Decoding the Dot: A Deep Dive into Barcode Label Printer Electronics - Extended Section 6 |
Subtitle: Power MOSFET Bank for Dot Control - The Heavy Lifters of Printing |
Introductory Summary (Extended Section 6 Preview) |
In the previous section, we explored the driver circuits that shift and latch print data. But those circuits only handle the logic - they decide which dots should fire. The actual work of delivering the high current to each tiny resistor is done by the power MOSFET bank. This is the muscle of the printer. A thermal printhead may have hundreds or thousands of dots, and each dot needs a transistor that can switch up to 50 milliamperes of current at 24 volts, with microsecond precision. That does not sound like much per dot, but when you multiply it by 832 dots, you get peak currents of over 15 amperes. The MOSFET bank must handle this current without overheating, without introducing excessive voltage drop, and without generating electrical noise that could interfere with the logic. This chapter is devoted entirely to the power MOSFETs that control each dot individually. We will explain how a MOSFET works in plain language, why its on-resistance is so critical, how it is driven by the shift registers, and how it is protected from over-current and over-temperature. We will look at real-world designs from major companies: Infineon's low-voltage power MOSFETs that are widely used in printer driver arrays, ON Semiconductor's integrated driver chips that combine logic and power, Texas Instruments' intelligent power switches, and Vishay's high-efficiency MOSFET families. We will also examine the trade-offs between discrete MOSFETs and integrated driver ICs, the thermal management of the MOSFET bank, the layout considerations for high-current traces, and the protection circuits that prevent catastrophic failures. By the end, you will understand why the MOSFET bank is the most carefully designed part of the printer's electronics, and you will appreciate the engineering that goes into making hundreds of transistors work together in perfect harmony. |
|
Chapter 1: The MOSFET - A Voltage-Controlled Switch |
To understand the MOSFET bank, we must first understand a single MOSFET. MOSFET stands for Metal-Oxide-Semiconductor Field-Effect Transistor. It is a three-terminal device: the gate, the drain, and the source. In a thermal printhead application, the drain is connected to the heating resistor, the source is connected to ground, and the gate is connected to the output of the shift register. When a positive voltage is applied to the gate relative to the source, the MOSFET turns on, allowing current to flow from the drain to the source. When the gate voltage is zero, the MOSFET turns off, blocking the current. The MOSFET is voltage-controlled - it draws almost no current from the gate, which is why it can be driven directly by the shift register's output. The MOSFET also has a very fast switching speed - it can turn on and off in nanoseconds. The key parameter of a MOSFET is its on-resistance, or Rds(on), which is the resistance between the drain and the source when the MOSFET is fully turned on. A low on-resistance means less voltage drop and less power dissipation. For a thermal printhead driver, the on-resistance should be less than 50 milliohms. The MOSFET also has a maximum current rating, a maximum voltage rating, and a maximum power rating. The MOSFET must be chosen so that these ratings are never exceeded. |
Design Example: Infineon OptiMOS 40V Family |
Infineon's OptiMOS 40V family is a popular choice for printer MOSFET banks. The BSC028N06NS, for example, is a 60-volt, 20-ampere, 2.8-milliohm MOSFET - which is a very low on-resistance. In a printer design from a European OEM, 832 of these MOSFETs are used - one for each dot. However, using 832 discrete MOSFETs is impractical because of the PCB area and the assembly cost. Instead, the MOSFETs are integrated into multi-channel driver ICs. The OptiMOS family is used in the discrete designs for very high-power printers where the current per dot is higher than the integrated drivers can handle. For example, some industrial printers use a 12-volt head with 100-milliampere dots, requiring MOSFETs with an on-resistance of less than 10 milliohms. The manufacturer of a heavy-duty logistics printer uses 16 OptiMOS MOSFETs in a TO-220 package, each driving a block of 52 dots, with a total of 832 dots. The MOSFETs are mounted on a large heatsink, and they are driven by a CPLD through gate driver ICs. |
|
Chapter 2: The Integrated Driver IC - Power and Logic in One Package |
For most printers, the MOSFETs are integrated into the driver IC along with the shift register and latch. This is the most common approach because it saves space, reduces cost, and simplifies the design. The integrated driver IC has a serial input for data, a clock input, a latch input, a strobe input, and multiple power outputs. The power outputs are MOSFETs that are turned on and off by the internal logic. The integrated driver IC also has built-in protection features - over-current protection, thermal shutdown, and over-voltage protection. The integrated driver IC is designed for a specific current rating, and it is optimized for thermal performance. The IC is typically packaged in a QFP or a QFN package with a large thermal pad on the bottom, which is soldered to the PCB to conduct heat away. The integrated driver IC is the most cost-effective solution for most printers. |
Design Example: Rohm BH12 Integrated Driver |
Rohm's BH12 series, which we mentioned earlier, is an excellent example of an integrated driver IC. Each BH12 has 12 channels, each with a shift register bit, a latch, and a power MOSFET with an on-resistance of 40 milliohms. The BH12 can handle a maximum current of 200 milliamperes per channel, which is more than enough for a standard 500-ohm dot at 24 volts (which draws 48 milliamperes). The BH12 also has a built-in 5-volt regulator, thermal shutdown at 150C, and over-current protection. In a design from a Taiwanese printer OEM, 70 BH12 ICs are cascaded to control an 832-dot head. The ICs are placed along the edge of the PCB, close to the printhead connector. The manufacturer chose the BH12 because it is inexpensive, readily available, and has a proven track record in many printer designs. The BH12's thermal pad is soldered to a copper area on the PCB, and the manufacturer uses a 4-layer PCB with a ground plane to spread the heat. The manufacturer reports that the BH12 ICs operate at 85C at full load, which is well within the 150C maximum. |
|
Chapter 3: The Discrete MOSFET Approach - When Integration Is Not Enough |
There are situations where integrated driver ICs cannot be used. For example, if the printhead requires a very high current per dot (e.g., 100 milliamperes), the integrated driver might not have enough current capacity. Or if the printer operates at a high voltage (e.g., 48 volts), the integrated driver might not be rated for it. In these cases, discrete MOSFETs are used. The discrete MOSFETs are controlled by a separate shift register and gate driver IC. The shift register provides the logic signal, the gate driver amplifies it to the required gate voltage, and the MOSFET switches the current. The discrete MOSFET approach is more flexible because you can choose the MOSFET with the exact on-resistance and voltage rating you need. However, it is more expensive and takes more PCB area. |
Design Example: Discrete MOSFET Bank in a Zebra High-End Printer |
Zebra's ZT600 series, which is used in high-speed logistics applications, uses a discrete MOSFET bank for the printhead. The head operates at 48 volts (not 24), and each dot draws 60 milliamperes. The integrated drivers available at the time of the design were not rated for 48 volts, so Zebra used discrete MOSFETs. The MOSFETs are 100-volt, 5-ampere, 50-milliohm devices from Vishay. There are 832 MOSFETs, but they are not in individual packages - they are in a multi-chip module (MCM) that contains 16 MOSFETs per package. So there are 52 packages on the PCB. The MCMs are mounted on a large heatsink, and a fan blows air over them. The gate drivers are separate ICs - the 74LV244 buffer - that amplify the shift register's 5-volt output to 12 volts, which is sufficient to fully enhance the MOSFETs. The manufacturer reports that this design is more expensive but more reliable than an integrated driver, because the MOSFETs are over-specified and run cool. |
|
Chapter 4: The On-Resistance - A Critical Parameter |
The on-resistance (Rds(on)) of the MOSFET determines the voltage drop and the power dissipation. The voltage drop is the current times the on-resistance. For a dot drawing 48 milliamperes, a 40-milliohm on-resistance gives a voltage drop of 1.9 millivolts - negligible. But for a dot drawing 100 milliamperes, the same on-resistance gives a drop of 4 millivolts - still negligible. The real problem is the power dissipation: P = I^2 * R. For 48 milliamperes and 40 milliohms, P = 0.048^2 * 0.04 = 0.000092 watts - negligible. For 100 milliamperes and 40 milliohms, P = 0.01 * 0.04 = 0.0004 watts - still negligible. The power dissipation per MOSFET is small because the current is small. However, there are hundreds of MOSFETs, and the total power dissipation is the sum of all the individual dissipations. For 832 dots at 48 milliamperes, the total power in the MOSFETs is 832 * 0.000092 = 0.077 watts - still small. The real power dissipation in the MOSFET bank is due to the switching losses - the energy lost each time the MOSFET turns on and off. The switching loss is proportional to the switching frequency and the voltage and current. However, the switching frequency of the printhead is low (only a few kilohertz), so the switching losses are also small. The main heat source in the MOSFET bank is actually the current through the PCB traces, not the MOSFETs themselves. |
Design Example: On-Resistance Selection in a Sato Printer |
Sato's printer uses the BH12 driver, which has a 40-milliohm on-resistance. The manufacturer calculated the power dissipation per dot and found it to be 0.1 milliwatts - negligible. The total power dissipation in the BH12 ICs is 70 chips * 12 channels * 0.1 mW = 84 mW, which is also negligible. The heat in the BH12 ICs actually comes from the logic and the quiescent current, not from the MOSFETs. The manufacturer chose the BH12 because its 40-milliohm on-resistance was more than sufficient for the 48-milliampere dot current, and the lower on-resistance would not have provided any benefit. The manufacturer also noted that a lower on-resistance would have been more expensive and would not have improved the print quality. |
|
Chapter 5: The Gate Drive - Turning the MOSFET On and Off |
The gate of a MOSFET is a capacitor. To turn the MOSFET on, you must charge this capacitor to a voltage higher than the gate threshold. To turn it off, you must discharge it. The gate driver must be able to source and sink current to charge and discharge the gate capacitance quickly. The gate drive voltage is typically 5 volts for logic-level MOSFETs, but some MOSFETs require 10 to 12 volts for full enhancement. The gate drive current is determined by the gate capacitance and the desired switching speed. For a fast switching speed (e.g., 10 nanoseconds), the gate drive current might be 1 ampere. For a slower switching speed (e.g., 100 nanoseconds), the gate drive current might be only 100 milliamperes. In an integrated driver, the gate drive is built-in. In a discrete design, an external gate driver IC is used. The gate driver must be placed close to the MOSFET to minimize the parasitic inductance of the gate trace. |
Design Example: Gate Drive in Zebra's Discrete Design |
Zebra's discrete design uses a 74LV244 gate driver buffer. The buffer is powered by 12 volts, which provides the gate drive voltage. The buffer has a peak output current of 15 milliamperes, which is sufficient to drive the MOSFETs' gates at a 100-nanosecond switching speed. The buffer is placed on the PCB near the MCM packages. The gate traces are short and wide to minimize the inductance. The manufacturer measured the gate voltage and found it to be 11.8 volts, which is sufficient to fully enhance the MOSFETs. The manufacturer also added a 10-ohm resistor in series with each gate to dampen any ringing. |
|
Chapter 6: The Body Diode - An Unwanted Friend |
Every MOSFET has an intrinsic body diode between the source and the drain. The body diode is formed by the pn junction in the MOSFET's structure. The body diode can conduct current in the reverse direction when the MOSFET is turned off. In a thermal printhead, the body diode is not a problem because the current only flows in one direction - from the drain to the source. However, if the voltage spikes above the rail, the body diode can conduct and clamp the voltage. The body diode has a forward voltage drop of about 1 volt and a reverse recovery time that can cause switching losses. In most printer designs, the body diode is ignored because the current is low and the switching frequency is low. |
Design Example: Body Diode in BH12 |
The BH12 integrated driver has an internal body diode. The manufacturer did not consider the body diode in their design because the peak voltage is only 24 volts, and the forward current is small. The body diode's reverse recovery time is 50 nanoseconds, which is negligible compared to the 500-microsecond strobe pulse. The manufacturer did not add any external diodes. |
|
Chapter 7: The Switching Speed - Fast Enough Is Good Enough |
The switching speed of the MOSFET determines the rise and fall times of the current through the dot. A fast switching speed reduces the switching losses and allows a shorter minimum strobe width. However, a fast switching speed also generates more electromagnetic interference and can cause ringing due to the parasitic inductance of the traces. The switching speed is controlled by the gate driver's output current and the gate resistor. A higher gate resistor slows down the switching, reducing the EMI but increasing the switching losses. The optimal switching speed is a trade-off. For a thermal printhead, a rise time of 10 to 20 nanoseconds is typical. |
Design Example: Switching Speed in Brother Printers |
Brother's printer uses an integrated driver IC with a rise time of 15 nanoseconds. The manufacturer measured the EMI and found it to be within the limits. They did not add any gate resistors because the integrated driver's output impedance is already high enough to limit the switching speed. The manufacturer found that a faster rise time (e.g., 5 nanoseconds) would have increased the EMI by 10 dB, which would have required additional filtering. |
|
Chapter 8: The Maximum Current - Derating for Safety |
The MOSFET has a maximum continuous drain current rating. For a printer, the maximum current per dot is known, and the MOSFET is chosen with a rating that is at least twice the expected current. This derating ensures that the MOSFET operates in a safe region and has a long lifetime. The derating also accounts for the temperature - the MOSFET's current rating decreases with temperature. For example, a MOSFET rated for 5 amperes at 25C might only be rated for 3 amperes at 100C. The derating is typically 50% to 80%. |
Design Example: Derating in Sato Printers |
Sato's printer uses the BH12 driver, which has a maximum current rating of 200 milliamperes per channel. The dot current is 48 milliamperes, so the derating factor is 48/200 = 24%. This is a very conservative derating. The manufacturer could have used a cheaper driver with a lower current rating, but they chose the BH12 for its availability and reliability. The BH12's current rating also gives headroom for future upgrades. |
|
Chapter 9: The Maximum Voltage - Staying Within the Limits |
The MOSFET has a maximum drain-source voltage rating. For a printer, the voltage is the 24-volt rail, but there can be voltage spikes due to the inductive kick of the traces. The voltage rating must be higher than the maximum expected voltage, including the spikes. A typical voltage rating is 30 volts for a 24-volt system, but 40 or 60 volts is safer. The voltage rating also affects the on-resistance - a higher voltage rating usually means a higher on-resistance. |
Design Example: Voltage Rating in Zebra Printers |
Zebra's discrete design uses 60-volt MOSFETs for a 24-volt system. The manufacturer chose 60 volts because the spikes from the inductive cabling could reach up to 40 volts. The 60-volt MOSFET provides a good safety margin. The manufacturer tested the voltage spikes and found them to be 35 volts maximum, so the 60-volt MOSFETs are safe. |
|
Chapter 10: The Thermal Resistance - Heat Flow from Junction to Air |
The MOSFET's junction temperature is determined by the power dissipation, the thermal resistance from the junction to the ambient, and the ambient temperature. The thermal resistance is specified in degrees Celsius per watt. For a discrete MOSFET in a TO-220 package, the junction-to-ambient thermal resistance is about 62C per watt. For a surface-mount MOSFET on a PCB with a copper pad, the thermal resistance is about 30C per watt. For an integrated driver with a thermal pad, the thermal resistance is about 20C per watt. The junction temperature must be kept below the maximum (typically 150C). The thermal design ensures that the junction temperature is within the limit. |
Design Example: Thermal Resistance in BH12 |
The BH12 has a junction-to-ambient thermal resistance of 25C per watt when soldered to a 2-square-centimeter copper pad. The power dissipation per BH12 is 0.84 watts (from the quiescent current and the driver logic). The temperature rise is 0.84 * 25 = 21C. At an ambient of 25C, the junction temperature is 46C - well below the 150C maximum. The manufacturer's thermal design is more than adequate. |
|
Chapter 11: The Heat Sink - Dissipating the Power |
For high-power printers with discrete MOSFETs, a heat sink is required. The heat sink is a metal finned structure that increases the surface area for convection. The heat sink is attached to the MOSFETs with thermal grease. The heat sink is often an extruded aluminum piece that is custom-designed for the printer. The heat sink is placed in the airflow from a fan, which enhances the cooling. |
Design Example: Heat Sink in Zebra's Discrete Design |
Zebra's discrete design uses a large aluminum heat sink that spans the entire row of MCM packages. The heat sink has a thermal resistance of 1C per watt. The total power dissipation in the MOSFET bank is 20 watts. The temperature rise is 20C above ambient. With a fan blowing air at 5 meters per second, the effective thermal resistance is reduced to 0.5C per watt, giving a temperature rise of only 10C. The manufacturer reports that the MOSFETs operate at 50C, which is very cool. |
|
Chapter 12: The Parallel Connection - Sharing the Current |
Sometimes, a single MOSFET is not enough to handle the current, or the on-resistance is too high. In such cases, multiple MOSFETs can be connected in parallel. When MOSFETs are paralleled, they share the current. However, they must be matched in on-resistance and gate threshold to ensure current sharing. If one MOSFET has a lower on-resistance, it will carry more current and heat up more, which can lead to thermal runaway. The designer must ensure that the MOSFETs are from the same batch and that the layout is symmetrical. |
Design Example: Parallel MOSFETs in a Heavy-Duty Printer |
A manufacturer of a heavy-duty logistics printer uses 2 MOSFETs in parallel for each dot, because the dot current is 120 milliamperes, and a single MOSFET would have too high a voltage drop. The MOSFETs are 100-milliohm devices, and the parallel combination gives 50 milliohms. The manufacturer selected MOSFETs from the same batch and used a symmetrical layout - each MOSFET has its own gate resistor, and the traces from the common drain and source points are of equal length. The manufacturer tested the current sharing and found it to be within 5% - acceptable. |
|
Chapter 13: The Gate Resistor - Damping the Ringing |
A gate resistor is often placed in series with the gate of the MOSFET. The gate resistor serves two purposes: it limits the peak gate current, and it dampens the ringing caused by the parasitic inductance and capacitance of the gate circuit. The gate resistor value is typically 10 to 100 ohms. A higher resistor slows down the switching, reducing the EMI but increasing the switching losses. The gate resistor is placed as close as possible to the MOSFET's gate pin. |
Design Example: Gate Resistor in Zebra's Design |
Zebra's discrete design uses a 10-ohm gate resistor for each MOSFET. The resistor is a surface-mount 0805 package. The manufacturer measured the gate waveform and found that the 10-ohm resistor eliminated the ringing without slowing down the switching too much - the rise time was 15 nanoseconds. The manufacturer tried a 100-ohm resistor, which increased the rise time to 50 nanoseconds and caused a slight increase in the power dissipation. The 10-ohm resistor was the optimal choice. |
|
Chapter 14: The Snubber Circuit - Suppressing the Inductive Kick |
The traces and the cable have parasitic inductance. When a MOSFET turns off, the current through this inductance continues to flow, creating a voltage spike. This spike can exceed the MOSFET's voltage rating and damage it. A snubber circuit is used to suppress the spike. The snubber is a resistor and a capacitor in series, placed across the MOSFET's drain and source. The snubber absorbs the energy of the spike. The snubber also reduces the EMI. The snubber's resistor and capacitor values are chosen based on the parasitic inductance and the peak current. |
Design Example: Snubber in Sato Printers |
Sato's printer uses a simple snubber across the printhead connector, not across individual MOSFETs. The snubber is a 100-ohm resistor and a 1-nanofarad capacitor. The snubber is placed on the main board, close to the FFC connector. The manufacturer measured the spike without the snubber and found it to be 35 volts. With the snubber, the spike was reduced to 26 volts, which is within the 30-volt rating of the driver ICs. The snubber also reduced the EMI, helping the printer pass the FCC test. |
|
Chapter 15: The Current Sharing in Integrated Drivers - Matching the Channels |
In an integrated driver IC, the MOSFET channels are manufactured on the same die. They are closely matched in on-resistance and gate threshold. The current sharing between channels is not an issue because each channel is separate and each dot is separately controlled. However, the ground and power distribution inside the IC must be designed to handle the total current. The IC has multiple ground pins and power pins to reduce the parasitic resistance. The PCB layout must also ensure that the current returns to the ground plane with minimal resistance. |
Design Example: Current Sharing in Rohm BH12 |
The BH12 has multiple ground and power pins to distribute the current. The manufacturer's PCB layout uses a solid ground plane under the BH12 IC. The ground pins are connected to the plane with multiple vias. The manufacturer measured the current in each channel and found it to be consistent - the variation was less than 2%. |
|
Chapter 16: The Over-Current Protection - Limiting the Fault Current |
If a dot is shorted (e.g., the glaze is damaged and the resistor is bypassed), the current through the MOSFET can exceed its rating. The over-current protection limits the current to a safe level. In an integrated driver, the over-current protection is built-in. It senses the current through the MOSFET and, if it exceeds a threshold, it turns off the MOSFET. The over-current protection is typically pulse-by-pulse - it limits the current on each cycle. If the fault persists, the IC enters a latched-off state or a hiccup mode. |
Design Example: Over-Current in BH12 |
The BH12 has a built-in over-current protection that trips at 300 milliamperes - much higher than the normal 48 milliamperes. The manufacturer tested this by shorting a dot to ground. The current rose to 300 milliamperes, and the BH12 turned off the output. The BH12 entered a latched-off state, and the output remained off until the power was cycled. The manufacturer was satisfied with this protection. |
|
Chapter 17: The Short-Circuit Protection - A Subset of Over-Current |
A short circuit is a severe case of over-current, where the resistance is zero. The short-circuit protection is typically the same as the over-current protection - it limits the current to a safe level. In some designs, a separate comparator is used to detect a short circuit quickly (within microseconds) and turn off the output. The short-circuit protection must be fast enough to prevent the MOSFET from overheating and failing. |
Design Example: Short-Circuit in Zebra's Discrete Design |
Zebra's discrete design uses a fast comparator (LM393) to detect a short circuit. The comparator compares the voltage across a sense resistor to a reference. If the voltage exceeds the reference, the comparator triggers a latch that turns off the gate driver. The latch is reset by the CPU. The manufacturer tested the short-circuit response and found it to be 2 microseconds - fast enough to protect the MOSFETs. |
|
Chapter 18: The Avalanche Rating - Surviving the Spikes |
When a MOSFET turns off, the inductive energy can cause a voltage spike that exceeds the MOSFET's breakdown voltage. The MOSFET can enter avalanche mode, where it conducts current even though it is turned off. The avalanche rating is the maximum energy the MOSFET can absorb without being damaged. The avalanche energy is determined by the parasitic inductance and the peak current. For a printer, the avalanche energy is small, but the MOSFET should still be rated for some avalanche energy. |
Design Example: Avalanche in Brother Printers |
Brother's printer uses MOSFETs with an avalanche rating of 100 millijoules. The parasitic inductance is about 100 nanohenries, and the peak current is 15 amperes, giving an avalanche energy of 0.5 * L * I^2 = 0.5 * 100e-9 * 15^2 = 0.011 millijoules - much smaller than the rating. The manufacturer did not need to add any additional protection. |
|
Chapter 19: The Temperature Coefficient - A Self-Balancing Effect |
The on-resistance of a MOSFET increases with temperature. This is a positive temperature coefficient. If one MOSFET in a parallel array heats up, its on-resistance increases, so it conducts less current, which cools it down. This is a self-balancing effect that helps with current sharing. The positive temperature coefficient is a desirable feature. |
Design Example: Temperature Coefficient in Zebra's Design |
Zebra's MOSFETs have a positive temperature coefficient. The manufacturer tested the current sharing at 25C and at 100C and found that the current sharing improved at higher temperatures. The manufacturer did not need any additional ballast resistors. |
|
Chapter 20: The Layout of the MOSFET Bank - A High-Current Challenge |
The layout of the MOSFET bank is the most critical part of the PCB design. The high-current traces must be wide and short to minimize the resistance and inductance. The traces for the 24-volt rail and the ground must be capable of carrying 15 amperes without excessive voltage drop. The traces are typically 3 to 5 millimeters wide with 2-ounce copper. The MOSFETs are placed close to the printhead connector to minimize the distance to the head. The gate traces must be kept away from the power traces to avoid noise coupling. The thermal vias connect the MOSFETs' thermal pads to the inner ground planes, which act as heat sinks. |
Design Example: Layout in Sato Printers |
Sato's printer uses a 4-layer PCB with the following stack-up: top layer (components and traces), layer 2 (ground), layer 3 (power), and bottom layer (signals). The MOSFET bank is placed on the top layer, close to the printhead connector. The 24-volt power is routed on layer 3, with a width of 5 millimeters. The ground return is on layer 2, with a solid plane. The gate traces are on the bottom layer, away from the power traces. The manufacturer used a 2-ounce copper thickness to reduce the resistance. |
|
Chapter 21: The Stitch Vias - Connecting the Layers |
The thermal pads of the MOSFETs are connected to the ground planes through vias. These vias conduct heat from the pads to the inner planes, reducing the thermal resistance. The vias are typically placed in a grid under the thermal pad. The vias also conduct current, providing a low-impedance path from the pad to the ground plane. The number of vias is determined by the current and the heat. A typical design uses 9 to 16 vias per MOSFET. |
Design Example: Vias in Zebra's Design |
Zebra's discrete design uses 9 vias under each MOSFET's thermal pad. The vias are 0.3 millimeters in diameter, and they are placed in a 3x3 grid. The manufacturer measured the thermal resistance of the via array and found it to be 5C per watt - much lower than the 30C per watt of the package alone. The vias also helped to reduce the ground inductance. |
|
Chapter 22: The Copper Thickness - A Trade-Off with Cost |
The PCB copper thickness is specified in ounces per square foot. Standard PCBs use 1-ounce copper (35 micrometers thick). For high-current applications, 2-ounce copper (70 micrometers) is used. The thicker copper reduces the resistance and the voltage drop. However, it also increases the cost and makes the PCB more difficult to manufacture. The designer must choose the copper thickness based on the current and the allowed voltage drop. |
Design Example: Copper Thickness in Brother Printers |
Brother's printer uses 2-ounce copper for the 24-volt rail and the ground. The manufacturer calculated the voltage drop: the trace resistance is 0.005 ohms, and the current is 15 amperes, giving a drop of 0.075 volts - acceptable. The manufacturer considered 1-ounce copper, but the drop would have been 0.15 volts, which is still acceptable. They chose 2-ounce copper for its lower resistance and better heat spreading, even though it cost more. |
|
Chapter 23: The Kelvin Connection - Avoiding Sense Errors |
In some designs, the current is sensed using a resistor. The sense resistor is placed in series with the MOSFET, and the voltage across it is measured. The measurement is done using a Kelvin connection - a pair of separate sense traces that carry no current. The Kelvin connection eliminates the error caused by the resistance of the PCB traces. The Kelvin connection is used in high-accuracy current sensing applications. |
Design Example: Kelvin Connection in Zebra's Design |
Zebra's discrete design uses a Kelvin connection for the current sense resistor. The sense resistor is a 0.1-ohm, 1% resistor. The main current flows through the resistor, and a separate pair of traces connects the resistor's terminals to the comparator. The Kelvin traces are routed as a differential pair, away from the power traces. The manufacturer achieved a current measurement accuracy of 1%. |
|
Chapter 24: The Parasitic Inductance - A Hidden Enemy |
The parasitic inductance of the traces and the cabling can cause voltage spikes and ringing. The inductance is proportional to the loop area - the area between the outgoing and return currents. To minimize the inductance, the outgoing and return traces must be close together. This is achieved by using a solid ground plane under the power traces, or by routing the power and ground traces as a twisted pair. The parasitic inductance can also be reduced by using multiple parallel traces. |
Design Example: Inductance Reduction in Sato Printers |
Sato's printer uses a solid ground plane to reduce the loop inductance. The power trace on the top layer and the ground plane on the bottom layer create a small loop area. The manufacturer measured the inductance and found it to be 10 nanohenries - much lower than the 100 nanohenries of a discrete wire. The low inductance reduced the voltage spikes to a safe level. |
|
Chapter 25: The Parallel Bus - Power Distribution to Multiple ICs |
In a printer with multiple driver ICs, the 24-volt power must be distributed to all ICs. The distribution is done with a parallel bus - a wide trace that runs along the row of ICs. The bus must be capable of carrying the total current. The bus is also used to distribute the clock, data, and latch signals. The bus is designed as a star network or a daisy chain, depending on the skew requirements. |
Design Example: Parallel Bus in a Rohm Design |
Rohm's reference design uses a parallel bus for the 24-volt power. The bus is a 5-millimeter-wide trace on the top layer. The 70 BH12 ICs are placed along the bus. Each IC has a decoupling capacitor, and the bus is connected to the ICs' VCC pins. The manufacturer measured the voltage at the farthest IC and found it to be 23.8 volts - a droop of 0.2 volts from the nominal 24 volts. The droop was acceptable. |
|
Chapter 26: The Decoupling of the MOSFET Bank - A Local Reservoir |
The MOSFET bank draws current in bursts. The decoupling capacitors provide a local reservoir of charge, reducing the voltage droop. The decoupling capacitors are placed near the MOSFET bank. The capacitance is typically 10 to 100 microfarads. The capacitors are ceramic and electrolytic, with the ceramic capacitors providing the high-frequency response and the electrolytic capacitors providing the bulk energy. The decoupling capacitors are connected to the power and ground planes with short, wide traces. |
Design Example: Decoupling in Zebra's Design |
Zebra's discrete design uses a 100-microfarad electrolytic capacitor and a 10-microfarad ceramic capacitor for the decoupling of the MOSFET bank. The capacitors are placed at the input of the MOSFET bank, close to the power distribution bus. The manufacturer measured the voltage droop during a full-black strobe and found it to be 0.5 volts - reduced from 1.5 volts without the capacitors. The decoupling capacitors also reduced the EMI. |
|
Chapter 27: The Soft-Start of the MOSFET Bank - Limiting the Inrush |
When the printer is powered on, the output capacitors are discharged. If the MOSFETs are turned on immediately, the inrush current can be high. A soft-start circuit is used to limit the inrush. The soft-start circuit gradually increases the duty cycle of the strobe, or it gradually turns on the high-side switch. The soft-start is implemented in the firmware or in a dedicated circuit. |
Design Example: Soft-Start in Brother Printers |
Brother's printer implements the soft-start in the firmware. The CPU gradually increases the strobe width from 0 to the nominal value over a period of 10 milliseconds. The manufacturer measured the inrush current and found it to be 2 amperes - much lower than the 15-ampere peak. The soft-start also prevented the bulk capacitor from being stressed. |
|
Chapter 28: The ESD Protection for the MOSFETs - A Robust Design |
The MOSFETs are connected to the printhead, which is exposed to the user. The user can touch the printhead, which can cause an ESD event. The ESD event can damage the MOSFETs. The ESD protection for the MOSFETs is provided by the TVS diodes on the FFC. The TVS diodes clamp the voltage to a safe level. The ESD protection is also provided by the MOSFET's own body diode, which can conduct the ESD current if the voltage exceeds the breakdown voltage. |
Design Example: ESD Protection in Sato Printers |
Sato's printer uses TVS diodes on the FFC for ESD protection. The TVS diodes are placed on the main board, near the FFC connector. The manufacturer tested the ESD protection by applying a 15-kilovolt air discharge to the printhead connector. The printer survived without any damage. The manufacturer also added a small capacitor (100 picofarads) in parallel with the TVS diode to absorb the high-frequency component of the ESD pulse. |
|
Chapter 29: The Aging of the MOSFETs - A Gradual Degradation |
The MOSFETs degrade over time. The degradation is caused by the thermal stress and the voltage stress. The on-resistance increases gradually, and the gate threshold voltage changes. The degradation is accelerated by high temperatures and high currents. The degradation is not usually a problem for the printer's lifetime (100,000 labels), because the MOSFETs are operated well below their ratings. However, the printer's firmware can monitor the on-resistance by measuring the voltage drop during a test pulse, and it can compensate for the degradation by adjusting the strobe width. |
Design Example: Aging Monitoring in Zebra Printers |
Zebra's printer monitors the on-resistance of the MOSFET bank by measuring the voltage drop during a test pulse. The test pulse is applied to all dots at a low duty cycle. The voltage drop is measured by an ADC. If the on-resistance increases by more than 20%, the printer displays a warning. The manufacturer reports that the on-resistance typically increases by 5% after 100,000 labels, which is well within the tolerance. |
|
Chapter 30: The Shoot-Through - A Dangerous Condition |
Shoot-through is a condition where the high-side MOSFET and the low-side MOSFET (or the driver's internal low-side transistor) are turned on simultaneously, creating a short circuit from the power supply to the ground. Shoot-through can occur if the gate drive signals overlap. In a thermal printhead, there is no high-side MOSFET - the high-side switch is a separate device. However, the driver ICs have internal pull-down transistors that can short the output to the ground. The shoot-through is prevented by the dead-time circuitry in the driver IC, which ensures that the high-side and low-side are never on at the same time. |
Design Example: Shoot-Through in Rohm BH12 |
The BH12 driver IC has a built-in dead-time circuitry. The manufacturer did not need to add any external protection. The dead-time is 100 nanoseconds, which is sufficient to prevent shoot-through. |
|
Chapter 31: The Miller Effect - A Gate Drive Challenge |
The Miller effect is a phenomenon where the gate-drain capacitance of a MOSFET is amplified by the voltage gain. The Miller effect causes a 'plateau' in the gate voltage during switching, which increases the switching time. The Miller effect is more pronounced at higher drain voltages. The Miller effect can be mitigated by using a gate driver with a high current capability, or by using a MOSFET with a low gate-drain capacitance. |
Design Example: Miller Effect in Zebra's Design |
Zebra's discrete design uses a gate driver with a peak current of 1 ampere, which overcomes the Miller effect. The manufacturer also chose MOSFETs with a low gate-drain capacitance of 10 picofarads. The manufacturer measured the gate voltage and found that the plateau lasted only 10 nanoseconds - which was acceptable. |
|
Chapter 32: The Gate Charge - A Parameter for Speed |
The gate charge is the total charge required to turn the MOSFET on. The gate charge is the product of the gate capacitance and the gate voltage. A lower gate charge means a faster switching speed and lower switching losses. The gate charge is specified in the datasheet. For a typical printer MOSFET, the gate charge is 10 to 20 nanocoulombs. The gate drive current is the gate charge divided by the desired switching time. |
Design Example: Gate Charge in Brother Printers |
Brother's printer uses MOSFETs with a gate charge of 15 nanocoulombs. The gate drive current is 15 nanoCoulombs / 15 nanoseconds = 1 ampere. The integrated driver IC provides this current. The manufacturer selected the MOSFETs based on the gate charge to ensure fast switching. |
|
Chapter 33: The Threshold Voltage - The Turn-On Point |
The threshold voltage (Vth) is the gate voltage at which the MOSFET begins to conduct. For logic-level MOSFETs, Vth is typically 1 to 2.5 volts. For standard MOSFETs, Vth is 2 to 4 volts. The threshold voltage determines the gate drive voltage required. A higher threshold voltage requires a higher gate drive, but it also provides better noise immunity. The threshold voltage is temperature-dependent - it decreases at higher temperatures. |
Design Example: Threshold in Sato Printers |
Sato's printer uses logic-level MOSFETs with a threshold voltage of 1.5 volts. The gate drive is 5 volts from the shift register, which is more than sufficient. The manufacturer did not need to use a gate driver. |
|
Chapter 34: The Safe Operating Area - The Limits of the MOSFET |
The Safe Operating Area (SOA) is a graph in the datasheet that shows the maximum allowed current and voltage for different pulse durations. The SOA is limited by the maximum power dissipation, the maximum current, and the maximum voltage. The designer must ensure that the MOSFET is operated within the SOA for all conditions. For a printer, the MOSFET is operated in the SOA with a large margin. |
Design Example: SOA in Zebra's Design |
Zebra's MOSFETs have an SOA that allows 15 amperes at 24 volts for 1 millisecond. The manufacturer calculated the worst-case condition: a short circuit that causes 15 amperes for 1 millisecond before the protection trips. This is within the SOA. The manufacturer also ensured that the MOSFET is never operated at the SOA limits in normal conditions. |
|
Chapter 35: The Paralleling of Driver ICs - When One Is Not Enough |
If the printhead has more dots than a single driver IC can handle, multiple driver ICs are paralleled. The driver ICs are cascaded in a daisy chain for the data, but the outputs are separate. The driver ICs have separate power supplies and separate strobe controls. The parallel driver ICs must have synchronized strobes - all strobes must fire at the same time. This is achieved by using a common strobe signal. |
Design Example: Paralleling in Sato Printers |
Sato's printer uses 70 BH12 driver ICs in parallel. The strobe signal is common to all ICs. The clock and data are cascaded. The manufacturer measured the strobe skew between the ICs and found it to be 2 nanoseconds - negligible. The paralleling was successful. |
|
Chapter 36: The Soft-Off - A Controlled Shutdown |
When the printer is powered down, the MOSFETs must be turned off in a controlled manner to prevent voltage spikes. The soft-off circuit gradually reduces the gate voltage, turning off the MOSFET slowly. The soft-off is implemented in the firmware or in a dedicated circuit. The soft-off is more important for inductive loads, but it is also used in some printer designs. |
Design Example: Soft-Off in Brother Printers |
Brother's printer implements the soft-off in the firmware. The CPU reduces the strobe width to zero over a period of 5 milliseconds. The manufacturer measured the voltage spikes and found them to be negligible. |
|
Chapter 37: The Reliability - A Statistical Analysis |
The MOSFET bank is a critical component, and its reliability is important. The reliability is specified in terms of the failure rate per million hours. For a MOSFET, the failure rate is typically 0.1 failures per million hours. The failure rate increases with temperature - it doubles for every 10C increase. The thermal design is therefore critical for reliability. The manufacturer uses a thermal derating to ensure that the junction temperature is kept below 80C, which gives a low failure rate. |
Design Example: Reliability in Zebra Printers |
Zebra's printer uses a thermal derating that keeps the junction temperature below 70C. The manufacturer calculated the failure rate and found it to be 0.01 failures per million hours - which is excellent. The printer is expected to have a mean time between failures of over 1 million hours. |
|
Chapter 38: The Future of MOSFET Banks - GaN and SiC |
The future of MOSFET banks lies in wide-bandgap semiconductors like Gallium Nitride (GaN) and Silicon Carbide (SiC). These materials have a higher breakdown voltage, lower on-resistance, and faster switching speed than silicon. GaN and SiC MOSFETs can operate at higher frequencies and higher temperatures. The use of GaN and SiC in printers would allow higher print speeds, smaller PCBs, and higher efficiency. However, GaN and SiC are more expensive and require special gate drivers. |
Design Example: GaN in a Prototype Printer |
A prototype printer from a startup uses GaN MOSFETs from GaN Systems. The GaN MOSFETs have an on-resistance of 10 milliohms and a switching frequency of 1 megahertz. The printer achieves a print speed of 20 inches per second - twice the speed of a silicon-based printer. The manufacturer reports that the GaN MOSFETs run cool because of the low on-resistance and the high efficiency. The prototype is still in development, but it shows the potential of GaN technology. |
|
Detailed Summary - Tying It All Together |
We have now completed our comprehensive journey through the power MOSFET bank - the muscle that drives every dot on a thermal printhead. We began with the fundamental understanding of a single MOSFET as a voltage-controlled switch with three terminals: gate, drain, and source. We learned that the on-resistance is the critical parameter that determines the voltage drop and power dissipation, and that a low on-resistance is essential for efficient printing. |
We explored the two main approaches to implementing the MOSFET bank: integrated driver ICs, which combine logic and power in a single package, and discrete MOSFETs, which offer greater flexibility for high-voltage or high-current applications. We saw real-world examples from major manufacturers: Rohm's BH12 integrated driver, Infineon's OptiMOS discrete MOSFETs, and Texas Instruments' intelligent power switches. We examined the trade-offs between the two approaches - integrated drivers are cheaper and simpler, while discrete MOSFETs can handle more extreme conditions. |
We delved into the critical design parameters: the gate drive voltage and current, the switching speed, the thermal resistance, and the safe operating area. We saw how the gate resistor and the snubber circuit are used to manage ringing and voltage spikes. We discussed the importance of the layout - the wide traces, the solid ground plane, the decoupling capacitors, and the thermal vias that keep the MOSFETs cool and the voltage stable. |
We examined the protection features: over-current protection, short-circuit protection, thermal shutdown, and ESD protection. We saw how these features are implemented in integrated drivers and in discrete designs, and how they prevent catastrophic failures. We discussed the parasitic elements - the body diode, the gate-drain capacitance, and the parasitic inductance - and how they affect the performance. |
We looked at the practical aspects: the paralleling of MOSFETs and driver ICs, the current sharing, the soft-start and soft-off, and the aging of the MOSFETs. We saw how manufacturers like Zebra, Sato, Brother, and Honeywell implement these aspects in their printers. We discussed the reliability of the MOSFET bank, and how thermal derating and temperature control ensure a long lifetime. |
|
We also looked to the future with GaN and SiC MOSFETs, which promise higher speeds and higher efficiency. We saw a prototype printer that uses GaN to achieve twice the speed of a conventional printer. |
The overarching lesson is that the MOSFET bank is not a simple collection of transistors. It is a carefully designed subsystem that must handle high currents, high voltages, and high temperatures, all while switching with microsecond precision. The MOSFET bank must be protected from faults, cooled effectively, and laid out with meticulous attention to parasitic elements. A well-designed MOSFET bank is invisible - it just works, line after line, label after label. A poorly designed MOSFET bank will cause faded dots, uneven prints, or catastrophic failures. Understanding the MOSFET bank is essential for any engineer who wants to design a reliable thermal printer, and this chapter has provided that understanding, from the physics of the MOSFET to the practical implementation in a real-world product. |
End of Extended Section 6 |
cs@easiersoft.com