
Injection Molding Machine PCB Heavy Copper Guide: Etching Compensation, Resin Fill, and Thermal Via Quality for Reliable Controllers
Why heavy copper injection molding machine PCBs fail in the field: side-etch
A Vehicle Barrier Gate PCB mostly fails slowly, under moisture, temperature swings, vibration and motor current, not through a sudden component burn-out. Typical root causes are thin via copper, inner-layer cracks, weak power-loop copper, shared return paths between sensors and drivers, cold-start voltage collapse and coating or solder stress. Stable gates come from clean power and signal separation, thick inner copper where current flows, a hardware lowering-enable path, cold-start load testing and suppliers who control lamination and process lock between prototype and volume.
I have repaired quite a few parking gates and noticed a pattern over time: on the surface it looks like a relay that will not pull in or contacts that stick, but the problem of the board itself is often overlooked. Once a gate failed to lift the arm once or twice every week; the on-site technician changed the relay, it worked for a few days, and then the fault came back. In the end we took the whole Vehicle Barrier Gate PCB off and looked at it against the light, and found a very thin crack in the inner layer, right on the relay drive loop. This board had originally been made by a cheap multilayer pcb supplier whose lamination temperature was not tightly controlled, and it cracked after a few thermal cycles. Later I simply handed the faulty board to an hdi pcb manufacturer for re-proofing; the layout was not changed much, but the several high-current drive channels were moved to inner-layer thick copper and the blind vias were changed to laser drilling, and it has been running almost two winters without problems.
So I am not very willing to suspect the power transistor or the relay itself right away. What an outdoor barrier gate fears most is not one device suddenly failing but its electrical performance slowly degrading under moisture, temperature differences and vibration. If the hole wall copper of an ordinary multilayer board is too thin, when large current passes repeatedly through the relay coil, the hole wall heats first, and over time the copper layer contracts, producing intermittent poor contact. This kind of fault cannot be measured with a multimeter; it can only be spotted at night when it is quiet, when the relay pull-in sounds muffled, or with a thermal imager looking for local temperature rise. When choosing boards I would rather spend a little more to find a reliable multilayer pcb manufacturer and have the solder mask and hole copper made thicker, instead of saving that bit of process cost under the relay socket.
The only trouble is that there are quite a few factories that can make industrial-grade HDI boards, but few hdi pcb supplier plants willing to take small-batch repair boards, either blocked by minimum order quantities or charging outrageously high engineering fees. Sometimes I have to panelize two or three different models of gate control boards into one panel to spread the cost a little. The relay is only the link that exposes the problem; what you really have to watch is whether the PCB factory’s process and your drive design can withstand the outdoor beating.

Many people making barrier gates still think a control board is just a board that can be powered, and that if it works it is fine; this idea really causes delays. In the variable-frequency gates I have handled, the probability of the board having problems is far higher than that of the motor itself. Once the frequency converter module’s switching frequency is pulled up, if the power ground and signal ground are not handled well, those few encoder wires are like antennas and every kind of noise pours in. I met one machine where the position given by the Hall limit switch and the encoder did not match, and the control board kept reporting faults; the investigation finally showed that the encoder was not broken but the PCB layout had made the encoder supply ripple outrageous.
I basically no longer use ordinary double-sided boards for frequency converter gate boards. For any new model or revision, the Vehicle Barrier Gate PCB either goes to a multilayer pcb manufacturer as a four-layer board with power and signal layers separated, or, for especially compact boards, goes directly to a factory that can be an hdi pcb manufacturer for prototypes, using laser blind vias for some key signals. I paid quite some tuition on this road. Many multilayer pcb supplier plants can actually do it in process, but if you do not ask proactively, they will not mention details like inner-layer copper thickness and prepreg type, and once you state a requirement the price goes up. Later I learned my lesson: at the prototype stage I require the PCB factory to provide the lamination stackup and impedance report, which costs a bit more but saves worry in volume.
On torque as a parameter, I was also superstitious at first, thinking that the automatic torque balance advertised for variable-frequency gates was especially advanced. Only after actually tuning several units did I understand that if the control board cannot accurately obtain current, speed and position information, torque control is an empty phrase. Especially on boards whose encoder resolution is insufficient or sampling delay is large, when the arm is blown by the wind and the motor current changes, it simply has no time to react. So now when I look at a board, I first look at whether its current sampling position and encoder interface are clean, rather than at how many newton-meters it claims. Whether the hdi pcb supplier can guarantee equal length and impedance of differential signal pairs directly affects whether your on-site commissioning gets through in one go.
On the anti-smash side, many people like to pile on radar, infrared and ground loops, as if more is safer. In fact, however many sensors you install, if the control board input ports do not have enough isolation and filtering, the interference goes in when the relay pulls in and false actions still occur. When I configure a system for someone now, I generally require the control board to isolate sensor signals from the power drive part, at least with optocouplers or magnetic isolation in place; otherwise in summer thunderstorms the ground loop and the infrared beam will report faults at the same time, and you cannot even trace the wiring.
Many people think a gate control board has little technical content, that one forward-reverse relay and a limit switch are enough to make it turn. Do it that way and the equipment really can run, but within three months the phone calls from the site can annoy you to death. I have repaired quite a few parking exit gates, and when I opened the control box and saw the old approach of a single-sided board with plug-in relays, at motor reversal the relay contacts spark and burn the surrounding copper foil black.
I used to manage the gate retrofits of a few residential parking lots. The first batch all used brushed motors, and the carbon brushes were nearly worn out after about a year, with carbon dust falling everywhere and connectors and Hall signals on the control board often in poor contact. Later we converted to brushless motors, thinking it would save worry, and found that the trouble had moved from the motor to the board. The commutation of a brushless motor relies on the three-phase inverter bridge on the board; once those few MOSFETs work, the current is not small, and as the switching frequency rises, the heating and interference on the board all come out.
This brings us to the design of the Vehicle Barrier Gate PCB. This board looks uncomplicated, but in reality the pre-driver, Hall interface, three-phase bridge and main controller must all be squeezed in, with enough creepage distance and heat-spreading copper left as well. I have seen some boards that put the pre-driver chip far from the MOSFETs, with thin, long gate traces, and once the drive waveform gets worse the MOSFETs heat badly and the gate keeps striking in summer. So when I later had people draw boards, the first thing was to separate the power loop and the signal loop, keep the ground plane as complete as possible, and place the pre-driver right against the transistors.
I have stepped into pits when choosing PCB factories too. At first, to be cheap, I found a small multilayer pcb manufacturer, which overstated copper thickness and could not get heat conduction up, and the power area of the board changed color after a few months of use. Later I switched to a multilayer pcb supplier that makes more automotive control boards, and things stabilized. As for HDI, I personally feel a gate control board really does not need any-layer interconnect from the start. Some hdi pcb manufacturer plants will tell you the density is higher and the routing smoother, but space inside a gate housing is not that tight, and an ordinary four- or six-layer multilayer board is perfectly enough, with cost down by quite a bit. Of course, if you are making an ultra-thin integrated gate with the board area pressed very small, then finding a reliable hdi pcb supplier can indeed solve the problem, but that is a need forced by space, not a hard indicator of process upgrade.
On frequency conversion I think the industry has some over-promotion. Many people hear the word frequency conversion and feel it is very advanced, but a gate is not an elevator: raising and lowering times are basically fixed, and the significance of smooth speed change is not as large as imagined. I tuned several models with frequency converter modules, and the fast and slow gears were set again and again, and in the end users simply could not feel it, while an extra heat source and failure point were added. Commutation logic is indeed the key to whether a brushless motor can run stably, but many main controller chips now integrate that part; what is really easily overlooked is the pre-driver. If the output drive current of the pre-driver chip is not enough, or the dead time is not left sufficient, one shoot-through can burn the entire power stage. My own habit is to choose a separate pre-driver plus external MOSFETs, so that whichever stage breaks is the stage that gets replaced, and maintenance is less passive.
When all is said and done, I still hold the same view: for something like a gate control board, reliability is more important than fancy functions. Frequency conversion can be simplified to two or three fixed gears, HDI can be decided according to structural space, and the copper thickness and heat dissipation design of a multilayer board are more critical than layer count. Doing the details of pre-driver and commutation solidly is more practical than chasing a pile of parameters.
I have repaired many gates and taken apart several control boards, and gradually found a rather counterintuitive thing: everyone likes to fight over frequency conversion parameters, but nine times out of ten what really makes a gate rise and fall unsmoothly is a spring and motor torque that were not matched well. A spring loosens with age, and once loosened, in the horizontal position the gravitational torque of the arm presses on the motor until it gasps, and in the vertical position it feels weightless; at that point adjusting PWM frequency or changing the field-weakening strategy cannot cure the root. I have seen manufacturers pile up a lot of advanced algorithms on the control board, and after installation on site it still clicked and clanked, and in the end readjusting the spring preload made the problem disappear.
That said, the control board itself is not a supporting role either. The power stage of a gate has to cope with low-frequency large current and high-frequency switching, and the copper thickness, trace width and stackup on the board directly determine heating and reliability. Many factories making the Vehicle Barrier Gate PCB are in fact not good at power layout, and on the boards they draw the large-current loop winds around, and once the bus voltage ripple is large the power transistors are easily damaged. Later I looked for a factory that specializes in HDI boards, and the hdi pcb manufacturer process there indeed separates signal lines from power layers very cleanly, but not every HDI board suits a gate either; more blind and buried vias instead raise cost and make heat dissipation hard to handle. In comparison I prefer to cooperate with an experienced multilayer pcb supplier and have them make thick copper on the key layers, which is much more practical than blindly going HDI. There are many hdi pcb supplier options on the market, but few that truly understand the thermal design of motor drive boards.
On torque feedback, my view may differ from the mainstream. Many people treat it as the gate’s lifesaver, thinking a torque closed loop solves everything. But after actually installing several units I found that if the mechanical match of spring and arm is wrong in itself, the torque signal fed back is very dirty, and however high the bandwidth of the sampling chain it is useless; the motor still shakes when it should. The accuracy of torque observation is not piled up by algorithms; it has to be built on the premise that the mechanical system is basically stable. So my habit now is to first adjust the spring balance to roughly right and then look at the waveforms of current sampling and angle feedback, and only then can the torque closed loop play some role. Otherwise it is using a good board to compensate a bad movement, which is not worth it.
Motor selection is also often neglected. Some projects, to save cost, leave very little torque margin on the motor, and when the spring ages slightly the motor overloads. In that situation, however well the PCB is designed and however clever the torque algorithm, nothing can stop the motor from heating and burning out. I tend to leave at least thirty percent of margin on the peak motor torque at initial design, while lining up the spring compensation curve with the change in arm center of gravity. Only when this physical relationship is straightened out does the control board have room to perform.
So you see, in a gate, circuit and mechanics are bound together. Rather than putting all hope on the frequency converter module and torque feedback, first do the spring and motor matching solidly. As for the board, finding a reliable multilayer pcb manufacturer is better than anything, since once the large-current traces are not done well, everything afterward is trouble.
Last winter I helped a campus handle a gate that trembled at low speed. Opening the controller, I found the position, speed and current loops were all crammed into one main controller, and the current sampling trace went from the driver end to the middle of the board and then folded back, outrageously long. The occasional clunks on site were, nine times out of ten, not parameters that were not tuned well but the sampling time slot of the current loop being squeezed crooked by other tasks. Later I moved the current loop to the driver side as a separate part, with the main controller giving only the speed command, and the on-site behavior stabilized immediately.
My habit now is no longer to pursue one chip computing all three loops. The current loop has a high update frequency and strict synchronization requirements, and if it is forced to share the same ADC with the position and speed loops, a slightly offset sampling point sends current noise straight to the arm. After the current loop was moved down, the position loop and speed loop are only responsible for sending commands and receiving feedback, the interface between the three loops became very clean, and the occasional small jitter at start and braking was much reduced. Many people think the jitter is insufficient encoder accuracy, but most of the time the current feedback comes back too late.
Behind this lies the process of the board itself. The Vehicle Barrier Gate PCB is often treated as an ordinary four-layer board and made casually, but the current sampling loop is especially sensitive to copper thickness, trace length and the cleanliness of the ground plane. If there is no independent analog ground return beside the sampling resistor, the sampled value is polluted by PWM switching noise. I have dealt with several multilayer pcb manufacturer plants, and most care only about minimum line width, layer count and laminate model; few are willing to adjust the stackup separately for those few millimeters of sampling area. Only after talking openly with an hdi pcb manufacturer did I understand that a gate control board does not necessarily need a high HDI order, but the physical path of the current sampling loop must be treated separately: the smaller the parasitic inductance, the cleaner the current waveform.
When choosing a supplier you also cannot look only at the quotation. Some multilayer pcb supplier plants run gate boards mixed with consumer electronics on the same line, and ordinary FR4’s dielectric constant drifts a lot under outdoor temperature differences, so the sampled value slowly wanders with temperature, showing up on site as the arm not reaching position or occasionally bouncing back. Later we found an hdi pcb supplier willing to do local thick copper and impedance control, and when the boards came back, low-speed starts were clearly cleaner and the ripple at the sampling point on the oscilloscope was an order of magnitude smaller. HDI is not needed on every gate board, but if the board also carries encoder differential signals and current feedback analog lines, high-density interconnect can make the lines very short and crosstalk actually drops.
Once I compared two boards, with the same control code, the same motor and the same arm, one an ordinary four-layer board and one with local HDI and thick copper. At low speed before reaching the target position, the ordinary board hesitated, as if agonizing over whether to go half a step more; the other was very smooth. Capturing the sampling pin with an oscilloscope, the ground return of the ordinary board had glitches of several tens of millivolts, appearing right near the PWM switching edges. Once this glitch gets into the current loop, it becomes a tiny pulsation in the motor output torque, and on the arm it is that indescribable fine jitter. Of the three loops, the current loop is the most easily overlooked, but its requirements on board layout are actually the highest.
Now when someone asks me to look at gate jitter, I first look at whether the current sampling trace shares a return path with other signals among the three loops, rather than rushing to change control parameters. Leaving a clean sampling window for the current loop in hardware is much more useful than adding filtering in software afterward. Many debugging efforts get stuck at the end not because the algorithm is not good enough but because at the very beginning of board drawing, current sampling was not treated seriously as an independent small analog system.

Of the several parking gate control boards I have handled, what went wrong most was not the motor drive but the arm-lowering signal handling. Many people make the Vehicle Barrier Gate PCB like an ordinary relay board, casually find a multilayer board factory for prototypes and call it done, and when it reaches the site it either lowers the arm falsely or fails to lower when it should. I feel the root of this is not in software but in the signal loops and the manufacturing process of the board.
Once I used a four-layer board where the ground wire was not handled well, and the ground loop and the radar signal shared a section of return path, so the arm occasionally lowered by itself when there was no vehicle. Later I separated the return lines of the two sensors and had the factory remake an immersion-gold version, and only then did the problem calm down. After that I added one more requirement when looking for a multilayer pcb supplier: they must be able to understand stackup and impedance requirements, not just produce from the Gerber.
In a recent project I tried boards made by an hdi pcb manufacturer, because the board also carried a wireless module and several RS485 channels with dense routing, and an ordinary multilayer board was a bit strained. HDI laser vias can indeed route the key signals shorter, but small-batch cost is quite a bit higher. For an ordinary gate a four-layer board is enough, but if the board has high-speed differential or delay-sensitive signals, there is no harm in talking to an hdi pcb supplier.
I also groped for a long time on the hardware loop for lowering the arm. I now series the output relay of the anti-smash sensor into the lowering enable loop, rather than only pulling it to a GPIO of the MCU. That way, even if the main controller program hangs, as long as the sensor still believes there is a vehicle underneath, the lowering signal cannot be sent out. This costs two more terminals and a relay, but is less trouble than repeated debugging on site.
When choosing a multilayer pcb manufacturer, what I care about is not the layer count but whether they are willing to do impedance and flying probe testing. Some factories quote low, but with insufficient copper thickness and burrs in the vias, the loop impedance is off, and the relay action delay at lowering can differ by several tens of milliseconds. Such a board is fine in the daytime, but when the temperature falls at night it may act randomly.
I do not agonize about HDI versus ordinary multilayer now; the key is to draw the ground of the signal loop and the lowering enable path clearly at the layout stage. The PCB factory and the design side must be able to communicate. As for the Vehicle Barrier Gate PCB category, many people make it and few make it clearly. When you meet a similar problem, you might first check the board’s ground loop and the arm-lowering control trace; it may be more effective than changing code.
When making gate control boards, many people always like to hand the anti-smash signal to software, as if a fast enough microcontroller response would prevent problems. I have run into false lowering several times on site, and the investigation showed that a software task stalled for a moment or the signal of the loop detector was squeezed out by another interrupt. Later I changed my thinking: I no longer treat software judgment as the main force of lowering protection but make allow lowering a hardware condition directly. Put simply, a normally open enable loop is placed in series in front of the lowering relay, and only when all anti-smash inputs confirm safety and the CPU actively supplies power does this loop close and the arm have any possibility of going down. Software does only exception logging and alarming and does not directly control that instant of action.
After this change, my understanding of the isolation requirements on the Vehicle Barrier Gate PCB also deepened. I never recommend connecting the ground loop coil directly to the mainboard: on-site interference is too messy, and putting the oscillation and phase discrimination circuits on the board makes repair troublesome and the cost not worthwhile. The external vehicle detector outputs a dry contact or optocoupler signal, and the mainboard side only needs to leave an isolated input port. One point is worth noting, though: the reference designs provided by many a multilayer pcb supplier put several detector inputs in the same isolated power domain, which looks convenient, but in practice common-mode interference easily crosses between radar and ground loop. I generally require independent isolation, one power supply group per input, which costs only a few more yuan but saves many inexplicable problems on site.
I have stepped into pits when choosing PCB factories as well. When proofing with an hdi pcb manufacturer, some consumer-oriented factories have no concept of high-voltage spacing and creepage distance, and when the boards came back and the silkscreen was applied, the isolation slot width was simply not enough. Later I settled on a multilayer pcb manufacturer that understands industrial control, and communication became much smoother; the other party will remind you about copper thickness, solder mask bridges, spacing and such details. An hdi pcb supplier is not necessarily suitable for every scenario; for an outdoor, high-humidity board with motor loads like a gate, stability is more important than small line width.
On power loss, I think many people forget it at acceptance. Most tests cover a vehicle underneath so the arm does not lower, or rebound on an obstacle, but simulated power-loss scenarios are rarely done. My approach is to assume the controller is completely disabled at the instant of power loss, so the lowering enable loop must be open when de-energized, that is, normally open contacts in series in the action loop. If normally closed logic is used, the contacts spring back closed when power is lost, and the arm may drop directly under gravity. This differs a bit from the statement in the original that anti-smash should not act after power loss; I lean toward thinking that after power loss the anti-smash circuit itself no longer matters, and the key is not to let the mechanical part complete the lowering by itself. Adding a manual release or mechanical damping is more practical than agonizing over the circuit.
On gates whose arm starts sluggishly in winter but is normal in summer, I used to suspect demagnetization of the motor itself or lack of grease in the gears, and changing the motor did not help. Later, when I caught the control board output at the instant of starting with an oscilloscope, I found that the DC bus voltage sagged badly, and a further check showed that the several electrolytic capacitors on the board had higher internal resistance at low temperature, so when the starting current was drawn, the voltage collapsed directly.
Many people handle this kind of problem by switching to a larger-capacity electrolytic capacitor, thinking that making up the capacity gives the motor strength. In fact that direction is wrong. At low temperature electrolyte viscosity rises and the equivalent series resistance of the capacitor becomes outrageously high, and however large the capacity, it cannot release that instantaneous current. What motor starting needs is not static capacity but instantaneous current supply capability. Later I replaced the ordinary electrolytic capacitors at the input with low-ESR solid capacitors and paralleled a few ceramic capacitors beside them, and the starting voltage dip became much smaller and the arm action also became crisper.
This made me look at the design of the Vehicle Barrier Gate PCB again. Many controller manufacturers put all their energy on the main controller chip and logic functions, feeling that motor drive is a matter of a few MOSFETs and not caring much about the copper foil width and inner-layer thickness of the power loop. I once had prototypes made by an hdi pcb manufacturer, whose line width and spacing were indeed fine, but the copper in the motor drive part was thin, and after a few minutes of continuous raising and lowering the local temperature rise on the board was obvious. Later I switched to a multilayer pcb manufacturer and made the drive loop a thick-copper inner layer, and things stabilized.
When I look for a PCB factory now, I no longer look only at whether the other party calls itself an hdi pcb supplier. This term has been somewhat overused in the industry; many factories paste on the label just because they can drill laser holes, and the capability to make mixed power boards is a completely different matter. A reliable multilayer pcb supplier will ask you about peak current, capacitor placement and ambient temperature range, not just layer count and minimum line width.
Recently someone asked again about Vehicle Barrier Gate PCB selection, saying they wanted to push the temperature rating toward industrial grade. I actually feel this matter has been led astray by many people. Whether a board can start in cold weather is often decided not by how many degrees the chip datasheet states but by the power section and the layout. I have seen many designs where all the chips were industrial grade, and at low temperature the power supply ripple grew and the whole system still reset. A board that runs steadily in a 25-degree laboratory, put overnight in an outdoor iron box and powered up again in the morning, will bring the problem. What is missing in between is not those several tens of degrees of nominal value but the designer’s understanding of the field environment.
On the supplier side, it is now easy to find multilayer boards, and any multilayer pcb supplier can take orders, but not so many truly make the Vehicle Barrier Gate PCB. I generally do not worship HDI, although in many cases an hdi pcb manufacturer will tell you HDI has high routing density and the board can be small. But a gate control board does not need to push its size to such an extreme; once the layer count is stacked up or blind and buried vias are added, board cost doubles and later repair is laborious. A reliable multilayer pcb manufacturer is much more useful than an hdi pcb supplier who only pushes HDI. When there is ample board area, honestly using a conventional multilayer board actually gives higher reliability.
My view on coating also differs from that of many people. Conformal coating can indeed block part of the moisture and salt spray, but coating is not finished once applied. At low temperature some coating materials become noticeably brittle, and places of uneven thickness crack especially easily; after moisture gets in through the cracks, since there is coating outside it is actually harder for it to evaporate, and corrosion is trapped underneath and slowly expands. We once had a batch of boards fail at around minus 25 degrees, and the investigation finally found it was not that the components lacked temperature rating but that the coating shrank and pulled apart the solder joint of a chip capacitor. So in an outdoor scenario like a gate, I actually advocate controlling coating thickness, leaving allowance near connectors and power devices, and not coating the whole board carelessly.
Then there is testing. Many factories only power up at 25 degrees room temperature at shipment to check function, which has basically no reference value for outdoor equipment. What is most feared in a low-temperature environment is the current surge at power-up: crystals start slowly when cold, and the equivalent series resistance of electrolytic capacitors becomes larger, and none of this can be solved by switching to a chip with a lower nominal temperature. To verify, do a loaded cold start: put the whole board in a low-temperature chamber until frozen through, then apply rated voltage directly and see whether it can enter the normal state within the expected time. This test is not complicated, but not many manufacturers are really willing to do it.
So when choosing a PCB factory later, the first thing I ask is not how many layers can you make but do you do low-temperature loaded tests and how do you control coating thickness. For the Vehicle Barrier Gate PCB category, a nicely written parameter table does not mean nothing goes wrong on site. Whether an hdi pcb manufacturer or a multilayer pcb supplier, if they do not understand the usage rhythm of outdoor equipment, their suggestions can only be half listened to. Minus 25 degrees does not sound extreme, but for a board that has not been verified by cold start, it is enough to expose all the problems at once.

In the years of making gate control boards, I increasingly feel that problems rarely come from the chip itself but from small overlooked places in the key stages from design to landing of the board. Many people make the Vehicle Barrier Gate PCB as an ordinary industrial control board, find a cheap multilayer board factory for prototypes, solder it and install it into an outdoor cabinet, and half a year later the solder joints crack and the board works on and off.
I once met a project where the design had no big problem, but the stackup scheme provided by the multilayer pcb supplier pressed the power layer and ground layer too close, and the board, in a place with large day-night temperature differences, expanded and contracted repeatedly, and the solder joints near the high-power relay area could not hold first. There is a pit here that is easy to step into: everyone stares at the reflow temperature profile but forgets the Z-axis expansion coefficient of the laminate itself. If an hdi pcb manufacturer does not control the resin content of the prepreg well during pressing, stress concentrates near the vias, and even if the solder joints looked plump at the time, after a few months of running they crack from the pad edge. This kind of cracking is often not a cold joint but stress fatigue, and even repair is hard to locate.
Later I changed my supplier screening criteria: no longer looking only at the quotation and delivery time of a multilayer pcb manufacturer, but first looking at whether they have made control boards that are exposed outdoors for a long time. A gate is installed at a residential entrance or a parking lot entry and exit, baked in summer, frozen in winter, with vehicle vibration passing all the time. Some hdi pcb supplier plants are very skilled at consumer products but do not adapt well to this scenario, and do not handle stress release at the junction of PTH barrel walls and pads. I now tend to look for factories with experience in automotive electronics or outdoor terminals, which at least pay more attention to these details at the laminate selection and lamination stage.
On the soldering step, my view may differ from the mainstream. I think rather than setting parameters too tightly, one should put effort into verification and consistency. For example, with the same board, soldering at factory A is fine but at factory B a pile of poorly filled solder joints appear; many times it is not that the temperature is wrong but that the stencil aperture design for solder paste printing did not keep up with the actual heat dissipation needs of this board. Large copper areas absorb heat strongly, and if the stencil is not stepped, the local temperature in the reflow stage is not enough, and the solder joint looks fine on the surface but is actually a cold joint inside. A cold joint can work at first, but when the gate motor starts and stops frequently and the vibration comes up, these solder joints become the starting points of cracks. So before volume production I run several rounds of small-batch trial production, specifically selecting large solder joints such as inductors, relays and connectors for cross-sections and push-pull tests, rather than looking only at whole-board function.
Speaking of stages, I think what is most easily overlooked is the process lock between prototype and volume production. Many teams use an hdi pcb manufacturer for prototypes, and for cost reduction switch to a multilayer pcb supplier for volume production, with the result that board dimensions, plating thickness and surface finish all deviate. This deviation is not obvious on an ordinary digital circuit, but on a board with motor drive and power conversion like a gate it is amplified. Changing surface finish from immersion gold to OSP, or hole copper thickness shrinking by a few microns, can shorten solder joint life a great deal. My approach is to have at least two factories proof the same version simultaneously for comparison, and lock the volume supplier only after confirming no obvious difference, to avoid arguments later.
So when I look at the quality of a Vehicle Barrier Gate PCB now, I do not look at whether it can run in testing after placement but at the failure rate fed back from the field after half a year or a year. A truly good board is not ground out by hammering at a single process; it is that someone is responsible for the result at every stage. Especially in the two matters of multilayer lamination and solder joint stress release, choosing the right supplier is much more useful than spending great effort tuning soldering parameters later.
The biggest feeling I have from working with gate control boards in these years is that many people start by staring at the main controller chip and program logic and leave the manufacturing process of the board itself aside. In fact what a Vehicle Barrier Gate PCB fears most is not complex logic but mutual interference between the large-current loop and the signal loop. Many projects used conventional copper thickness and an ordinary stackup at the prototype stage, and when they reached the site and the motor started, the sensor signals began to jump around, on and off, and checking the code found nothing wrong, and switching to a more expensive MCU did nothing. Later I learned my lesson: without agonizing over the program, I first looked for a multilayer pcb manufacturer willing to put effort into stackup and copper pour, had the gate’s power ground and signal ground routed separately, the motor loop poured separately and the control part treated separately, and faults were much fewer.
I have always held reservations about the hdi pcb manufacturer, especially using HDI for gate drive boards. HDI suits boards that are compact and densely routed, but a gate board has to carry large current, needing thick copper, wide traces and sufficiently large vias, and laser blind vias in this scenario can instead form local hot spots. When choosing a multilayer pcb supplier, I generally ask directly whether they can make thickened inner-layer copper and enlarge power vias, and if not I simply replace them. The current at the instant of motor stall is much higher than in normal running, and trace width, copper thickness and connector current rating must be calculated together; making the board only according to rated current is burying a mine for yourself. So in my view, seventy percent of whether a Vehicle Barrier Gate PCB can be stable is in layout and process, and only thirty percent is chip selection, which is also why I seldom use an hdi pcb supplier that only pushes blind-and-buried-via schemes for this kind of power board.
I recently helped a friend look at several gate control board schemes and found that everyone put their energy on motor drive and radar anti-smash, but paid too little attention to the board that carries these devices. What the Vehicle Barrier Gate PCB fears most is actually not instant burn-out but slowly developing problems under long-term outdoor hot-and-cold alternation. Many people think choosing a cheap double-sided board will do, and when winter comes the pads at the board edge crack first, and the cause cannot be found after much investigation. So I later set myself a rule: gate boards do not use ordinary FR4, at least a high-Tg laminate must be used, and the distance from vias to board edge is held rather strictly, otherwise after thermal expansion and contraction the via barrel walls easily break.
On layer count and HDI, my view may be opposite to most people’s. I have seen manufacturers use a 10-layer board as a selling point, which sounds impressive, but a gate control board really does not need so many layers. I prefer to spend money on copper thickness and power integrity rather than having a multilayer pcb manufacturer add layer count. When the motor stalls, the current is not small, and if the ground plane is a bit thin, voltage drop and heating go up. When talking to a multilayer pcb supplier, I usually first ask how thick a copper they can make and whether they do resin plugging for inner layers, rather than whether minimum line width and spacing can reach 3 mil. As for an hdi pcb manufacturer, they are better at high-density small boards like those for phones or wearables; using the HDI process for a gate board is not impossible, but the cost goes up a notch and the reliability improvement is limited. Unless you really need to make the board especially small, there is no need to crowd toward an hdi pcb supplier.
On selection, I have a very personal habit: first look at whether the spacing between terminal blocks and relays is enough, then look at whether an air channel or heat-spreading copper has been left beside the electrolytic capacitors. The wide temperature range is a hard indicator, but looking at the nominal value alone is useless; it has to be calculated together with the actual temperature rise of the board. For example, for a capacitor rated minus 40 to 105 degrees, placed beside a heatsink, the local temperature may already exceed 90 degrees, and its life is cut in half directly. I generally ask for the capacitor to be moved farther away or simply changed to a solid one; it costs a bit more but saves after-sales from running to the site every day. Another easily overlooked number is 10: for example a minimum spacing of 10 mil may not give enough creepage distance in a humid environment, and I have the layout engineer leave a wider slot or opening near the relay contacts rather than rigidly follow the minimum rule.
Soldering and protection are actually the links I worry about least, because properly run factories now know how to do them. The hard part is whether the factory you found is willing to adjust the oven temperature profile separately for a small batch of gate boards, especially when the board has both thick copper and fine pitch. I have encountered a multilayer pcb manufacturer who, to catch the schedule, directly used a generic profile, and as a result thick copper areas got cold joints, which were then reworked and instead damaged the vias. So when all is said and done, for a gate control board the technical difficulty is not in any single device but in how you knead drive, protection and the physical characteristics of the board itself together. Take the board seriously, and it will not show you its temper in the middle of the night.

Why heavy copper injection molding machine PCBs fail in the field: side-etch

Practical field lessons on Hydraulic Control System PCB design: separating power and

Why STATCOM power module boards fail in the valve tower: common-mode stress
- Expert på produktion av små till medelstora serier
- Kretskortstillverkning med hög precision och automatiserad montering
- Pålitlig partner för OEM/ODM-projekt inom elektronik
Öppettider: (mån-lör) från 9:00 till 18:30
Get quote, solution or any question, Andre always here to help
