PCB layout basics
The intention of this article is to discuss common mistakes made by the developers of printed circuit boards, a description of the impact of these mistakes on quality indicators and recommendations for solving the problems encountered.
GENERAL CONSIDERATIONS
Due to the significant differences between analog circuitry and digital circuitry, the analog part of the circuit must be separated from the rest, and its wiring must follow special methods and rules. The effects arising from the imperfect characteristics of printed circuit boards become especially noticeable in high-frequency analog circuits, but the general form errors described in this article can affect the quality characteristics of devices operating even in the audio frequency range.
The circuit board is a component of the circuit.
Only in rare cases, the circuit board of an analog circuit can be diluted so that the effects it makes have no effect on the operation of the circuit. At the same time, any such impact can be minimized so that the characteristics of the analog circuit of the device are the same as those of the model and the prototype.
Layout
Developers of digital circuits can correct small errors on the manufactured board, complementing it with jumpers or, conversely, removing unnecessary conductors, making changes to the work of programmable chips, etc., moving very soon to the next development. For an analog circuit this is not the case. Some of the common mistakes discussed in this article cannot be corrected by adding jumpers or removing extra conductors. They can and will render the entire PCB inoperative.
It is very important for a digital designer who uses such remedies to read and understand the material outlined in this article well in advance of the project’s transfer to production. A little attention paid during development and a discussion of possible options will help not only prevent the PCB from becoming waste, but also reduce the cost due to blunders in a small analog part of the circuit. Finding bugs and fixing them can result in losses of hundreds of hours. Layout can reduce this time to one day or less. Make all your analog circuits.
Sources of noise and interference
Noise and interference are the main elements that limit the quality characteristics of the circuits. Interference can both be emitted by sources and also be induced by circuit elements. An analog circuit is often located on a printed circuit board along with high-speed digital components, including digital signal processors (DSP).
High-frequency logic signals create significant radio frequency interference (RFI). The number of sources of noise emission is enormous: key power sources for digital systems, mobile phones, radio and television, power sources for fluorescent lamps, personal computers, lightning, etc. Even if the analog circuit operates in the audio frequency range, radio frequency interference can create noticeable noise in the output signal.
PCB CATEGORIES
The choice of PCB design is an important factor in determining the mechanical characteristics when using the device as a whole. For the manufacture of printed circuit boards used materials of different levels of quality. The most suitable and convenient for the developer would be if the manufacturer of printed circuit boards is located nearby. In this case, it is easy to control the resistivity and the dielectric constant - the main parameters of the material of the printed circuit board. Unfortunately, this is not enough and it is often necessary to know other parameters, such as flammability, high temperature stability and hygroscopicity coefficient. These parameters can be known only by the manufacturer of components used in the manufacture of printed circuit boards.
Laminated materials are designated with FR (flame resistant, flame retardant) and G. Materials with indexes G10 and G11 have special characteristics. Materials printed circuit boards are given in Table. one.
Do not use the FR-1 category PCB. There are many examples of the use of FR-1 printed circuit boards, which are damaged by heat from powerful components. Printed circuit boards in this category are more like cardboard.
FR-4 is often used in the manufacture of industrial equipment, while FR-2 is used in the manufacture of household appliances. These two categories are industry standardized, and the FR-2 and FR-4 printed circuit boards are often suitable for most applications. But sometimes the imperfection of the characteristics of these categories leads to the use of other materials. For example, for very high frequency applications, fluoroplastic and even ceramics are used as printed circuit board material. However, the more exotic the material of the PCB, the higher the price can be.
When choosing a printed circuit board material, pay special attention to its hygroscopicity, since this parameter can have a strong negative effect on the desired characteristics of the board - surface resistance, leakage, high-voltage insulation properties (breakdown and sparking) and mechanical strength. Also pay attention to the operating temperature. High temperature areas may occur in unexpected places, such as near large digital integrated circuits that switch at high frequency. If such areas are located directly below the analog components, the temperature increase may affect the characteristics of the analog circuit.
Table 1
Category
|
Components, comments
|
FR-1
|
paper, phenolic composition: pressing and stamping at room temperature, high hygroscopicity coefficient
|
FR-2
|
paper, phenolic composition: applicable for one-sided printed circuit boards of household appliances, low coefficient of hygroscopicity
|
FR-3
|
paper, epoxy composition: developments with good mechanical and electrical characteristics
|
FR-4
|
fiberglass, epoxy composition: excellent mechanical and electrical properties
|
FR-5
|
fiberglass, epoxy composition: high strength at elevated temperatures, no ignition
|
G10
|
fiberglass, epoxy composition: high insulating properties, the highest strength of fiberglass, low hygroscopicity
|
G11
|
fiberglass, epoxy composition: high bending strength at elevated temperatures, high solvent resistance
|
After the material of the printed circuit board is selected, it is necessary to determine the thickness of the foil of the printed circuit board. This parameter is primarily selected based on the maximum amount of current flowing. If possible, try to avoid using very thin foil.
NUMBER OF PCB LAYERS
Depending on the total complexity of the scheme and quality requirements, the developer must determine the number of layers of the printed circuit board.
Single Layer PCB
Very simple electronic circuits are carried out on one-sided boards using low-cost foil materials (FR-1 or FR-2) and often have many jumpers, resembling double-sided boards. This method of creating printed circuit boards is recommended only for low-frequency circuits. For reasons that will be described below, one-sided printed circuit boards are highly susceptible to interference. A good one-sided PCB is difficult to develop due to many reasons. Nevertheless, good boards of this type are found, but when developing them, a lot of things need to be thought out in advance.
Double Layer PCB
At the next level, there are two-sided printed circuit boards, which in most cases use FR-4 as the substrate material, although FR-2 is sometimes found. The use of FR-4 is more preferable, since the holes in the printed circuit boards from this material are of better quality. Schemes on double-sided printed circuit boards are divorced much easier, because in two layers it is easier to carry out the layout of intersecting tracks. However, for analog circuits, the intersection of routes is not recommended. Where possible, the bottom layer (bottom) should be taken under the landfill, and the rest signals should be spread in the top layer (top). Using the landfill as a ground tire provides several advantages:
- common wire is the most frequently connected wire in the circuit; therefore, it is reasonable to have a lot of common wire to simplify wiring.
- increases the mechanical strength of the board.
- resistance of all connections to the common wire decreases, which, in turn, reduces noise and pickup.
- the distributed capacitance for each circuit of the circuit increases, helping to suppress the radiated noise.
- the polygon, which is a screen, suppresses interference emitted by sources located on the side of the polygon.
Despite their advantages, double-sided printed circuit boards are not the best, especially for low-signal or high-speed circuits. In general, the thickness of the PCB, i.e. the distance between the metallization layers is 1.5 mm, which is too much to fully realize some of the advantages of the two-layer printed circuit board listed above. Distributed capacity, for example, is too small due to such a large interval.
Multilayer printed circuit boards
For responsible circuit design requires multilayer printed circuit boards (WFP). Some reasons for their use are obvious:
- the same convenient as for the common wire bus, the power bus wiring; if polygons on a separate layer are used as power tires, it is quite simple to make a power supply to each circuit element using vias;
- the signal layers are released from the power rails, which facilitates the wiring of the signal conductors;
- Distributed capacitance appears between the polygons of the earth and the power supply, which reduces high-frequency noise.
In addition to these reasons for the use of multilayer printed circuit boards, there are other, less obvious:
better suppression of electromagnetic (EMI) and radio frequency (RFI) interference due to the image plane effect, known in Marconi times. When a conductor is placed close to a flat conductive surface, most of the return high-frequency currents will flow along the plane directly below the conductor. The direction of these currents will be opposite to the direction of the currents in the conductor. Thus, the reflection of the conductor in the plane creates a signal transmission line. Since the currents in the conductor and in the plane are equal in magnitude and opposite in direction, some reduction of radiated noise is created. The reflection effect effectively works only with inseparable solid polygons (they can be both polygons of the earth, and polygons of food). Any breach of integrity will result in reduced interference suppression.
reduced total cost for small-scale production. Despite the fact that the manufacture of multilayer printed circuit boards is more expensive, their possible radiation is less than that of single and double layer boards. Therefore, in some cases, the use of only multilayer boards will allow to fulfill the radiation requirements set during development and not to carry out additional tests and tests. The use of WFPs can reduce the radiated noise level by 20 dB compared to dual layer boards.
reduced total cost for small-scale production. Despite the fact that the manufacture of multilayer printed circuit boards is more expensive, their possible radiation is less than that of single and double layer boards. Therefore, in some cases, the use of only multilayer boards will allow to fulfill the radiation requirements set during development and not to carry out additional tests and tests. The use of WFPs can reduce the radiated noise level by 20 dB compared to dual layer boards.
The order of the layers
Inexperienced developers often have some confusion about the optimal order of the layers of the PCB. Take for example a 4-layer chamber containing two signal layers and two polygon layers - the earth layer and the power layer. What is the best layer order? Signal layers between polygons that will serve as screens? Or make the polygon layers internal to reduce the interference of the signal layers?
When addressing this issue, it is important to remember that often the arrangement of the layers does not matter much, since all the same the components are located on the outer layers, and the tires bringing signals to their conclusions sometimes pass through all the layers. Therefore, any screen effects are only a compromise. In this case, it is better to take care of creating a large distributed capacity between the food and land polygons, placing them in the inner layers.
Another advantage of the location of the signal layers outside is the availability of signals for testing, as well as the ability to modify connections. Anyone who has ever changed connections of conductors located in the inner layers will appreciate this possibility.
For printed circuit boards with more than four layers, there is a general rule to place high-speed signal conductors between the ground and power polygons, and low-frequency conductors for outer layers.
GROUNDING
Good grounding is a common requirement for a rich, multi-layered system. And it should be planned from the first step of the design development.
The main rule: the division of land.
The division of land into analog and digital parts is one of the simplest and most effective methods of noise suppression. One or more layers of a multi-layer printed circuit board is usually retracted under the ground layer. If the developer is not very experienced or inattentive, then the land of the analog part will be directly connected to these polygons, i.e. analog return current will use the same circuit as digital return current. Auto-distributors work approximately as well and unite all the lands together.
If a previously developed printed circuit board with a single earth ground uniting analog and digital lands is processed, then you must first physically divide the lands on the board (after this operation, the board becomes almost impossible) After that, all connections to the analog earth ground of the analog circuit components are made (analog ground is formed) and to the digital earth ground of the digital circuit components (digital ground is formed). And only after this, the source combines digital and analog earth.
Other land formation rules:
Power and ground must be at the same potential for alternating current, which implies the use of decoupling capacitors and distributed capacitance.
Avoid overlapping analog and digital polygons. Locate buses and analog power polygons over an analog ground landfill (similarly for digital power buses). If in any place there is an overlap of the analog and digital polygons, the distributed capacitance between the overlapping sections will create an AC coupling, and pickups from the operation of the digital components will fall into an analog circuit. Such overlaps cancel isolation of polygons.
Separation does not mean electrical isolation of analog from digital ground. They should be connected together in some, preferably one, low-impedance node. The correct, from the point of view of the ground, the system has only one earth, which is the grounding terminal for systems with mains-powered voltage or a common terminal for systems with DC-voltage (for example, battery). All signal and supply currents in this circuit must return to this earth at one point, which will serve as the system earth. Such a point may be the output of the device. It is important to understand that when connecting the general output of the circuit to several points of the body, earth loops can form. Creating a single common point of land unification is one of the most difficult aspects of system design.
If possible, separate the pins of the connectors intended for transferring the return currents - the return currents should be combined only at the point of the system ground. Aging of the contacts of the connectors, as well as frequent uncoupling of their counterparts, leads to an increase in the resistance of the contacts, therefore, for more reliable operation, it is necessary to use connectors with a certain number of additional leads. Complicated digital circuit boards have many layers and contain hundreds or thousands of conductors. Adding another conductor rarely creates a problem in contrast to the additional connector pins added. If this cannot be done, then it is necessary to create two return current conductors for each power circuit on the board, taking special precautions.
It is important to separate digital signal buses from the places on the printed circuit board where the analog circuit components are located. This implies isolation (shielding) by polygons, creation of short paths of analog signals and careful placement of passive components in the presence of adjacent high-speed digital and responsible analog bus signals. Tires of digital signals must be bred around areas with analog components and not overlap with tires and polygons of analog ground and analog power. If this is not done, the design will contain a new unintended element - an antenna, the radiation of which will affect the high-impedance analog components and conductors.
Avoid overlapping analog and digital polygons. Locate buses and analog power polygons over an analog ground landfill (similarly for digital power buses). If in any place there is an overlap of the analog and digital polygons, the distributed capacitance between the overlapping sections will create an AC coupling, and pickups from the operation of the digital components will fall into an analog circuit. Such overlaps cancel isolation of polygons.
Separation does not mean electrical isolation of analog from digital ground. They should be connected together in some, preferably one, low-impedance node. The correct, from the point of view of the ground, the system has only one earth, which is the grounding terminal for systems with mains-powered voltage or a common terminal for systems with DC-voltage (for example, battery). All signal and supply currents in this circuit must return to this earth at one point, which will serve as the system earth. Such a point may be the output of the device. It is important to understand that when connecting the general output of the circuit to several points of the body, earth loops can form. Creating a single common point of land unification is one of the most difficult aspects of system design.
If possible, separate the pins of the connectors intended for transferring the return currents - the return currents should be combined only at the point of the system ground. Aging of the contacts of the connectors, as well as frequent uncoupling of their counterparts, leads to an increase in the resistance of the contacts, therefore, for more reliable operation, it is necessary to use connectors with a certain number of additional leads. Complicated digital circuit boards have many layers and contain hundreds or thousands of conductors. Adding another conductor rarely creates a problem in contrast to the additional connector pins added. If this cannot be done, then it is necessary to create two return current conductors for each power circuit on the board, taking special precautions.
It is important to separate digital signal buses from the places on the printed circuit board where the analog circuit components are located. This implies isolation (shielding) by polygons, creation of short paths of analog signals and careful placement of passive components in the presence of adjacent high-speed digital and responsible analog bus signals. Tires of digital signals must be bred around areas with analog components and not overlap with tires and polygons of analog ground and analog power. If this is not done, the design will contain a new unintended element - an antenna, the radiation of which will affect the high-impedance analog components and conductors.
Almost all clock signals are fairly high-frequency signals, so even small capacitances between tracks and polygons can create significant connections. It must be remembered that not only the main clock frequency can cause a problem, but also its higher harmonics.
A good concept is to place the analog part of the circuit close to the input / output connections of the board. Developers of digital printed circuit boards using powerful integrated circuits often tend to spread tires 1 mm wide and several centimeters long to connect analog components, believing that low track resistance will help get rid of interference. What turns out at the same time is an extended film capacitor, which will be induced by spurious signals from digital components, digital ground and digital power, exacerbating the problem.
There is only one case when it is necessary to combine analog and digital signals over the area of the analog ground polygon. Analog-digital and digital-analog converters are located in housings with analog and digital ground leads. Taking into account the previous considerations, it can be assumed that the digital ground output and the analog ground output must be connected to the digital and analog ground buses, respectively. However, in this case it is not true.
The names of the outputs (analog or digital) refer only to the internal structure of the converter, to its internal connections. In the circuit, these pins must be connected to the analog ground bus. The connection can be made inside the integrated circuit, however, to obtain a low resistance of such a connection is rather difficult due to topological limitations. Therefore, when using converters, an external connection of analog and digital ground leads is assumed. If this is not done, the chip parameters will be much worse than those given in the specification.
It is necessary to take into account that the digital elements of the converter can degrade the quality characteristics of the circuit, introducing digital interference into the circuits of analog ground and analog power. When developing converters, this negative impact is taken into account so that the digital part consumes as little power as possible. In this case, the interference from switching logic elements is reduced. If the digital outputs of the converter are not heavily loaded, then the internal switching usually does not cause any special problems. When designing a printed circuit board containing an ADC or DAC, it is necessary to properly consider the decoupling of the digital power supply of the converter to the analog ground.
FREQUENCY CHARACTERISTICS OF PASSIVE COMPONENTS
For correct operation of analog circuits, the correct choice of passive components is very important. Begin design development with careful consideration of the high-frequency characteristics of the passive components and the preliminary placement and layout of them on the board sketch.
A large number of developers completely ignore the frequency limitations of passive components when used in analog circuitry. These components have limited frequency ranges and their operation outside the specified frequency domain can lead to unpredictable results. Someone might think that this discussion concerns only high-speed analog circuits. However, this is far from the case - high-frequency signals have a rather strong effect on the passive components of low-frequency circuits by means of radiation or direct connection via conductors. For example, a simple low-pass filter on an operational amplifier can easily turn into a high-pass filter when it is exposed to a high frequency input.
Resistors
Three types of resistors are commonly used: 1) wire, 2) carbon composite, and 3) film. You don't have to have a lot of imagination to understand how a wirewound resistor can turn into an inductance, since it is a coil with a high-resistance metal wire. Most developers of electronic devices do not have a clue about the internal structure of film resistors, which also represent a coil, however, from a metal film. Therefore, film resistors also have an inductance that is smaller than that of wire resistors. Film resistors with a resistance of less than 2 kΩ can be freely used in high-frequency circuits. The terminals of the resistors are parallel to each other, so there is a noticeable capacitive coupling between them.
Capacitors
The high-frequency characteristics of the capacitors can be represented by an equivalent circuit shown in Figure 6.
Capacitors in analog circuits are used as decoupling elements and filtering components.
A 10 μF electrolytic capacitor has a resistance of 1.6 Ohms at a frequency of 10 kHz and 160 μΩ at a frequency of 100 MHz. Is it so?
In fact, no one has ever seen an electrolytic capacitor with a reactive resistance of 160 µOhm. The plates of film and electrolytic capacitors are twisted foil layers that create a parasitic inductance. The effect of self-inductance in ceramic capacitors is much smaller, which allows them to be used when working at high frequencies. In addition, capacitors have a leakage current between the plates, which is equivalent to a resistor connected in parallel to their terminals, adding its parasitic effect to the effects of a series-connected resistance of the terminals and plates. In addition, the electrolyte is not an ideal conductor. All these resistances add up to create equivalent series resistance (ESR). Capacitors used as decoupling should have a low ESR, since series resistance limits the effectiveness of ripple and noise suppression. An increase in operating temperature rather significantly increases the equivalent series resistance and may lead to a deterioration in the capacitor characteristics. Therefore, if it is intended to use an aluminum electrolytic capacitor at an elevated operating temperature, then it is necessary to use capacitors of the appropriate type (105 ° C).
Capacitor leads also contribute to an increase in parasitic inductance. For small capacitance values, it is important to keep the length of the leads short. The combination of parasitic inductance and capacitance can create a resonant circuit. Assuming that the pins have an inductance of the order of 8 nH per one centimeter of length, a capacitor with a capacity of 0.01 μF with pins one centimeter long will have a resonant frequency of about 12.5 MHz. This effect is known to engineers who developed electronic vacuum devices decades ago. The one who restores antique radios and does not know about this effect faces many problems.
When using electrolytic capacitors it is necessary to monitor the correct connection. A positive lead must be connected to a more positive constant potential. Incorrect connection leads to the flow through the electrolytic DC capacitor, which can damage not only the capacitor itself, but also part of the circuit.
In rare cases, the potential difference in DC between two points in the circuit may change its sign. This requires the use of non-polar electrolytic capacitors, the internal structure of which is equivalent to two polar capacitors connected in series.
Inductance
For example, an inductance of 10 mH will have a reactance of 628 Ohms at a frequency of 10 kHz, and at a frequency of 100 MHz - a resistance of 6.28 MΩ.
In fact, there is no inductance with a reactance of 6.28 MΩ. The nature of the parasitic resistance is easy to understand - the turns of the coil are made of wire with some resistance per unit length. The parasitic capacitance is perceived more difficult until it is taken into account that the next turn of the coil is located close to the previous one, and a capacitive coupling occurs between closely spaced conductors. Parasitic capacitance limits the upper operating frequency. Small wire inductances begin to become ineffective in the range of 10 ... 100 MHz.
In fact, there is no inductance with a reactance of 6.28 MΩ. The nature of the parasitic resistance is easy to understand - the turns of the coil are made of wire with some resistance per unit length. The parasitic capacitance is perceived more difficult until it is taken into account that the next turn of the coil is located close to the previous one, and a capacitive coupling occurs between closely spaced conductors. Parasitic capacitance limits the upper operating frequency. Small wire inductances begin to become ineffective in the range of 10 ... 100 MHz.
Printed circuit board
The printed circuit board itself has the characteristics of the passive components discussed above, but not so obvious.
The pattern of conductors on a printed circuit board can be both a source and receiver of interference. Good wiring of conductors reduces the sensitivity of the analog circuit to the radiation sources.
A printed circuit board is susceptible to radiation, since the conductors and the terminals of the components form peculiar antennas. The theory of antennas is a fairly complex subject to study and is not covered in this article. However, some basics are provided here.
A bit of antenna theory
One of the main types of antennas is a pin or straight conductor. This antenna works because the direct conductor has a parasitic inductance and therefore can concentrate and pick up radiation from external sources. The full impedance of a direct conductor has resistive (active) and inductive (reactive) components:
At a direct current or low frequencies the active component prevails. As the frequency rises, the reactive component becomes more and more significant. In the range from 1 kHz to 10 kHz, the inductive component begins to influence, and the conductor is no longer a low-resistance connector, but rather acts as an inductor.
Typically, tracks on a printed circuit board have values from 6 nH to 12 nH per centimeter length. For example, a 10 cm conductor has a resistance of 57 mΩ and an inductance of 8 nHH per cm. At a frequency of 100 kHz, the reactance becomes 50 mΩ, and at higher frequencies the conductor will be inductance rather than resistance.
The whip antenna rule says that it begins to interact noticeably with the field at its length of about 1/20 of the wavelength, and the maximum interaction occurs at the length of the pin equal to 1/4 of the wavelength. Therefore, the 10 cm wire from the example in the previous paragraph will start to become a pretty good antenna at frequencies above 150 MHz. It must be remembered that despite the fact that the clock frequency generator of the digital circuit may not work at a frequency above 150 MHz, there are always higher harmonics in its signal. If the PCB contains components with pin pins of considerable length, such pins can also serve as antennas.
Another main type of antenna is loopback antenna. The inductance of a straight conductor increases greatly when it bends and becomes part of the arc. Increasing inductance lowers the frequency at which the antenna begins to interact with the field lines.
Experienced PCB designers, who are well versed in the theory of loop antennas, know that you cannot create loops for critical signals. Some developers, however, do not think about it, and the conductors of the return and signal current in their circuits are loops.
The theory of reflection and matching of signals is close to the theory of antennas.
When the PCB conductor rotates through a 90 ° angle, a signal reflection may occur. This is mainly due to a change in the width of the current path. At the top of the corner, the width of the path increases 1.414 times, which leads to a mismatch of the characteristics of the transmission line, especially the distributed capacitance and the own inductance of the path. Quite often it is necessary to rotate the circuit by 90 ° on the printed circuit board. Many modern CAD-packages allow you to smooth the corners of the tracks or to conduct the tracks in the form of an arc. Figure 9 shows two steps to improve the shape of the angle. Only the last example maintains a constant width of the path and minimizes reflections.
Tip for experienced PCB distributors: leave the smoothing procedure to the last stage of work before creating droplet-shaped leads and filling polygons. Otherwise, the CAD package will take longer to smooth because of more complex calculations.
PARASITIC EFFECTS OF THE PCB
Capacitive coupling occurs between the conductors of a printed circuit board located on different layers when they intersect. Sometimes this can create a problem. Conductors placed on top of each other on adjacent layers create a long film capacitor.
For example, a printed circuit board may have the following parameters:
- 4 layers; Signal and ground polygon layer - adjacent,
- interlayer spacing - 0.2 mm,
- conductor width - 0.75 mm,
- conductor length - 7.5 mm.
- 4 layers; Signal and ground polygon layer - adjacent,
- interlayer spacing - 0.2 mm,
- conductor width - 0.75 mm,
- conductor length - 7.5 mm.
The typical value of the dielectric constant ER for FR-4 is 4.5.
The capacitance value between the two tires is 1.1 pF. Even such a seemingly small capacity is unacceptable for some applications.
There is a doubling of the amplitude of the output signal at frequencies close to the upper limit of the frequency range of the OS. This, in turn, can lead to generation, especially at the operating frequencies of the antenna (above 180 MHz).
This effect gives rise to numerous problems, for which, however, there are many ways. The most obvious of these is a reduction in the length of the conductors. Another way is to reduce their width. There is no reason to use a conductor of this width for connecting the signal to the inverting input, since a very small current flows through this conductor. Reducing the length of the track to 2.5 mm, and width to 0.2 mm will lead to a decrease in capacitance to 0.1 pF, and such capacity will not lead to such a significant increase in frequency response. Another solution is to remove a part of the polygon under the inverting input and the conductor suitable for it.
The inverting input of an operational amplifier, especially a high-speed one, is largely prone to generation in high gain circuits. This is due to the unwanted capacitance of the input stage OU. Therefore, it is extremely important to reduce the parasitic capacitance and position the feedback components as close to the inverting input as possible. If, despite the measures taken, the amplifier is excited, it is necessary to proportionally reduce the resistance of the feedback resistors to change the resonant frequency of the circuit. An increase in resistors can also help, though much less frequently, since the excitation effect depends on the circuit impedance. When changing the feedback resistors, we should not forget about the change in the capacitance of the correction capacitor. Also, do not forget that
The width of the conductors of the printed circuit board can not be infinitely reduced. The maximum width is determined by both the process and the thickness of the foil. If two conductors pass close to each other, then a capacitive and inductive coupling is formed between them.
The dependencies describing these parasitic effects are complex enough to result in this article, but they can be found in the literature on transmission lines and strip lines.
Signal conductors should not be run parallel to each other, except when wiring differential or microstrip lines. The gap between the conductors must be at least three times the width of the conductors.
The capacitance between the tracks in analog circuits can create difficulties with large resistances of resistors (several MOhm). The relatively large capacitive coupling between the inverting and non-inverting inputs of the operational amplifier can easily lead to self-excitation of the circuit.
Whenever, when wiring a printed circuit board, there is a need to create a via, i.e. interlayer connection, it must be remembered that this also causes parasitic inductance. When the diameter of the hole after metallization d and the length of the channel h inductance can be calculated by the following approximate formula:
For example, when d = 0.4 mm and h = 1.5 mm (fairly common values), the inductance of the hole is 1.1 nH.
Keep in mind that the inductance of the hole together with the same parasitic capacitance form a resonant circuit, which can affect when working at high frequencies. The inductance of the hole is rather small, and the resonant frequency is somewhere in the gigahertz range, but if the signal is forced to pass through several vias during its path, then their inductances are added (series connection), and the resonant frequency decreases. Conclusion: try to avoid a large number of vias when wiring critical high-frequency conductors of analog circuits. Another negative phenomenon: with a large number of vias in the landfill, loopbacks can be created. Best analog wiring - all signal wires are located on the same PCB layer.
In addition to the parasitic effects discussed above, there are also those that are related to the insufficiently clean surface of the board.
Remember that if there are large resistances in the circuit, then special attention should be paid to cleaning the board. At the final operations of the PCB manufacturing, flux and dirt residues should be removed. Recently, when installing printed circuit boards, water-soluble fluxes are often used. Being less harmful, they are easily removed with water. But at the same time, washing the board with insufficiently clean water can lead to additional contaminations that impair the dielectric characteristics. Therefore, it is very important to wash the PCB with a high-impedance circuit with fresh distilled water.
SIGNAL DISCHARGE
As already noted, interference can penetrate into the analog part of the circuit through the power supply. To reduce such interference, decoupling (blocking) capacitors are used, reducing the local impedance of the power supply bus.
If it is necessary to dissolve a printed circuit board, on which there are both analog and digital parts, then you need to have at least a small understanding of the electrical characteristics of logic elements.
A typical output stage of a logic element contains two transistors connected in series with each other, as well as between the power and ground circuits.
In the ideal case, these transistors operate strictly in antiphase, i.e. when one of them is open, at the same time the second is closed, forming at the output either a signal of a logical unit or a logical zero. In steady state, the power consumption of the logic element is small.
The situation changes dramatically when the output stage switches from one logical state to another. In this case, for a short period of time, both transistors can be opened simultaneously, and the power supply current of the output stage increases greatly, since the resistance of the current path section from the power bus to the earth bus through two series-connected transistors decreases. The power consumption increases abruptly and then also decreases, which leads to a local change in the supply voltage and the appearance of a sharp, short-term change in current. Such changes in current lead to the emission of radio frequency energy. Even on a relatively simple printed circuit board there may be dozens or hundreds of considered output stages of logic elements, therefore the total effect of their simultaneous operation can be very large.
It is impossible to accurately predict the range of frequencies in which these current surges will be located, since the frequency of their occurrence depends on many factors, including the propagation delay of the switching transistors of a logic element. The delay, in turn, also depends on many random reasons arising during the production process. The switching noise has a broadband distribution of harmonic components over the entire range. There are several ways to suppress digital noise, the use of which depends on the spectral distribution of noise.
Table 2 presents the maximum operating frequencies for common types of capacitors.
table 2
Type of
|
Maximum frequency
|
aluminum electrolytic
|
100 kHz
|
tantalum electrolytic
|
1 MHz
|
mica
|
500 MHz
|
ceramic
|
1 GHz
|
It is obvious from the table that tantalum electrolytic capacitors are used for frequencies below 1 MHz, and ceramic capacitors should be used at higher frequencies. It must be remembered that capacitors have their own resonance and their wrong choice can not only help but exacerbate the problem. Figure 15 shows typical intrinsic resonances of two general-purpose capacitors — 10 μF of tantalum electrolytic and 0.01 μF of ceramic.
Actual characteristics may differ from different manufacturers and even from batch to batch from one manufacturer. It is important to understand that in order for the capacitor to work effectively, the frequencies it suppresses must be in a lower range than its own resonance frequency. Otherwise, the nature of the reactance will be inductive, and the capacitor will stop working efficiently.
Do not be mistaken about the fact that one 0.1 microfarad capacitor will suppress all frequencies. Small capacitors (10 nF and less) can operate more efficiently at higher frequencies.
Isolation of power IC
Isolating the power supply of integrated circuits to suppress high-frequency noise consists in using one or more capacitors connected between the power and ground terminals. It is important that the conductors connecting the leads to the capacitors are short. If this is not the case, then the own inductance of the conductors will play a noticeable role and negate the benefits of using decoupling capacitors.
The decoupling capacitor must be connected to each chip package, regardless of how many operational amplifiers are inside the case — 1, 2, or 4. If the op-amp is powered by bipolar power, then it goes without saying that the decoupling capacitors should be located at each power outlet. The capacitance value must be carefully selected depending on the type of noise and interference present in the circuit.
In particularly complex cases, it may be necessary to add an inductance connected in series with the power supply. Inductance should be located before and not after the capacitor.
Another cheaper way is to replace the inductance with a resistor with low resistance (10 ... 100 Ohms). In this case, together with the decoupling capacitor, the resistor forms a low-pass filter. This method reduces the power range of the operational amplifier, which also becomes more dependent on power consumption.
Usually, to suppress low-frequency noise in the power supply circuits, it is enough to use one or more aluminum or tantalum electrolytic capacitors at the input power connector. An additional ceramic capacitor will suppress high-frequency interference from other cards.
TRIMMING INPUT AND OUTPUT SIGNALS
Many noise problems are the result of direct connection of input and output pins. As a result of the high-frequency limitations of the passive components, the response of the circuit to the effects of high-frequency noise can be quite unpredictable.
In a situation where the frequency range of the induced noise is significantly different from the frequency range of the circuit, the solution is simple and obvious - placing a passive RC filter to suppress high-frequency interference. However, when applying a passive filter, one must be careful: its characteristics (due to the nonideality of the frequency characteristics of the passive components) lose their properties at frequencies 100 ... 1000 times higher than the cutoff frequency (f3db). When using series-connected filters tuned to different frequency ranges, the higher-pass filter should be closest to the source of interference. Inductances on ferrite rings can also be used for noise reduction; they retain the inductive nature of the resistance up to a certain frequency, and above their resistance becomes active.
The leads on the analog circuit can be so large that it is possible to get rid of (or at least reduce) them only by using screens. To work effectively, they must be carefully designed so that the frequencies that cause the most problems cannot get into the circuit. This means that the screen should not have holes or cuts with dimensions larger than 1/20 of the wavelength of shielded radiation. It is a good idea to allocate enough space for the intended screen from the very beginning of the PCB design. When using the screen, you can additionally use ferrite rings (or beads) for all connections to the circuit.
CASE OPERATIONAL AMPLIFIERS
In one case usually one, two or four operational amplifiers are placed.
A single opamp often also has additional inputs, for example, to adjust the bias voltage. Dual and quad op amps have only inverting and non-inverting inputs and output. Therefore, if necessary, to have additional adjustments it is necessary to use single opamps. When using additional conclusions, it is necessary to remember that in their structure they are auxiliary inputs, therefore, they should be managed carefully and in accordance with the manufacturer’s recommendations.
In a single opamp, the output is located on the opposite side of the inputs. This can make it difficult for the amplifier to operate at high frequencies due to long feedback conductors. One way to overcome this is to place the amplifier and feedback components on different sides of the printed circuit board. This, however, results in at least two additional holes and cuts in the landfill. Sometimes it is worth using a dual opamp to solve this problem, even if the second amplifier is not used (its outputs must be properly connected).
Dual op amps are especially often used in stereo amplifiers, and quad-op amps are used in multi-stage filter circuits. However, there is a rather significant minus. Despite the fact that modern technology provides decent isolation between the signals of amplifiers located on the same silicon chip, there are still some crosstalk between them. If it is necessary to have a very small amount of such interference, then it is necessary to use single operational amplifiers. Cross interference occurs not only when using dual or quad amplifiers. Their source can be a very close location of the passive components of different channels.
Dual and quad op amps, in addition to the above, allow for a more dense installation. Separate amplifiers are mirrored relative to each other.
It is necessary to pay attention to the fact that the conductors of the driver of a half supply voltage are located directly under the integrated circuit body, which allows reducing their length. This example does not illustrate how it should be, but what should be done. A mid-level voltage, for example, could be the same for all four amplifiers. Passive components can be the appropriate size. For example, planar components of frame size 0402 correspond to the distance between the outputs of a standard SO enclosure. This allows the conductor length to be very short for high frequency applications.
It is necessary to pay attention to the fact that the conductors of the driver of a half supply voltage are located directly under the integrated circuit body, which allows reducing their length. This example does not illustrate how it should be, but what should be done. A mid-level voltage, for example, could be the same for all four amplifiers. Passive components can be the appropriate size. For example, planar components of frame size 0402 correspond to the distance between the outputs of a standard SO enclosure. This allows the conductor length to be very short for high frequency applications.
The types of op amp housings include mainly DIP (dual-in-line) and SO (small-outline). Together with a decrease in the size of the body, the lead pitch is also reduced, which allows the use of smaller passive components. Reducing the size of the circuit as a whole reduces the parasitic inductance and allows you to work at higher frequencies. However, this also leads to stronger crosstalk due to the increase in capacitive coupling between components and conductors.
SURFACE AND SURFACE INSTALLATION
When placing operational amplifiers in DIP-type enclosures and passive components with wire leads, printed circuit boards must have vias for their installation. Such components are currently used when there are no special requirements for PCB sizes; they are usually cheaper, but the cost of the printed circuit board during the manufacturing process increases due to the drilling of additional holes for the conclusions of the components.
In addition, when using mounted components, the size of the board and the length of the conductors increase, which does not allow the circuit to operate at high frequencies. The vias have their own inductance, which also imposes restrictions on the dynamic characteristics of the circuit. Therefore, the mounted components are not recommended for the implementation of high-frequency circuits or for analog circuits located close to high-speed logic circuits.
Some developers, trying to reduce the length of the conductors, place the resistors vertically. At first glance it may seem that it reduces the length of the track. However, this increases the path of current flow through the resistor, and the resistor itself is a loop (inductance loop). The radiating and receiving ability increases many times.
For surface mounting, it is not necessary to place a hole for each component pin. However, problems arise when testing the circuit, and it is necessary to use vias as control points, especially when using small size components.
UNUSED OU SECTIONS
When using dual and quad op-amps in the circuit, some of their sections may remain unused and must be correctly connected in this case. An erroneous connection can lead to an increase in power consumption, more heat and more noise used in this OU package. The outputs of unused op amps can be connected like this: the output of the amplifier is connected to the inverting input.
CONCLUSION
Keep the following points in mind and keep them constantly in the design and layout of analog circuits.
Are common:
- think of a printed circuit board as a component of an electrical circuit;
- Have an understanding and understanding of the sources of noise and interference;
- model and layout schemes.
Printed circuit board:
- use printed circuit boards from quality materials only (for example, FR-4);
- circuits made on multi-layer printed circuit boards are 20 dB less susceptible to external interference than circuits made on two-layer boards;
- use divided, non-overlapping polygons for different lands and feeds;
- Locate land and power landfills on the inner layers of the PCB.
Components:
- be aware of the frequency limits imposed by the passive components and the conductors of the board;
- try to avoid vertical placement of passive components in high-speed circuits;
- for high-frequency circuits, use components for surface mounting;
- conductors should be shorter the better;
- if longer conductor length is required, reduce its width;
- unused leads of active components must be properly connected.
Layout:
- place the analog circuit near the power connector;
- never separate the conductors transmitting logic signals through the analog region of the board, and vice versa;
- make short conductors suitable for the inverter input of an op-amp;
- make sure that the conductors of the inverting and non-inverting OU inputs are not parallel to each other over a large distance;
- Try to avoid using extra vias, because their own inductance may lead to additional problems;
- Do not spread the conductors at right angles and smooth the tops of the corners, if possible.
Decoupling:
- use the correct types of capacitors to suppress interference in the power supply;
- To suppress low-frequency noise and noise, use tantalum capacitors at the power input connector;
- To suppress high-frequency noise and noise, use ceramic capacitors at the power input connector;
- use ceramic capacitors at each power supply pin; if necessary, use several capacitors for different frequency ranges;
- if the circuit is excited, it is necessary to use capacitors with a smaller value of capacitance, and not more;
- in difficult cases in power circuits, use series-connected resistors of low resistance or inductance;
- Analog power decoupling capacitors should only be connected to analog ground, not digital.
Comments: 0
Post a Comment