Power delivered to sensitive analog circuitry must be treated differently than power for digital circuitry. All circuits are affected by noise delivered through the power supply, yet analog loads tend to be more sensitive. The actual type of circuitry and application will determine the tolerable noise limits. Powering digital circuitry today is a fairly straightforward task and can be handled with available power design tools such as National Semiconductor’s WEBENCH tools (www.national.com/webench).
Sensitive analog designs such as those found within medical and video systems often require lower-noise power sources. System designers must pay special attention to ensure that the power-supply rails don’t contain noise above amplitude limits within specific areas of concern. Noise may need to be limited to specific amplitudes over frequency ranges that the system is dealing with (or harmonics of these frequencies).
This article will describe the issues that emerge when designing the power section of an analog system. Focus will be on generating power-supply rails for analog systems and discuss some specific power-supply options for these systems. Another topic is the noise associated with common power circuits and methods of reducing noise. Further, the article will offer tips on good practice when it comes to measuring power-supply noise.
Split-Rail Powering of Analog Circuits
When biasing sensitive analog circuitry, designers often turn to split-rail power circuits to get optimal performance. Analog-system designers utilize differential power rails centered around a system ground to maintain a low-impedance analog reference. Creating an analog reference above system ground (often called a virtual ground) can result in subsequent signal-distortion problems.
A virtual ground can be created with resistors or an active circuit that sets a reference voltage for analog operations (amplifiers, integrators, comparators, etc.). When employing a virtual ground, a designer must deal with the fact that it will vary with ground currents passing through this node (which will always be the case).
As ground currents vary, the virtual ground will also vary by the product of the ground current and the impedance of the node, resulting in unwanted distortion in the signal path. When a true ground is used for the analog ground, this distortion doesn’t exist because the impedance of the ground itself is 0 (or close to it). Therefore, currents in/out of the node won’t cause any voltage change.
Another advantage of designing with a true ground reference is the ability to drive analog signals biased above and below earth ground. This provides clear advantages when it comes to driving differential signals between different earth ground potentials, as seen when driving signals between systems on different power grids.
No matter what tolerances and differences exist between the driving circuit and receiving circuit, the receiver knows where the analog signal can be found. Also, the analog signal biased around earth ground can expect lower leakage currents as compared to a signal that’s always at a potential above (or below) earth ground. Leakage current can result in higher distortion and attenuation of the analog signal being carried. True analog grounds also reduce the risk of signal problems during power up or power down of a circuit. Audio designs often need to deal with clicks and pops generated during powering up or down a circuit when using a virtual ground.
Different physical locations are often at different ground potentials. When power and/or a signal (analog or digital) is connected to a remote location, a designer must be aware that the ground differences between locations could cause system-related problems. Grounds connected together across cables may create unwanted ground currents that increase noise, negatively affecting system performance.
Designers often need to isolate analog signals using coupling capacitors or transformers. For dc signals, this may require special signal coding to ensure spectral density for proper energy transfer. Isolating a power supply makes it possible to float an entire circuit, thus eliminating potential noise issues. In addition, floating the power supply and associated circuitry can create better noise immunity, especially when it comes to induced noise such as ESD or EMI. The circuit discussed later in this article provides the option of isolating the ground and analog rails being generated.
Noise can enter an analog system from many paths, including the power supply itself. Though a properly designed circuit will reject certain amounts of power-supply noise (often specified at an IC level as power-supply rejection ratio, or PSRR), the lower the power-supply noise, the lower the signal distortion caused by such noise.
Ideally a power supply provides a pure dc signal, yet this is unrealistic. All circuits generate noise and regulators generate many types, including thermal, shot, and in the case of switching regulators, switching noise.
In general, from a noise perspective, linear voltage regulators are desirable over switching regulators when powering sensitive analog circuits. Yet, because of the power-efficiency advantage, system designers are being forced to employ switching power supplies and power-supply subsystems (dc-dc converters). The move to switching power in sensitive analog systems will continue to increase as designers are forced to be more conscious of power-efficient design. Today, IC and system designers try to get the most functionality for the lowest amount of power (see www.national.com/powerwise for more information on power-efficient designs).
Switching-regulator noise is generated from multiple sources, with the highest contributions related to the “switching” of current flow. Switching of the current flow transfers energy between passive storage elements (inductors and capacitors) utilizing diodes and transistors to perform the switching. The devices charge during some cycles and discharge during others, resulting in ripple currents and ripple voltages. Switching ripple on power-supply output voltages may be intolerable when biasing analog circuitry (yet often acceptable with digital circuits).
When current flowing through switches (FETs, diodes, etc.) changes direction, as with a switching regulator, the current may instantaneously flow into a high impedance, resulting in high transient voltages. These transient voltages can be somewhat controlled with circuit design (often called snubbers), yet these transitions always result in unwanted conducted and/or coupled noise.
Later we will discuss switching noise (ripple and transient noise), methods of filtering noise, and methods of measuring noise.
The ability to establish a fixed frequency in a switching regulator(s) provides the designer with the ability to move the switching noise to a frequency that doesn’t cause unwanted interference. Locking a regulator’s oscillator to a fixed (or variable) frequency can provide significant performance advantages in analog systems.
System designers can lock the regulator frequency to an available clock or to a software-controlled counter/timer and, as needed, employ software to dynamically move the switching frequency. As an example, this technique has been utilized in radio designs to ensure the regulator doesn’t cause interference while adjusting the radio frequency.
Negative Rail Options
Numerous approaches exist to create a split-rail power supply, all which have tradeoffs. Design techniques include switch capacitor, buck-boost, Cuk, and flyback architectures.
Switch-cap converters: Switched-capacitor regulation is often used to create a negative rail from an existing positive rail. The basic concept is to charge a capacitor followed by reversing the capacitor’s polarity and connecting it to a second output capacitor. The faster the switching takes place, the lower the switching ripple (noise). The higher the load, the higher the ripple for a fixed-frequency and capacitor value, resulting in fairly large capacitors required for high-output currents.
For low-power analog rail generation, a switched-capacitor approach may be found acceptable. For sensitive analog designs, switched-cap approaches may carry more noise than can be tolerated and/or there may be limitations on the amount of deliverable current.
Buck-boost: A buck-boost converter is a switching topology that creates a negative rail from a positive voltage. This type of converter simply steers current through an inductor, followed by steering the inductor current backwards into an output capacitor (Fig. 1).
Using a buck regulator for a positive rail along with a separate buck-boost converter can create a positive and negative split-rail system. This approach requires two separate integrated circuits and two inductors. Both regulators can be synchronized if at least one of the switching controllers provide an external sync input, yet the inherent switching noise should be further filtered when powering most analog loads.
Such an approach is fairly straightforward, yet can be more costly and have higher noise than other approaches. It also doesn’t offer the option of isolating the input from the output.
For more on buck-boost converters, see www.national.com/an/AN/AN-1157.
Cuk: A Cuk regulator architecture utilizes two inductor elements to create the negative rail (Fig. 2). The Cuk converter inverts and can step up or step down the input voltage. The design uses two stages coupled by a storage capacitor. The first inductor stage acts like a switch-mode boost regulator, while the second inductor stage acts like an inverted buck regulator where the inversion of the current flow takes place by steering current on both sides of a storage capacitor (CCUK).
By using inductors on both the input and output, the Cuk converter produces less input and output current ripple compared to other inverting topologies, such as the buck-boost and flyback architectures. The lower noise of the Cuk converter may still be higher than can be tolerated for sensitive analog designs, resulting in the need for additional output filtering.
The Cuk converter’s operating states are shown in (Figure 2. During the first cycle, the transistor switch is closed and the diode is open. L1 is charged by the source and L2 is charged by CCUK, while the output current is provided by L2. In the second cycle, L1 charges CCUK and L2 discharges through the load. By applying the volt-second balance to either of the inductors, the relationship of VOUT to the duty cycle (D) is found to be:
VOUT = –(VIN x D)/(1 – D)
While the Cuk regulator is attractive for creating a negative rail for some analog designs, it does require two inductors and does not provide the ability to fully isolate the output from the input.
For more information on creating a negative rail using a Cuk regulator, see www.national.com/ds/LM/LM2611.pdf. Flyback Topology
A flyback switching topology converts a wide input voltage into one or more output voltages with the option of isolating the output(s) from the inputs. As discussed earlier, output isolation is used when the power supply is delivering power to a remote location, where the ground reference voltages may differ. In place of the inductor found in other switching-regulator architectures, a transformer (or coupled inductors) is used as the inductive storage element.
The flyback topology employs a switch between the primary winding of the transformer and ground (Fig. 3). When the switch is closed, energy is transferred from the input to the primary winding. When the switch is opened, energy is moved from the secondary of the transformer to the outputs.
A PWM modulator controls the switch S1. The modulator is stimulated by an error amplifier, which creates an error term related to the output voltage in relation to a reference voltage. When the flyback control loop is in steady-state continuous-conduction mode, the PWM duty cycle (D) relates to the VIN, VOUT, and the turns ratio of the transformer and the forward-voltage drop of secondary catch diodes.
D = (VO + VF)/(VO + VF + ((NS / NP) x VIN))
NS = turns on the secondary windings
NP = turns on the primary windings
VF = forward voltage drop of the catch diodes
The modulator and loop error amplifier are internal to the flyback regulator IC. Some designers consider flyback regulators problematic and shy away from using them. By using recent integrated circuits such as National Semiconductor Corp.’s LM5001 (see “LM5001 Backgrounder”), and an available transformer, one can produce a relatively simple design and realize a superb power source for analog circuitry. As with any design, one does need to be aware of potential issues, including the inherent switching noise generated by a switching power circuit.
The LM5001 regulator toggles between two phases at the switching frequency (Fig. 4). During phase 1, the primary switch S1 is closed causing current flow into the transformer core (shown in blue); the secondary catch diodes (D+ and D–) remain off; and energy is delivered to the output from the output capacitors (C+ and C–). During phase 2, the secondary delivers energy to the outputs and output capacitors via the forward-biased catch diodes (shown in red). The output capacitors are charged during phase 2 to ensure continuous delivery of energy during phase 1.
Current changes paths during transitions between the two phases, causing voltage excursions that must be considered. As the regulator enters phase 2, S1 is opened and the primary current continues to flow as the field collapses. The result of this current flow will be a high voltage induced across the switch.
This transient voltage can result in two different problematic issues: the voltage may exceed the limits of the IC itself, and the high voltage transients can cause coupled and emitted noise. Placing a circuit across the primary winding of the transformer (often called a snubber) can limit this voltage, significantly reducing both of these issues. A simple RC and/or Zener diode is often used.
Snubber circuits can dramatically reduce voltage transients and noise, yet the designer must analyze the extra power dissipated in the snubber circuit and may need to trade off between noise and power efficiency. A snubber circuit may also be needed across the secondary catch diodes to limit voltage excursions when the diode turns off, although this may be a minor contribution to the noise budget.
No power supply is perfect. Yet when properly designed, a flyback regulator topology has some key advantages when developing split analog rails.
Key Design Elements
One can design a physically small flyback circuit to deliver very low-noise rails for split-rail analog systems. The flyback circuit described below employs an LM5001 integrated flyback regulator with a very small surface-mount EP5 transformer. This circuit inputs 10 V to 30 V and comfortably delivers 250 mA on each output (±5 V). A system requiring higher current can use the LM5000 to deliver about twice this amount of power.
Isolation: When input to output isolation is needed, one also must add isolation from the output(s) back to the regulator’s feedback input. This can be accomplished by simply adding an LM431 (or LMV431) voltage-reference/amplifier IC to monitor the output voltage and feedback an error current through an optoisolator.
Primary and secondary ground planes should be sufficiently spaced yet connected with a high-voltage capacitor to reduce unwanted output noise during switching. Isolation limits are set by the breakdown limits of the transformer, the optoisolator, and the capacitor across the ground planes.
Transformer: The transformer should be selected properly for ideal operation, smallest size, and lowest cost. Transformer design involves theory that is not included in this article. Popular inductor companies are capable of using the key parameters from a design to recommend an available transformer or produce a transformer optimized for an application.
The key parameters for the transformer include physical size, inductance, and turns ratio. The physical size and type of core determines the amount of energy that can be stored (in phase 1) before saturation. A larger-sized core also provides the ability to lower the equivalent series resistance (ESR) by using lower gauge wire, resulting in lower losses. In general, the larger the transformer, the more power that can be delivered and/or the better the efficiency.
The inductance determines the transformer’s rate of storage and discharge and ultimately the magnitude of the transformer’s ripple currents. Higher inductance will lower ripple current and can improve operation. However, as inductance increases (more turns) the resistance of the winding increases, resulting in higher transformer losses. These tradeoffs should be considered when selecting a transformer.
As briefly discussed above, the turns ratio is determined by the range of VIN and VOUT. The limits of the turns ratio (NS/NP) is determined by the IC duty cycle limit and the range of VIN. I target a PWM duty cycle (D) between 25% and 45%. The minimum and maximum pulse width can easily be calculated at a specific operating frequency and compared to the limits specified for the IC being used. Reference the regulator’s datasheet for more details on setting the operating frequency.
Tradeoffs exist between size, inductance, and series resistance. All of these tradeoffs can be worked through by any application engineer working for one of the inductor companies.
Circuit protection: Current limiting and thermal limiting is provided by the LM5001 series of regulator ICs. The current limit is fixed as per datasheet specification. Though a direct secondary current limit isn’t available with basic flyback circuits, power delivered to a heavily loaded secondary will be limited by the IC primary switch current limit. If the secondary outputs are shorted, the regulator will limit current. Still, the circuit needs to absorb additional power, leading to increased temperatures on the transformer, the catch diodes, and the regulator IC.
Bench testing of a circuit (described later) shows that a shorted output (V+ to V–) results in increased temperatures on the IC, resulting in thermal shutdown of the IC (as per specification). After this fault condition, the shorted output must be removed and the input voltage cycled before the circuit will return to normal operation. As with any design, a thorough reliability study should be performed to ensure safe and reliable operation in the specific application.
Most system implementations work safely without extra protection circuitry on the secondary side of the transformer, yet if the power rails might be inadvertently shorted, some designers may opt to add a fuse or other means of extra protection.
As described earlier, during phase 1, S1 is conducting and energy flows from CIN into the primary of the transformer. S1 turns off at the end of phase 1, causing the primary current to continue to flow into the high impedance of the FET. The result is a voltage equal to IPRIMARY x ZFET_OFF. Without a snubber, the switch voltage could exceed beyond the maximum voltage allowed, resulting in damage to the IC.
An optional steering diode reduces the energy losses in the snubber by only providing current flow during the transition from phase 1 to phase 2, while remaining off during other phases of operation. The Zener diode (Z1) provides a worst-case clamp to ensure the switch voltage doesn’t exceed the maximum allowed by the IC. The Zener will only turn on under worst-case extremes, as when the outputs of the circuit are shorted and then opened up.
Switching Noise and Frequency
Switching power supplies all exhibit noise associated with the charge and discharge cycle of the output capacitor (called ripple noise), as well as the transient noise caused by the turn on and turn off of the switching devices (discussed earlier). To reduce ripple noise, boost the switching frequency and/or increase the inductance of the transformer (Fig. 5).
Higher frequency reduces ripple noise, yet there’s always a limit to the switching frequency. Frequency limit is a function of the switches being employed (often inside the IC) and the pulse-width limit of the regulator itself. At higher frequencies, the output switches will dissipate more power caused by the ac losses. This loss should be considered when selecting a switching frequency.
Because the transformer turns ratio in a flyback converter can be selected to provide an ideal input/output voltage ratio, the minimum pulse-width limit is often not a limiting factor with a flyback design (as it is with other switching topologies). The flyback controller does have a maximum oscillator frequency that must be understood, yet switching losses often limit the frequency well before the limit of the oscillator itself.
In the design shown later, 600 kHz was chosen for the switching frequency based on the tradeoff of efficiency and size of the transformer. At 600 kHz, the design was able to obtain efficiencies above 80% while utilizing a very small EP5 transformer.
Filtering Switching Noise
Adding a noise filter to the output of a switching power supply is a good method of reducing switcher noise (Fig. 6). A simple LC low-pass filter can be employed to reduce ripple and transient noise to a level acceptable for most analog power supplies. This design adds a series inductor followed by capacitors to ground to provide the filtering needed. One can use various combinations of inductors and capacitors to do the job.
In the application described below, I started with a low value inductor, which provided a low ESR to minimize IR droop and power losses across the inductor (Fig. 7). I targeted a 40-db noise attenuation at an octave below the 600-kHz operating frequency, so I used a 60-kHz, 3-db point for my calculations. A 1-µH inductor value results in a capacitor value of 7 µF, so I used a small 10-µF ceramic capacitor that I had in my lab, C = 1/(f2 x 4π2L). If desired, you can also start with a capacitor value and calculate for the inductor value.
In lieu of an LC filter, some designers may opt for an additional linear regulator after a switch-mode regulator. The idea is to use a switching regulator to regulate to just above the voltage needed for the linear regulator to operate properly (called the dropout voltage). Driving the linear regulator input just above the dropout voltage creates an accurate voltage while minimizing energy losses.
Linear regulator circuits can be used as noise filter circuits, but be aware that a linear regulator itself doesn’t provide much switching noise attenuation. Linear regulators usually have minimal effect on noise above just a few kilohertz, yet properly chosen input and output capacitors can provide significant low-pass filtering. For very sensitive analog circuitry, pay attention to the noise generated by the linear regulator. Low-noise linear regulators are available for biasing of such circuitry (see the specific regulator datasheet for details).
Based on the additional efficiency loss and circuit cost, designers often find a passive filter acceptable for filtering switching noise for most low-power analog rails. For high-power analog rail generation, the IR drop across a filter inductor may result in unacceptable droop or load transient regulation, justifying the use of an additional active element.
All power supplies that employ feedback should be checked for stability over all operating conditions. As with any amplifier, a gain of greater than 1 with a phase shift of 180° results in oscillation. Most power supplies are compensated with a simple RC delay circuit placed across the error amplifier. The idea is to attenuate the gain of the amplifier at higher frequencies, where the phase delay approaches 180°. Because the loop bandwidth is limited, the circuit’s ability to react to fast load changes is also limited.
With fast-changing digital circuits, an over-compensated regulator circuit may inhibit the regulator’s ability to change load current fast enough, resulting in voltage droop, overshoot, and ringing as load currents change. For most analog rails, the load current is fairly constant, so this is usually not a problem. As a result, the loop bandwidth can be limited without any significant effect.
It’s a good idea with any design to verify the circuit’s stability over operating conditions. The loop stability can be checked using different methods. One of the simplest methods is to load-step the output current from its minimum to maximum over the operating conditions of the design (VIN range, temp, etc.).
During these load-steps, observe the output voltage to ensure the voltage(s) remains within tolerance and returns to the proper level without any significant ringing, overshoot, or undershoot. A good rule of thumb is to ensure that the voltage returns to a stable output within approximately three ring cycles, and the excursions aren’t outside of system limits. If excessive ringing is evident, adjustment of the compensations circuit may be necessary.
For those that have access to a network analyzer, one can verify operating extremes and directly read phase margin off of a Bode plot. More information concerning measurement of phase noise and stability can be found at www.national.com/an/AN/AN-1889.pdf.
Circuit Performance Measurements
The circuit in Figure 7 and Figure 12 is a working design that produces +5 V and -5 V rails on a small double-sided PCB. These power rails are appropriate for use in biasing most sensitive analog circuits. This circuit has been successfully employed in designs where the power is delivered from a remote power source (10 V to 30 V) that’s a significant distance away from this circuit. The outputs are fully isolated from the input, so the earth grounds at the two locations can vary without affecting the circuit being powered.
This design provides clean and stable rails for powering a remote analog (and/or digital) system with superior system performance. Other low-noise split-rail power designs often require multiple regulators, resulting in higher noise, larger PCB size, and higher cost. For a more detailed explanation of the circuit operation and the full design package (PCB, BOM, etc.) please go to www.national.com/rd/RDhtml/RD-171.html.
Output regulation: Not all analog systems that employ split-rail biasing are affected by variations of the actual rail voltages. Signals are often centered by ac coupling and biasing between V+ and V–, thus as long as the differential rails remain high enough in amplitude for the signal to remain within an acceptable common-mode range, all will operate properly. In any case, proper regulation must be considered and in this case the target was 5% of the nominal rail voltage.
A flyback design must take into account a few points when it comes to accurate regulation. Because the feedback control loop only comes off of the positive output, the outputs must maintain a small load at all times to ensure 5% regulation (30 mA with this design). If the load current is less, the negative output voltage may float below –5.25 V.
Also, the +5 V rail should carry a small load to ensure that the negative rail maintains proper regulation during high negative loads. If the positive output doesn’t carry this small load, when the negative output is loaded, the regulator’s on time may be reduced to a point to where the transformer core doesn’t receive sufficient energy to maintain the loaded negative output.
Because most designs will require a minimum current well above these low limits, no extra design effort is needed. For designs with very low shutdown currents ( Regulation of the positive rail stays within 5% tolerance over loads from 25 mA to 250 mA (Tables 1 and 2). The negative rail is slightly less accurate, yet also maintains 5% cross regulation with loads above 30 mA. Though the negative rail isn’t directly regulated with a feedback loop to the regulator, it provides acceptable regulation via mutual coupling of the transformer.
Output noise: In Figure 8, noise was measured on each rail with respect to the output ground and the differential +5 V to –5 V noise. In all cases, the transient noise was below 20 mV p-p and ripple noise below 5 mV p-p. Because of the symmetry of the flyback secondary design, some differential noise may cancel. Symmetrical layout of the secondary circuit and optimized transformer design improves differential noise cancellation.
The noise measurements were taken using a high-performance differential probe with ac coupling placed directly across the output connector P2. Be aware that some of the rounding of the triangle wave caused by the ripple voltage results from the ac-coupling capacitors within the probe used, and isn’t an effect of the circuit itself. The actual noise should be less than shown here.
Stability: Control-loop compensation is accomplished by the RC circuit, which is connected to the the LM5001’s COMP pin (R6/C13 and R9/C12). In this circuit, the feedback signal is safely fed into the same COMP pin of the regulator and the feedback (FB) pin is shorted to ground. This provides a method to bypass the voltage reference and error amplifier internal to the LM5001, allowing the use of a separate reference and amplifier on the secondary side of the transformer (U2 – LM431).
The design shown is very stable and provides over 45° of phase margin. Because the output filter is outside the feedback loop, it doesn’t affect the stability of the control loop (Fig. 9).
Efficiency: Efficiency of the flyback design shown was measured to be above 80% over most of the operating range (Fig. 10). Power losses arise from the transformer, the catch diodes, and the IC itself (internal switch and biasing). As mentioned above, the switching frequency was limited to minimize the ac losses, yet still provide the advantages of a small transformer core size.
At low loads and high input voltages, IC biasing dominates the losses. At higher loads, the transformer begins to saturate and starts to dominate the loss budget. The design as shown runs without any single element losing significant amounts of energy, thus no component runs at elevated temperatures when delivering full power.
Noise on any power bus often emits EMI, which may cause regulatory issues during system testing. Load transients, switching ripple noise, and switching transient noise on a power bus can create unwanted radiation. Reducing noise via the methods discussed may significantly reduce system EMI.
Too often, intermittent system problems relate back to power problems that could have been addressed during the early stages of a design. A solid study of the power planes early in the debug stage might uncover addressable noise issues, thereby eliminating potential residual problems before they surface.
When measuring noise on a power bus, understand what you see. Large amounts of energy can radiate from power circuits, so a less-than-ideal probe without proper ground connections can cause improper measurements. Use one or two high-frequency probes with a small ground stub; a ground wire of any length can result in false noise measurements.
The oscilloscope must clearly display and trigger up to the noise frequency to which your circuit might react. At minimum, I suggest a 500-MHz oscilloscope and a passive probe with a ground stub. For better results, use two probes with the oscilloscope set for a math function that adds the two inputs, with one probe set to invert (Fig. 11).
This differential method reduces the risk of ground currents that may occur when using a single probe. For best results, I use an active differential probe to measure power-supply noise. Again, use the shortest connections from the probe pins to the output and measure directly across the high-frequency output capacitor if possible.
When measuring noise on a split-rail power supply, you should measure the noise from the positive to the negative rail. Since the noise sensitive circuits are often powered from the differential rails, one should measure the noise differentially across these rails.
Trigger the oscilloscope using normal mode, with the threshold adjusted for extreme peaks. Make two measurements, one with the scope triggering on the positive edge and the other with the scope set for negative slope. The difference between the two numbers is the peak-to-peak noise voltage. Measure the frequency of the ripple noise to verify the switcher is running at the correct frequency. Keep in mind that the ripple voltage is measured to the peaks of the triangle wave, which is usually lower than the switching transient voltage peaks (Fig. 5, again).
You can measure the frequency of the transient noise by triggering on the voltage peaks and setting the time base higher. Be sure that no bandwidth limiting is enabled on the oscilloscope. As mentioned earlier, these transients can be limited by adding a snubber(s) across the switching elements. In some cases, noise components may be at frequencies higher than the system can respond to, and thus not cause problems.
Noise often results from system operation, possibly a function of a circuit turning on or off. Or in the case of digital systems, a processor may be executing a high-performance subroutine along with low-power idle modes. A thorough understanding of the system operation will help identify other sources of noise affecting analog circuitry. Synchronizing system software with analog operations can often improve analog performance.
Spectrum analyzers can be used to quickly identify system noise components on a power bus in a complex system. The spectrum analyzer helps identify the noise sources by providing the exact frequencies of various noise components.
Taking the time to properly observe the power supplied to analog and high-speed devices is worth the effort. Some bench time early in the prototype debug stage can save lots of debug time later on. Very often I see system power problems surface as unrelated system anomalies or performance limitations.
Good analog designs starts with a clean analog power supply. The circuit presented provides a very clean and fully isolated +5 V and – 5 V rail. It easily fits on a small double-sided PCB with all components placed on one side of the board (Fig. 12). The layout shown can shrink further by using components available in smaller packages. If isolation is unnecessary, size and cost reductions are possible by removing the feedback isolation circuit.
Higher voltage outputs are possible, and higher currents are attainable, by using the higher power LM5000. Though the transformer needed for a flyback design does cost more than a single inductor, other approaches will often require more than one inductor and more than one regulator, resulting in a higher overall solution cost, and likely higher noise. By employing an integrated flyback IC along with post filtering, one can design a high-performance split-rail power supply with superior system results.