolerances can affect the pressure applied during assembly. For example, if a component is slightly lower than its expected design height, it may not receive enough pressure to establish a reliable contact with the baseboard. Conversely, if the component is taller than expected, the housing may bend away from the PCB in that area. This could potentially allow neighboring components more vertical movement, resulting in loose connections.

We propose a cubic cavity design with a pair of flexible tabs to compensate for possible height variations caused by manufacturing tolerances (Figure 5). These flexible tabs are located on the side walls of the cavity, angled at 30° towards the PCB baseboard. Upon assembly, the tabs will deform inward, towards the cavity’s ceiling. Depending on the actual height of each two-terminal component, they will bend to varying degrees (Figure 5). This design will ensure adequate pressure between the components and the PCB, while at the same time absorbing energy through tab deformation to prevent the housing from bending away from the PCB.

In addition to two-terminal components, the majority of commonly used ICs for rapid prototyping are also available in standardized package designs. These include the Small Outline Integrated Circuit (SOIC), Thin Shrink Small Outline Package (TSSOP), Thin Quad Flat Package (TQFP), Ball Grid Array (BGA), and Quad Flat No-Lead Package (QFN). We categorize these package designs into two main types: Type 1, which includes SOIC, TSSOP, and TQFP, features pins extending from the plastic enclosure; and Type 2, which includes BGA and QFN, where pins are exposed only on the bottom of the chips. These two types of packages present different challenges for cavity design. We address them separately.

For Type 1 components, the extended metal pins are generally at the same height, although there may be some manufacturing tolerance errors. When a downward force is applied solely to the center of the IC enclosure, as with the cavity designed for two-terminal components, the pressure will not be evenly distributed across all the pins of the IC (Figure 6). Pins that are initially higher than others will not experience the same pressure as the lower ones and may not sufficiently connect to the baseboard. To account for potential manufacturing discrepancies, our cavity design for Type 1 ICs includes additional solid volumes at the top of each row of metal pins. These structures directly press on the metal pins, rather than applying pressure to the center of the IC enclosure (Figure 6 c). This approach ensures that all metal pins receive downward force, regardless of their individual height differences.

For Type 2 components, an exact negative volume is provided in the housing to accommodate the component, since the exposed metal connectors are not flexible and therefore do not require height compensation.

However, one challenge for Type 2 components is that the exposed metal connectors are not always the lowest point of the package design. Therefore, when pressing downwards, these metal connectors may not make contact with the baseboard. This can be mitigated by deliberately making the copper pads on the PCB baseboard taller than the surrounding areas, as illustrated in Figure 7.

For both Type 1 and Type 2 cavities, we employ the bolting technique introduced in Section 4.1.2 to generate localized downward pressure. A strong force applied to the corners of the ICs by the bolts may cause local deformation in the housing structure. This could affect neighboring small two-terminal components when multiple such components are required for a PCB design. This issue can be alleviated by carving a groove in the housing surrounding each component that requires bolting (Figure 8). The groove’s depth is set so that there is only a 0.3 mm thick of material connecting the component housing to the main housing. The 0.3 mm-thick connection will deform to absorb energy and prevent housing deformation from propagating to the neighboring components. The entire structure of the 3D housing remains as one piece during printing.

For SMD components that do not come in standard packaging, we individually design a customized cavity for each one. All these designs are tested for proper electrical connectivity and are stored in a component library.

In the IC component library, we have incorporated pre-allocated bolt holes into each IC footprint, and the same approach can also be extended to two-terminal components. However, it’s a common practice in PCB design to place two-terminal components closely together, as illustrated in Figure 10. In such scenarios, a set of predefined bolt holes for each two-terminal component may be unnecessary and could simply increase the manual assembly time. Instead, it is possible for a group of closely placed two-terminal components to share bolts (Figure 10 b). Here, we report on an experiment that determines the maximum allowable distance between neighboring bolts for a specific thickness of 3D-printed housing. The results of this experiment can serve as guidelines for the placement of bolts in two-terminal components that do not have pre-allocated holes.

As shown in Figure 11 a, we prepared a 23 mm by 91 mm PCB with a pair of bolt holes every 3 mm along the longitudinal axis. Between each pair of bolt holes, there is space for a 0805 two-terminal component. In total, the PCB can host 29 0805 SMD resistors. We also prepared nine 3D printed housings the size of the PCB, with the thickness ranging from 1 mm to 5 mm, at intervals of 0.5 mm.

In the experiment, we first placed all 29 0805 0-ohm resistors into one of the 3D printed housings. We then bolted the housing to the PCB, installing one pair of bolts at the very left end of the PCB, and another pair of bolts starting from the very right end and moving towards the first pair. While moving this second pair of bolts, we measured the conductivity of all resistors located between them and the left pair, until all resistors in between showed continuous connection. The distance between the two pairs is the maximum distance between bolts that can ensure reliable conductivity for the given thickness of the housing.

We repeated the same experiment for all nine housings, with each housing tested three times. As shown in the graph (Figure 11 c), the maximum distance between bolt holes monotonically increases with the thickness of the housing. This experiment enables us to determine the maximum distance between bolts for each housing thickness. For example, for a 3D printed housing with a thickness of 3 mm, as long as the distance between bolts is less than 27 mm, we can ensure that the electronic components in between have a reliable connection to the base PCB. All examples presented in this paper adhere to this design rule.

For two-terminal components, the flexible tab structure must be sufficiently pliable to absorb energy, thereby preventing the housing from bending. At the same time, it must exert enough pressure on the components to ensure good conductivity without causing breakage. We experimented with different geometric parameters and converged on the following empirical formula:

Where T represents tab thickness, α is the tab’s slope angle, W and L denotes width and length, and H indicates the height with respect to the housing surface. These parameters are determined by the specifications of a two-terminal component, where t, w, l represent the rated component thickness, width, and length, respectively.

We report that, using the aforementioned parameters and setup, we can reliably print housings for two-terminal components the size of or larger than the 0603 series. For the 0603 package, the tab has a length of 0.69 mm, a width of 0.48 mm, and a thickness of 0.45 mm.

To validate our housing cavity design for ICs, we developed a simple testing circuit incorporating ATtiny microcontrollers. We specifically selected three types of ICs: ATtiny85, ATtiny84A, and ATtiny828, chosen for their identical architecture, which enables them to run the same testing code. This selection enables us to test five different yet commonly used SMD packages: the ATtiny85 is available in BGA, SOIC, and TSSOP; the ATtiny84A in a QFN package, and the ATtiny828 in a TQFP package. Among the five different packages, the TSSOP has the minimal pitch between pins at 0.6 mm; the BGA has the smallest contacting pad area of 0.0625  m² m.

Our testing circuit comprises five LEDs, each connected to one of the five GPIO pins of the ATtiny ICs. Every LED features a 1206 footprint and is serially connected to a 100 Ω resistor in a 0805 package. Power is supplied to the circuit through two soldered through-hole pin headers. To test the housing cavity design, we uploaded code that controls the sequential flashing of each LED, as shown in Figure 13 and the accompanying video. We confirmed that all five packages establish reliable mechanical contact with the PCB baseboards. As a result, they effectively control the LED blinking, free from any glitches or ghost connections.

We measured 177 connection points created by SolderlessPCB across various packages of SMD components. For each measurement, we produced custom PCB boards and housings, enabling us to place two measurement probe leads: one at the IC’s metal lead, and the other at the contacting trace for ICs, or alternatively, at each side of the contacting traces for two-terminal components. Figure 14 illustrates our measuring setup. We measured the resistance using a Keysight 3446SECU digital multimeter and reported an average resistance of 0.46 Ω with a standard deviation of 0.139 Ω. The extra resistance should not affect the performance of common DC circuits when the electrical load is significantly larger than the connection point resistance. Note that the resistance data does not include connection points for any BGA package due to the lack of exposed conductors available for direct measurement, however, all prototypes made with UFBGA-15 package fulfill their functionalities.

Besides resistance, the quality of the electrical connection also affects signal transmission power, especially at higher frequencies  [14, 48].

To understand how SolderlessPCB preserves waveforms across different frequency ranges, we designed a test PCB equipped with two identical traces sharing the same input pin header. One trace includes a soldered 0 Ω resistor, while the other features a 0 Ω resistor assembled using the SolderlessPCB approach (Figure 15 a). We supplied both a sinusoidal wave and a quasi-square wave to the common pin header using a Function Generator BK PRECISION 4053, with frequencies ranging from 100 Hz to 10 MHz (the equipment’s maximum frequency). We then output the waveform from both output terminals and the input terminal as a benchmark, using an Oscilloscope SIGLENT SDS 1104X-E (Figure 15 b, c, and d). The results indicate no waveform attenuation through either the soldered or solderless circuit. For example, in the output graphs shown in Figure 15, both signals–the soldered trace (pink curve) and the solderless trace (blue curve)–exhibit the same amplitude level as the input signal (yellow curve), for both the sine wave and the zoomed-in view of the square wave.

To further assess the power loss of the high-frequency signals through mechanical connection points compared to soldered connections, we measured the energy loss between input and output signals using two 0 Ω resistors. Each resistor was assembled to close the circuit using either a soldered connection or the SolderlessPCB approach. To minimize energy loss on the trace itself, we redesigned the PCB with 65 mil traces (Figure 16 a). We also used individual input pins for each trace to prevent uneven splitting of signal energy at a shared input pin. For accurate power loss data, the test PCB was connected to a Vector Network Analyzer Keysight E5071C and fed with a pure sinusoidal wave signal within a range of 100 kHz to 5 GHz. The average signal energy loss across the full testing frequency was − 18.96 dB for SolderlessPCB and − 21.61 dB for the soldered PCB, with higher frequencies exhibiting greater energy loss in both cases (Figure 16 b).

The results of our experiment show that the SolderlessPCB is comparable to the soldered PCB even for high-frequency signal transmission.

Impedance is crucial in prototyping applications like radio, Bluetooth, and WiFi antennas, where impedance matching circuits are often developed through an iterative process [42]. In such scenarios, minimizing impedance variance at the connection interfaces between components and the PCB during assembly and disassembly is preferable. To assess impedance variance through multiple (dis)assembly iterations with SolderlessPCB, we designed and fabricated two identical PCBs carrying a pi-network circuit, commonly used for impedance matching [38]. We assembled the same set of components, both soldered and solderless, on these boards. Using a Vector Network Analyzer Keysight E5071C, we measured the impedance values for both PCBs across five (dis)assembly iterations. A pure sinusoidal wave signal ranging from 100 kHz to 5 GHz was supplied to the input end. It should be noted that frequencies below 100 kHz were not tested, as they are not relevant for impedance circuit applications [14, 48].

We report the results for both the soldered and solderless PCBs, along with their variances, in Figure 17. As shown in the figure, the impedance variance for solderless PCBs across different assembly iterations remains on par with that of the soldered ones. The results indicate that SolderlessPCB can be used for prototyping impedance matching circuits and other similar applications.

All prototypes described in Section 5.1 and Section 6, including those with both IC cavity and tab housings, remained fully functional three months post-assembly. This demonstrates that our SolderlessPCB methods can be reliably used for prototyping purposes.

As one potential application of SolderlessPCB is to facilitate the replacement of defective SMD components with new ones, we further investigated the housing reliability for component switching. Our focus was specifically on the tab structures of the two-terminal component housing, which are printed on a sub-millimeter scale and can be vulnerable to fatigue during the re-installation of a housing. To test the reliability of the printed tab structures, we repeatedly assembled and disassembled housings of three 0805 0 Ω resistors. For each iteration, we unbolted the housing and took out the resistors, and reassembled them from scratch. Using the same method as in Section  5.2.1, we measured the resistance values at the connection points after each assembly. We report the average resistance value across ten disassembly-assembly iterations in Figure 18. The resistance value remains consistent for the initial assembly cycles but begins to increase after the seventh iteration, indicating a failure of the tabs to exert adequate pressure beyond this point. We conclude that SolderlessPCB can be reliably used for component switching up to seven times before a new housing is required.

We conducted a drop test to investigate the robustness of SolderlessPCB. Specifically, we dropped the PCB assembly described in Section 5.1.3 from heights ranging from 0.3 m to 2 m, at 0.3 m intervals. We conducted three drops from each height; the PCB assembly remained fully functional after the entire series of drop tests. For an additional stress test, we dropped the same PCB assembly from a height of 6 m three more times. The PCB assembly continued to function properly. Our accompanying video documented one of these tests.

Alex is a maker and a designer. She recently moved to a new apartment and wanted to take up cooking as a new hobby. While the kitchen was well-equipped, it lacked one essential tool for Alex: a kitchen timer. Rather than using a smartphone, which could get greasy during cooking, or purchasing a brand-new kitchen timer, which might take time to be delivered, Alex decided to build her own timer using electronic parts she had saved previously.

Alex’s design was a PCB measuring 90 mm by 50 mm, featuring the following components: one ATtiny828 IC in TQFP-32 format, two 8-digit displays, two push buttons, and a battery holder (Figure 19 a). Adhering to the design workflow of SolderlessPCB, Alex designed the circuit and generated the corresponding 3D housing. She fabricated the circuit using a desktop CNC machine and printed the housing with an off-the-shelf resin printer. Figure 19 b showcases the completed assembly, which functions as a 60-minute timer.

Alex recently received two birthday gifts: a professional kitchen timer and a mini foosball table—the DIY timer she made earlier was no longer needed. However, noting all the electronic components in the DIY timer are perfectly functioning, she decided to reuse and repurpose them to create a digital scoreboard for the new foosball table.

Thanks to the SolderlessPCB approach, the kitchen timer could be easily disassembled using just a screwdriver. Alex then designed a new PCB for a scoreboard, utilizing all the components from the previous project. In the new design, the two 8-digit displays were positioned on either side, now serving to display the game scores.

Figure 19 d shows that the scoreboard installed on the foosball table. When a player scores a goal, pressing a button will update the display accordingly.

A bristlebot is a tiny robot built from a toothbrush head. With bristles on its bottom, it can move in arbitrary patterns or follow a predefined trajectory when vibrating. Alex saw it online and would like to build one as a toy for her newborn nephew.

Alex tested the circuit with a breadboard first. However, as a breadboard was too big for a toothbrush head, the finished toy had to have a compact design using a custom PCB.

Like before, Alex employed the SolderlessPCB approach and manufactured her own PCB baseboard and 3D-printed housing. Figure 20 a shows the first iteration. It featured an ATtiny84 as the microcontroller and utilized two top-open JST connectors, one reserved for a vibration motor and the other for a Li-Po battery.

Alex used a bench power supply to power the circuit for testing. Unfortunately, the two terminals of the supply were incorrectly connected to the board, resulting in the small IC on the circuit getting burned. Figure 20 b shows the fried IC.

In most cases, a burned multi-pin IC can pose a problem since it is challenging to desolder from a PCB and to clean all the pins properly. As a result, the fried PCB and all its onboard electronic components are often disposed of directly.

In Alex’s case, however, replacing the IC was much simpler. After disassembling the housing from the PCB baseboard, Alex popped out the broken IC. Noticing that the circuit traces were intact, she directly inserted a new IC into the design. This time, the circuit worked properly, and Alex had a functioning bristlebot (Figure 20 c).

Since the bristlebot lacked the capability to respond to its environment and was not controllable, Alex embarked on a third design iteration. In this iteration, she decided to include an ambient light sensor to sense environmental light as a means to control the bot’s motion. Additional features like an LED to indicate the bot’s status and a buzzer to make sounds were also added (Figure 20 d). Because these extra components could add weight, Alex also upgraded the bot from one brush head to two, and with two vibration motors.

Alex still needed to prepare a new PCB baseboard, but because she used SolderlessPCB approach, she was able to recycle all the SMD components from the second iteration. Alex also modified the PCB housing to include a few additional features, allowing it to clamp directly onto the two brush heads. The housing now serves two functions: it secures the SMD components to the PCB, and it also forms part of the bot’s structure. Figure 20 e shows the final design iteration.

SolderlessPCB can be used in circuits that deliver higher current beyond those used in typical logic signaling. Here, we demonstrate a tabletop mug heater designed to maintain liquid temperature. The mug heater composes a 3D printed mug base plate, an aluminum topping, and a PCB made using SolderlessPCB approach. As illustrated in Figures 21 a, b, and d, the main component of the PCB is a 12 V, 60 W heating element. This element is undervolted and powered by a 7.4 V, 2500 mAh Li-Po battery with a 35C rating, and is controlled through a relay.

The logic circuit is made with an ATtiny84A microcontroller in an SOIC package and powered by a 2032 coin cell battery. It reads the temperature of the aluminum and controls the relay accordingly. The PCB housing design is directly integrated into the cylindrical frame of the mug heater and was manufactured as a single piece, as shown in Figure 21 c. Upon turning on the device, the aluminum surface of the mug heater quickly reaches \(40 \,\mathrm{ \mathrm{^{\circ }\mathrm{\mathrm{C}}}}\) (as shown in Figure 21 e and f) within 1 minute and then stabilizes at around \(45 \,\mathrm{ \mathrm{^{\circ }\mathrm{\mathrm{C}}}}\). The current draw during heating is around 3.3 A.

In this example, we utilized the SolderlessPCB approach to create a mini game console, where the display data was transmitted through the I²C protocol. The PCB comprises an ATtiny84A microcontroller, four non-lock push buttons, and a 4-pin SMD JST connector that interfaces with an OLED display (Figure 22 a). The housing of the SolderlessPCB is assembled as part of the game console, with the top housing holding the OLED display screen while the bottom houses the circuits and the buttons (Figure 22 b and c). We implemented a snake game operated on the OLED display through I²C, transmitting data at a rate of 100 kHz with no data loss. The example demonstrates that SolderlessPCB can be used for prototyping circuits that require fast-speed data transmission.

In this example, we show that the SolderlessPCB approach can reliably support ICs transmitting data at even higher speeds. Future Technology Devices International (FTDI) modules, which convert USB to serial communication, are commonly used for code uploading in miniature Arduino boards lacking on-board UART (universal asynchronous receiver/transmitter) support (Figure 23 a). Here, we built a custom miniature FTDI adapter using SolderlessPCB, with a 28-pin FTDI232RL IC in an SSOP package and an SMD micro-USB connector. The 3D printed housing secures both the IC and the USB connector with six bolts (Figure 23 b). The custom FTDI module, made with the SolderlessPCB approach, tolerates repeated plugging and unplugging of the USB cable, and can communicate with the USB at the default rate of 12 Mbps and reliably upload code to an Arduino Pro mini (Figure 23 c).