Although semiconductor integration is becoming increasingly high, many applications have readily available system-on-chips, and many powerful and out-of-the-box development boards are becoming more easily accessible, electronic product applications still require custom PCBs in many use cases. In one-time development, even an ordinary PCB can play a very important role.
Although semiconductor integration is becoming increasingly high, many applications have readily available system-on-chips, and many powerful and out-of-the-box development boards are becoming more easily accessible, electronic product applications still require custom PCBs in many use cases. In one-time development, even an ordinary PCB can play a very important role.
PCB is the physical platform for design and the most flexible component for electronic system design using original components. This article will introduce several golden rules of PCB design. Most of these rules have not changed since the birth of commercial PCB design 25 years ago and are widely applicable to various PCB design projects. They are of great guiding significance for both young electronic design engineers and more experienced circuit board manufacturers.
Ten Most Effective Design Rules
The following are the ten most effective design rules that electronic design engineers should keep in mind and practice when using design software for PCB layout design and commercial manufacturing. Engineers do not need to implement these rules in chronological order or relative importance; just following all of them can greatly change product design.
Rule 1: Choose the Right Grid. Set and always use a grid spacing that matches the most components. Although multiple grids may seem effective, if engineers think a little more in the early stages of PCB layout design, they can avoid difficulties in spacing settings and maximize the use of the circuit board. Since many devices use multiple package sizes, engineers should use the product that is most beneficial to their own design. In addition, polygons are crucial for circuit board copper pouring. Multi-grid circuit boards generally produce polygon filling deviations when performing polygon copper pouring. Although not as standard as based on a single grid, they can provide more than the required circuit board service life.
Rule 2: Keep Paths Shortest and Most Direct. This sounds simple and common, but it should be kept in mind at every stage, even if it means modifying the circuit board layout to optimize routing length. This is especially true for analog and high-speed digital circuits where system performance is always partially limited by impedance and parasitic effects.
Rule 3: Use Power Planes to Manage Power and Ground Distribution as Much as Possible. Power plane copper pouring is a faster and simpler option for most PCB design software. By sharing connections for a large number of wires, it can ensure the provision of current with the highest efficiency and minimum impedance or voltage drop, while providing an adequate ground return path. If possible, multiple power supply lines can also be run in the same area of the circuit board to confirm whether the ground layer covers most of a certain layer of the PCB, which is conducive to the interaction between lines running on adjacent layers.
Rule 4: Group Related Components Together with Required Test Points. For example, place the discrete components required by an OpAmp operational amplifier close to the device so that the bypass capacitors and resistors can work in conjunction with it on the same ground, thereby helping to optimize the routing length mentioned in Rule 2, and also making testing and fault detection easier.
Rule 5: Repeat the Required Circuit Board Multiple Times on a Larger Circuit Board for PCB Panelization. Choosing the size that best suits the equipment used by the manufacturer helps reduce prototyping and manufacturing costs. First, perform circuit board layout on the panel, contact the circuit board manufacturer to obtain their preferred size specifications for each panel, then modify your design specifications and try to repeat your design multiple times within these panel sizes.
Rule 6: Integrate Component Values. As a designer, you will choose discrete components with higher or lower values but the same performance. By integrating within a smaller standard value range, you can simplify the bill of materials and potentially reduce costs. If you have a series of PCB products based on preferred device values, it will also be more conducive to making correct inventory management decisions in the long run.
Rule 7: Perform Design Rule Checks (DRC) as Often as Possible. Although running the DRC function on PCB software only takes a short time, in a more complex design environment, as long as you always perform checks during the design process, you can save a lot of time. This is a good habit to maintain. Every routing decision is critical, and by performing DRC, you can be prompted for the most important routings at any time.
Rule 8: Use Silkscreen Flexibly. Silkscreen can be used to mark various useful information for future use by circuit board manufacturers, service or test engineers, installers, or equipment commissioning personnel. Not only mark clear function and test point labels, but also mark the direction of components and connectors as much as possible, even if these comments are printed on the lower surface of the components used on the circuit board (after circuit board assembly). Fully applying silkscreen technology on the upper and lower surfaces of the circuit board can reduce duplication of work and streamline the production process.
Rule 9: Mandatory Decoupling Capacitors. Do not try to optimize your design by avoiding decoupling power lines and relying on the limit values in the component data sheet. Capacitors are inexpensive and robust. You can spend as much time as possible assembling capacitors, while following Rule 6 and using standard value ranges to keep inventory organized.
Rule 10: Generate PCB Manufacturing Parameters and Verify Before Submitting for Production. Although most circuit board manufacturers are happy to download and verify for you directly, it is better to output Gerber files first and check with a free viewer to see if they are as expected to avoid misunderstandings. By verifying in person, you may even find some careless mistakes and thus avoid losses caused by production according to wrong parameters.
As circuit design sharing becomes more widespread and internal teams increasingly rely on reference designs, basic rules like the above will remain a feature of printed circuit board design. We believe this is very important for PCB design. With these basic rules clarified, developers can flexibly enhance the value of their products and get the maximum benefit from the circuit boards they manufacture. Even novice circuit board designers can accelerate the learning process and increase confidence by keeping these basic rules in mind.
Engineers' Perspectives on Hardware Design
The following is a share from an engineer, which is worth reading:
Due to my weak technical foundation, I applied to work as an engineer on the production line. During this period, I looked at design from a production perspective and found that different perspectives lead to different content. Every time on the production line, employees would glare at me and complain: "Who the hell designed this product~" I would just walk past with a black face. I admire the members of our maintenance team very much. Comparatively speaking, my maintenance skills are broad but superficial, while theirs are down-to-earth. The reason is that I often start from theory, first look at the schematic diagram when repairing something, and then analyze which part of the circuit the problem may be in. They are different; they will take a good board and measure it, compare it, find where the problem is, and then figure out the reason. Of course, I think the main reason for the difference is that I don't have the pressure of repair quantity, while their performance is directly linked to results.
During this period, many problems occurred and many boards were repaired. However, the highest proportion of causes was still design problems, accounting for more than 90%. Especially in customer complaints about failures, they can all be attributed to design defects. Although there have also been problems such as PCB incoming material short circuits and equipment process issues, the number is extremely small. Design problems have many aspects, some are principle problems, some are layout problems, some are process problems, and some are material selection problems.
The most serious design defect I encountered was that there was no main switch designed in the circuit, that is, it could not be powered off. There was only a signal switch during normal use. There was no problem during normal operation, but it would burn directly after a failure. Therefore, no matter what product it is, if you want safety, you must have a switch that can quickly disconnect the circuit. I explained this problem in my previous blog questioning electric vehicles: in the event of an accident, can we ensure fast and effective power off? Several electric vehicle accidents, such as BYD, Zotye, Tesla, etc., ultimately resulted in burning because the power could not be disconnected, and the short circuit state would continue to heat up and catch fire until it burned out.
And the problem of microcontroller ports is a relatively common phenomenon. There was a product that had intermittent failure to start during production. After much puzzlement, the R&D engineer finally gave the answer—the light wire was broken! It turned out that the pin was multiplexed in the design. When powering on, it would detect whether the pin (open drain) was high level to enter the calibration state, and when entering the working state, it would be multiplexed as a driving pin (push-pull output). When the light wire was broken, the state was uncertain. If it was mistakenly considered high level, it would enter the calibration state and never work, looking like a crash. Therefore, in design, try not to assign too many possibilities to pins with special signals. This fault might have been improved if there was a pull-up resistor.
Similarly, our tooling also had a similar situation. The bouncing of buttons has always been a problem for us. For unknown reasons, the program mistakenly thought that the button was pressed and executed the wrong function. We tried various solutions such as routing layout, software anti-shake, and button filtering, but none of them completely solved the problem. Accidentally, during a chat, we found that the engineer did not add a pull-up resistor in the design~ All our other colleagues defaulted that pull-up resistors would be added when designing buttons. Empiricism kills people.
Most layout problems are related to ground wires, and problems are prone to occur in ground separation and single-point grounding. There was a product that burned one of the thin wires during the overcurrent test. After analysis, it was found that this was a ground wire, which was closer to the power supply than the large-current copper pouring. During the overcurrent test, the current would first choose this short path and burn the line. In addition, there are many cases where signal lines are routed near moving structural parts or card slots, leading to abrasion. These all test an engineer's familiarity with the product.
In another low-voltage product, a surge suppression diode was connected in reverse parallel between the positive and negative electrodes. The customer complained that the battery was drained when the main switch was not turned on. I picked up the circuit diagram and thought the problem must be with this diode. Then I checked the specification sheet and found that the device had a rated reverse leakage current of 500mA, which means that if a diode has 499mA, it is normal according to the specification sheet, but the battery will discharge all the time until it is completely dead. This is a typical poor material selection.
And there was an old product that was complained about easily damaging multimeters during production. After verification, it was found that one of the test contents was to test mA-level current, about 80mA, but if there was a PCB failure, it would soar to more than 1A. During testing, the process document required the mA range, which naturally burned the multimeter fuse. After seeing this, I just switched the multimeter to the A range, because the multimeter can effectively measure to 0.001A, which has the same effect as the mA range, but a current of 1A will not burn the fuse of the A range. It can be seen how important appropriate material selection is.
I generally do not use the current range because it is easy to damage. For debugging occasions, a clamp meter may be more convenient, but in production testing, I will recommend using a current sensing resistor. And the selection of the current sensing resistor is also very critical. There was a tooling that initially used a 220mR 3W current sensing resistor. This resistance value has a relatively high test voltage drop for small signals and is easy to measure, but it is easy to burn out in overcurrent conditions. Calculations found that when the overcurrent threshold is 30A, the instantaneous power on the resistor is nearly 200W, which will definitely burn out! So I suggested using a non-inductive resistor of 10mR 5W, which can withstand a current of 20A during normal operation, and has a ceramic package, eliminating the risk of fire.
Another product also had many problems, which were related to design. Since terminal design was not adopted in the design, all wires were soldered directly to the PCB for turnover during production testing. From the early stage of production, it was found that some units could not start, and the resistance value of the speed control knob dropped significantly. At first, we suspected a knob failure, but after multiple disassembly, we found that the internal circuit was burned, and the knob itself was not damaged. We never understood why the internal circuit would be damaged—short circuit? Soldering? Chip failure? Impact? After analysis, none of these were the case, and it could not be explained in principle. Finally, the reason was found: because the wires were dragged, the NTC wire touched the high-voltage line during testing, introducing high voltage into the low-voltage circuit and causing the chip to burn instantly. If terminal connection was adopted in the design, not only would testing be convenient, but there would also be no such a high damage rate. The damaged PCBs were enough to cover the purchase cost of the terminals, which was not worth the loss.
This product also had a strange test phenomenon—sparking. If the tester connected the wires with a bracelet, there would be no spark when connecting the live wire, but there would be a spark when connecting the neutral wire. In theory, the live wire is disconnected by the switch. When the test tooling was inspected by the tooling group, this phenomenon did not occur. I directly touched each wire and there was no problem. After much puzzlement, accidentally, when testing the voltage of each point to ground, I found that the neutral wire was actually high voltage! At this point, it was discovered that the input of the tooling's live and neutral wires was reversed, and the socket of the tooling group was also reversed. It seems that the appearance is a wiring problem, but it also indicates that there are risks in the equipment circuit control and the design is incomplete. Therefore, later AC test toolings are equipped with AC contactors or relays to disconnect both live and neutral wires, and are controlled by non-locking buttons to avoid the problem of PCB being charged during operation.
This product also selected a series of unsealed devices for applications requiring epoxy resin sealing, which is a design defect that has to be mentioned. The selected devices did not meet the usage requirements.
In fact, after disassembling many different types of domestic products, you will find that this is a common feature of domestic design: giving up some performance to ensure low cost.
I always remember what the teacher said during training: 99% of the cost of a product occurs in production, but the magnitude depends on the design.
Design should reduce the total cost of BOM + process as much as possible in production, not just look at the BOM cost. Using cheap raw materials does not necessarily mean that the product is cheap. Many large Chinese companies collapsed overnight because cheap raw materials tricked themselves. The teacher also said that now customers no longer blindly pursue low prices, but also consider the comprehensive cost of later maintenance. For design, production is our customer. No matter how advanced the design is, if it is difficult to manufacture, the workers will curse.
This year, this product broke out another hidden defect: it inexplicably soared to the limit speed and got out of control. When the PCB was removed for testing, everything was normal. After analysis, the problem was concentrated on the speed control knob, but the knob is only a weak current signal. Adjusting it to the bottom is the lowest speed, and to the top is the highest speed. There is no way to have a phenomenon of out of control to the limit speed. During the test, an adjustable resistor was specially used to simulate it, until an accidental opportunity to disconnect the middle pin~~ the fault reappeared. So the reason for the problem is that the structural imbalance causes the knob to be stressed and the middle pin to be suspended. But this is not the root cause. The root cause is that the program lacks a sentence to limit the maximum speed during PID speed control. Under normal conditions, it will stabilize at the maximum speed, but after a failure, it will run away.
In fact, there are many more defective phenomena. All defective products will eventually be sent for repair, so repair has become a place that greatly exercises an engineer's ability to analyze problems, and at the same time can discover many design defects of predecessors, so that you can take fewer detours on your own design road in the future. So students, if you want to become an excellent engineer, you can go to repair for exercise first.
About Maxipcb
Maxipcb empowers innovators to turn cutting-edge technologies into reality.
We offer one-stop solutions for design, simulation, testing, PCB manufacturing, component procurement and SMT assembly, enabling efficient development, rapid deployment and risk control across the full product lifecycle.Serving the world in communications, industrial automation, aerospace, automotive, semiconductor and beyond, we build a safer, more connected future together.