Penny Wise and Pound Foolish: One Man’s Experience with the Hidden Cost of Cheap Parts
Sometimes pennies saved in the design and manufacturing process can cost dollars later. This article visits the hidden costs of sourcing cheap components.
Fast international shipping has made it possible to source components from many places in the world for minimal cost. Unfortunately, those parts can have poor tolerances and low manufacturing standards. While it might be possible to save a small amount of money for an individual component, the time invested in troubleshoot when things go wrong can quickly eclipse any initial savings.
The Allure of a Bargain
Everyone loves a deal. It is irrational to pay one vendor more for an item when you can find an identical replacement for less elsewhere. However, when sourcing parts, you have to determine what you stand to lose and what you hope to gain by pinching pennies. The following article was inspired by a bad experience with shunt jumpers purchased from an overseas vendor through an online marketplace. While it focuses on a single bad part from a single supplier, the implication is a larger one. Pennies saved by purchasing cheap parts early in the design process can have hidden costs later in the development cycle in the form of wasted time, failed circuits, and failed products.
The Discovery of a Problem
This problem arose during the testing of a new ultra-low-power inductive comparator circuit. The board has four two-pin headers with shunt jumpers that can be used to electrically connect the two pins. Two headers enable indicator LEDs, one header enables an optional voltage divider, and one header toggles between a timer-based mode and an always-on mode.
Figure 1. 3D render of the circuit board in question. Four two-pin headers were used as switches.
The outputs of the two ICs on the circuit board were designed to activate LEDs that could be disabled to conserve energy. Unfortunately, the LEDs were not working.
Figure 2. Transistor switch subcircuit in question.
My first thought after powering the circuit on and not seeing the LEDs flash was that the LEDs were too dimly lit to be seen, so I moved the circuit to a dark room. Next, I used a multimeter to probe along the path that traces from the 3.3V supply through the LED (D2), the current-limiting resistor (R7), and the transistor (Q3) to ground. Moving on from there, I did some hot-air rework; I decreased the value R7, and then changed R8, and then replaced Q3, thinking it perhaps faulty. Still having no luck, I replaced D2 and changed its orientation on the board, in case polarity reversal was the problem. Absolutely nothing worked on this or the second identical subcircuit.
I had to trace the circuit from net to net to discover that the signal disappeared on the circuit board somewhere between the alarm output of the IC and R8. Only by chance did I happen to slide my oscilloscope probe across the top of the shunt jumper from one header pin to the next and notice the change in voltage—something that shouldn't be possible: if two conductors are electrically connected, they must have the same potential.
I resoldered the header pins to the board to eliminate the chance of a cold solder joint and grabbed another shunt jumper from the bag. Still no success. To the naked eye, and even under light magnification, it is impossible to tell that the jumpers are malformed. Only when they are installed and inspected at medium to high magnification, and at just the right angle, is the problem apparent.
The Problem of Bad Shunt Jumpers
Shunt jumpers are designed to short-circuit two header pins on a circuit board, thereby joining two disconnected nets. They are commonly used as a cheap alternative to SPST and SPDT switches on a circuit board. While less convenient than a switch, shunt headers provide a seemingly bulletproof way to electrically connect two adjacent pins.
Figure 3. Image of shunt jumper installed on two header pins, from SparkFun Electronics.
Shunt jumpers are inexpensive items: fifty shunt jumpers can be purchased from Digi-Key for less than two dollars, while fifty slide switches would cost closer to forty dollars. For a large production run, the cost difference between a shunt and a switch will quickly run into the thousands of dollars.
Figure 4. Shunt jumper price breakdown, from Digi-Key.
The potential problem is that cheap shunt jumpers have the same general appearance as quality shunt jumpers but do not work reliably, or in my case, at all. Microscopic inspection reveals a defined separation between one header pin and the conductor that is supposed to short-circuit the two encased header pins.
Figure 5. The image above shows a non-functioning shunt jumper on the left and a functioning shunt jumper on the right. Photo credit: Mark Hughes, AAC.
Figure 6. A close-up of the non-functioning shunt jumper shows a slight separation between the header pin and the metal shunt.
Implications and Avoiding Pitfalls
My use-case is a relatively simple one, and I had time for troubleshooting. But shunt jumpers are used widely throughout the industry to configure circuit boards. If I had used these same shunt jumpers to configure circuit boards in a production device, it is very likely that large numbers of products would not function as intended. A high failure rate would result and cause a great deal of angst in the quality assurance department. A great deal of labor could be invested in troubleshooting and determining the cause of the error.
What's worse, in the process of the hot-air rework and probing, I irreparably damaged the electrical connections between the main IC and the circuit board—it is a small 2mm × 2mm 8-WSON package that I inadvertently nudged, and it is beyond my ability to repair at home.
In all, I likely invested three or more hours of rework and troubleshooting, and I ordered parts for replacing the transistors, LEDs, and resistors. I have now spent close to $60, invested another $140 in time, all to have a non-functioning circuit. I will likely have to order a new circuit board from MacroFab as well as new jumpers. All because I wanted to save a half-penny on a half-nickel part.
Electronics design is complicated. Do not further complicate your life by purchasing low-quality parts from unknown vendors.
Wouldn’t a simple continuity test along the circuit trace prove no current thru the transistor base?
Such a simple circuit and you could not find a problem…
Next time add a pull down resistor to your transistor base. That’s proper engineering.