How to Choose a Reliable BLDC Motor Controller PCB Manufacturer for High Power Applications？

Later, I slowly figured out a truth: Many problems actually do not lie in the circuit design but in the “physical” aspects of PCB layout and routing.

Take that “land” for example. Many people think that just connecting all the lands together is enough! In fact, this is not the case at all.

You have to think about “power ground” and “signal ground” separately. “Power ground” is the channel for those large currents – such as the part of the circuit that supplies power to the MOSFET – and the current flowing through it is tens or hundreds of amps. What about “signal ground”? It is used for those sensitive sampling circuits – such as the op amp circuit behind the current sampling resistor – which require a very clean and stable reference point.

What would happen if you randomly connected these two “lands” together? That would be fun! Due to the large current switching action on the “power ground”, violent voltage fluctuations will occur – this fluctuation may only be a few tens of millivolts, but it is already a huge interference to the precision sampling circuit! This interference will be directly superimposed on your sampling signal, causing the value read by the ADC to jump around and be inaccurate.

The result? If your FOC algorithm controls the PWM output based on these erroneous readings, it will produce erroneous voltage commands, causing the motor to spin harshly or inefficiently.

So these two “places” must be isolated! How to isolate? Instead of completely disconnecting them, connect them at a specific point – usually using a magnetic bead or a 0 ohm resistor as a single-point ground near the negative pole of the bus capacitor. This ensures that the signal loop has a clean reference point while allowing the current to have a path.

Another problem that has troubled me for a long time is “ringing”. You may find that when the MOSFET is switching, its VDS waveform will be superimposed with some high-frequency oscillations that look like a string of small bells. This is “ringing”.

How did this phenomenon occur? To put it simply, the traces on the PCB are not ideal wires. They have parasitic inductance, and the MOSFET itself also has junction capacitance. These two things come together to form an LC oscillation circuit. When the switching action occurs, the oscillation of this circuit will be stimulated to produce high-frequency ringing.

This ring is no joke!
On the one hand, it will bring serious electromagnetic interference and your product will not pass the EMC test; on the other hand, it will also generate additional voltage stress on the MOSFET. If the ringing amplitude is too large, it may even exceed the withstand voltage value of the tube and directly damage the tube!

What to do? A common method is to connect an RC absorption circuit in parallel to both ends of the MOSFET. The function of this circuit is to add damping to the LC oscillation circuit so that it decays as quickly as possible instead of vibrating all the time.

However, the parameters of this RC absorption circuit are not randomly selected! You have to calculate it based on the actual ringing frequency, otherwise it may have no effect or may cause additional losses.

My own approach is to actually measure the ringing frequency on the VDS waveform, then add a known small capacitor and measure the new frequency. Based on these two frequency values, I can calculate the size of the parasitic inductance and capacitance, and then calculate the required resistance and capacitance values. Only in this way can the designed absorption circuit be more reliable.

When it comes to PCB manufacturing, I think many engineers now don’t understand the process deeply enough. They may think that they just need to send the Gerber file to the board factory and forget about the rest. In fact, there are many ways to do it!

Especially when you use some special boards, such as thick copper PCBs that need to carry large currents – what we often call Heavy copper PCBs – its processing technology is very different from ordinary boards!

The copper thickness of thick copper plates may reach more than 4 ounces or even 6 ounces. Such thick copper foil is prone to side etching problems during etching, resulting in inaccurate line width control; and when laminating multi-layer boards, uneven copper thickness is also prone to inter-layer alignment deviations. These problems will directly affect the reliability of the final product.

I have always thought that when it comes to making BLDC controllers, many people make the problem complicated. They always like to focus on those international standards and advanced theories, as if they can’t start work without getting a few certifications. In fact, if you actually make a controller that can rotate, you will find that many so-called “golden rules” are not that mysterious at all in actual operation.

Take PCB boards as an example. I have seen many engineers struggle with whether to use Heavy copper PCBs or not. It is true that high-current traces require sufficient copper thickness to carry them, but you also have to consider the actual heat dissipation structure and cost. Sometimes partially thickening the copper layer or inlaying copper on an ordinary board can produce better results and save a lot of money. Solutions that often use 4oz or even 6oz copper thickness are often over-designed. The board is heavy and expensive, and welding is particularly troublesome.

Speaking of soldering, I especially want to talk about the design of the pad. Many people are confused by various theories about this spacing and the window shape. My experience is that instead of memorizing standard parameters, it is better to spend more time understanding your production process and equipment capabilities. The accuracy of each factory’s placement machine is different, and the temperature curve of the reflow oven is also different. Blindly applying other people’s designs can easily lead to problems.
I once worked on a BLDC Motor Controller PCB project and suffered this loss – the pad spacing was designed according to the recommended parameters in an authoritative manual. However, during mass production, we found that our equipment could not meet such high precision requirements at all, resulting in a large number of false soldering.

There is another point that I think is very important: don’t be too superstitious about those complicated testing processes. Of course, I’m not saying that testing is unimportant! But you have to clearly distinguish which are truly necessary verification links and which are just formalistic flourishes. In order to appear professional, some companies come up with a lot of high-sounding test projects: a thousand times of thermal cycle, 100% coverage of X-Ray detection… But in fact, many tests have minimal improvement in the actual performance of the product, which is a pure waste of time and resources.

I prefer to do targeted testing in actual application scenarios. For example, if you are making a BLDC controller for power tools, simulate frequent starts and stops and high load impacts; if it is for home appliances, focus on testing the stability and noise control of long-term operation. This kind of testing method that is close to actual use has more reference value than those standardized laboratory data.

I would also like to say a few words about supply chain selection. Nowadays, many people choose suppliers based on who has more certificates and who is more famous. This is actually quite one-sided. Although some small factories I have worked with do not have ISO9001 or IATF16949 brands, the craftsmanship and experience of their masters are particularly solid and the products they produce are more reliable than the assembly line products of some large manufacturers.

Of course, I am not encouraging everyone to ignore qualifications at all, but I am saying that you should have your own judgment standards. Go to factories to see their production processes and equipment status, chat with their engineers, and feel their professionalism and technical understanding. These can reflect the true level better than a certificate.

After all, the most critical thing about making a BLDC controller is the ability to practice. You have to be willing to trial and error, adjust, and flexibly adapt to the actual situation. The theories and standards in books can only give you a general direction. The real path must be walked step by step.

I remember one time we were working on a new project. During the design stage, everyone felt that we should use the most reliable solution. During trial production, we found that the cost was too high and could not be brought to the market. Later, we readjusted our thinking and simplified the design in some non-critical parts. Instead, we made a product with higher cost performance and the market response was particularly good.

So my suggestion is don’t be intimidated by those complicated theories and don’t blindly pursue the so-called optimal solution. Start with actual needs and try more. The experience accumulated in practice is the most valuable. You can’t learn these things in any textbook. You can only understand them by doing it yourself.
That’s the thing about making hardware. It’s not like software, which can be patched at any time. Every modification requires real costs of time and money. But because of this, every success is particularly rewarding. When you see the BLDC controller you designed running smoothly, nothing can replace that sense of satisfaction. This is probably why I still like to do this business after so many years. Although it is hard work, it is worth it.

I have always felt that many people have a misunderstanding about motor controller design – they think that as long as the circuit diagram is correct, everything will be fine. In fact, the key to determining whether a BLDC controller can operate stably and how long it can be used is often hidden in the details that you cannot see.

Take PCB as an example. This is not something you can just find a board to deal with. I have seen too many projects where in the early stages of the project, in order to save money or catch up on schedule, they randomly found a factory to make prototypes, only to have things go wrong during high-temperature and high-load tests. Especially when you use those new wide bandgap devices, things become more subtle. These devices can run very fast, but it also means that your PCB must be able to keep up with them – any design oversight may greatly reduce the performance of the entire system. For example, an unreasonable via design or stacked structure may introduce excessive parasitic inductance, causing severe voltage spikes at the moment of high-speed switching, directly threatening the safety of power devices.

Speaking of which, we have to mention the importance of thick copper PCB. Many people may think that copper thickness only affects current carrying capacity, but it is actually much more than that. It is directly related to the efficiency of the heat dissipation path. If you think about it, heat will be generated when current flows. If this heat cannot be dissipated in time, no matter how good the chip is, it will overheat, reduce frequency, or even be damaged. My own experience is that for controllers that continue to work under high loads, it is better to spend more time planning the heat dissipation channels in the design stage than to wait until later to patch. For example, using 2 ounces or even thicker copper foil on the critical power path, combined with an array of inner thermal vias, can significantly reduce thermal resistance and control the chip junction temperature within a safe range.

Another point that is often overlooked is the layout of the controller. This sounds very simple, isn’t it just about arranging the components in order? But in actual operation, you will find that the placement of each component, the length and angle of the traces will affect the integrity of the signal. Especially if the impedance of the drive circuit is not well controlled or the parasitic parameters are too large, it can easily cause ringing or overshoot, which will put additional pressure on the switching device. For example, the driver IC should be as close as possible to the gate of the MOSFET and use short and thick traces to reduce loop inductance. This is crucial to suppress gate oscillation and ensure clean and fast switching action.

I experienced this deeply when I was working on a project recently. At first we designed the controller board according to conventional ideas, but during testing we found that the efficiency was always unsatisfactory and the temperature rise was much higher than expected. Later we readjusted the layout to more completely isolate the power section and control section and optimized the ground plane. The design effect was immediately different.
So sometimes the problem does not necessarily lie in the principle but in those seemingly inconspicuous physical implementation details. We use a star-shaped single-point grounding to separate the sensitive analog ground from the noisy power ground and merge at only one point, which greatly reduces the interference of ground wire noise on the sampling circuit.

Of course, finding a reliable PCB manufacturer is also crucial. Not only must they have the ability to make a board that meets your design requirements, but more importantly, they must understand the intention behind your design. For example, they need to know where special attention needs to be paid to impedance control, where the copper thickness needs to be strengthened, where the insulation needs special treatment, etc. A good manufacturer can help you better translate your design intentions into physical objects instead of just mechanically producing them according to drawings. They can provide professional advice on material selection (such as high-frequency plates), surface treatment technology (such as immersion gold is more beneficial to high-frequency signals) and processing tolerances to avoid compromise in design performance due to process limitations.

After all, designing a reliable BLDC controller is like cooking a delicate dish. You need good ingredients, which are high-quality components, but you also need a good pot and stove, which is a well-designed and well-made PCB. Both are indispensable and together determine the quality of the final product.

Nowadays, new technologies and new devices are emerging in the industry, which is both a challenge and an opportunity for those of us who design. The challenge is that we need to constantly learn to adapt to new requirements; the opportunity is that we can use these new tools to make more efficient, compact, and reliable products. The foundation of all this is inseparable from a deep understanding and careful polishing of the “cornerstone” of PCB. It may not directly bring cool functions, but it is a silent guarantee for the stable operation of the entire system. Every stackup planning, every wiring optimization, and even every via selection are silently laying the foundation for the long-term reliability and ultimate performance of the system.

I recently discovered a very interesting phenomenon. When many people mention BLDC motor controller design, they immediately start discussing various complex algorithms and software, as if the controller is generated by code. This is actually a bit putting the cart before the horse. After working on many projects myself, I feel that the hardware platform is the most basic and most prone to problems. No matter how beautifully written your code is or how fast it runs, if the circuit board itself that carries it is weak or unstable, everything will be in vain.

Take the core PCB of the controller as an example. Many people think that it is just a carrier that connects components. A conventional four-layer board or six-layer board is enough. I used to think so too. But later I found out that was not the case at all. Especially when the power you handle increases – such as driving an industrial fan or a small power tool – when the current is large, all problems will be exposed.

The biggest headache is the fever. Ordinary PCB copper thickness may only be 1 ounce or 2 ounces. If the current is slightly larger, the traces will become extremely hot, and the temperature rise of the entire board will be scary. Only then will you understand why special custom-made thick copper PCBs must be used in some situations.
It is not made to look good or to appear to be made of solid materials, it is purely determined by the laws of physics. A thicker copper layer means lower DC resistance and better current-carrying capacity, and heat can be conducted away faster instead of accumulating in a local hot spot.

But this brings new troubles. The processing technology of thick copper PCB is much more complicated than that of ordinary boards. Because the copper is thick, finer control is required during etching, otherwise the edges of the lines will easily become rough or even chipped. When laminating multi-layer boards, how to ensure bonding force and flatness between copper layers and dielectric layers of different thicknesses is also a technical task. Therefore, you must be particularly careful when choosing a manufacturer. Not all factories can perform this special process stably.

In addition to current carrying capacity, another easily overlooked but extremely critical point is the impact of layout and routing on signal integrity. Especially the design of the gate drive circuit. The gate of power switching tubes such as MOSFET or IGBT is very sensitive to the waveform of the driving signal. If your drive circuit is not designed well, the traces are too long or the loop area is too large and introduces too much parasitic inductance or capacitance… the consequences will be slower switching speed, increased loss or even severe oscillation causing the tube to blow up.

I have seen some designs that place the gate driver chip far away from the power tube to save trouble or reduce size, and then use long and thin traces to connect it. The result is a terrible switching waveform. A good design should be to place the gate driver as close to the gate pin of the power tube as possible and then the drive loop path should be short and thick to form a compact low-impedance loop. This ensures that the drive signal is clean and the power tube can be turned on and off quickly and reliably.

For BLDC motor controllers, this requirement is higher. Because it requires precise control of three half-bridges and six power tubes to switch in a specific sequence at the same time. Problems or interference with the switching timing of any bridge arm may lead to a reduction in the jitter efficiency of the motor or even damage. Therefore, the layout partitioning of the entire control board is very important. Usually, the high-power three-phase bridge arm part, the low-voltage digital control part and the sensitive gate drive part are clearly physically isolated, and the power supply and ground networks are well handled to avoid noise crosstalk.

This reminds me of my experience helping a friend adjust a small electric scooter. The controller boards they made themselves always burned the MOS tubes inexplicably. Later, I took the board over and took a closer look and found that they used a very ordinary double-layer board and squeezed all the power supply grounds on one layer to save costs. There is no separation or single point connection between the high-current power loop and the digital ground of the control chip. The result is that when a large current flows at the moment when the motor starts, a huge voltage fluctuation is generated on the ground plane, which directly interferes with the sensitive analog sampling circuit, causing sampling errors and triggering an erroneous PWM output.