- Created on Saturday, 15 October 2011 13:41
- Last Updated on Friday, 20 April 2012 06:59
The objective of Series 1 of the Missing Lecture Notes (MLN) series is to develop A LQ043T3DX02 (Sony PSP) Display Driver Board. In part 4, of the series, an improved backlight circuit is presented by using the MC34063A multi-purpose regulator. In this article, using the equations provided in the datasheet, a method to determine the values of the external components is demonstrated. The calculated passive component’s values and the MC34063A have then been simulated in PSPICE. The resultant simulation is used to confirm that the analytically derived results drive the MC34063A as expected. In the final part of the article an investigation is made into the optimum printed circuit board layout of the circuit to minimize ground rise and ground bounce.
At the end of the last part in the series the design had evolved to the point where a preliminary component list had been assembled and we were in a position to start thinking about the PCB design stage of the project. However, the design would only have lit the backlight of the TFT-LCD module to about 1/6 of its maximum brightness. [Or probably not at all – B.P]
A comment received from a reader, Igor, prompted me to revisit the backlight section of the design. In particular it was suggested that the MC34063A family of 1.5A step-up/step-down/inverting switching regulators may be a good choice to use as a step-up regulator in the LCD-TFT module interface board design. The device should be capable of generating a voltage that is just slightly less than the TFT-LCD module’s maximum backlight potential difference of 29.4V and a maximum current of 24mA.
Figure 4-1: Different flavours of the MC34063A Boost Regulator: Evaluation boards are usually laid out to optimally showcase a manufacturer’s product. They should be examined thoroughly to squeeze every last inch of performance out of them [Modified picture taken from the OnSemi MC34063A Evaluation Board datasheets].
Previously, by using the maximum available potential difference on the board, I had sacrificed brightness for simplicity and I was quite pleased too! However, I became intrigued about the comment of using a switching regulator in the backlight circuit design so I decided to analyse the MC34063A to assess its potential use in the project.
One of the encouraging things about undertaking an electronic design project using FPGA’s is that you are completely in the digital design domain. Any errors that occur in logical reasoning can be rectified by the design engineer re-thinking the logic and re-programming the device. This is not the case with analog electronic designs where any mistakes can lead to expensive re-spins when the error is not critical’ However, if the analog design engineer is unlucky, damaged equipment could be the end result if the error is critical.
Sometimes however a digital design engineer is called upon to do the unthinkable and enter into the analog design domain and PSPICE. A world where poor design work is quite quickly and ruthlessly exposed. Bearing this in mind, once I had decided to design the backlight circuitry using the MC34063A device, I decided to approach the design with extreme caution.
However, if I thought that the best thing to do in entering the murky world of analog electronics is to be prepared I was wrong; you need to do much more! By using the MC34063A you achieve the brightness required at the expense of a more complicated design, but you know what they say “There is nothing like a free lunch”.
4.2 The MC34063A Switching Regulator
The MC34063A can be configured to work in three different regulation modes which are as a step-up regulator, step-down regulator or as an inverting regulator. The backlight of the TFT-LCD module can accept a potential difference of up to a maximum voltage of 29.4V. Now, since the largest potential difference on the prototype board is 5V it is not unreasonable to attempt to use the device to step up the 5V supply to about 80% of the allowable maximum potential difference of the TFT-LCD module or roughly about 24V.
Internally, the MC34063A consists of the five major active functions (Figure 4-2) required in implementing a general purpose switching regulator. One of these functions is a comparator, (1), which is used to determine when the output voltage has fallen below a particular threshold voltage. The threshold voltage is provided by a 1.25V reference regulator, (5).
When the scaled output voltage has fallen below 1.25V, the control logic switches a Darlington transistor, (4), on. The control logic is provided by a controlled duty cycle oscillator, (2). The oscillator has an active peak current limit which is used to provide the Pulse Width Modulation (PWM) control of the Darlington transistor, (4). Hence, the Darlington transistor is used as a switch.
It is the rate at which the switch charges and discharges, determined by an external timing capacitor connected to pin 3 and the internal S-R flip-flop, (3), that determines the flow of current through an external inductor. Depending on the state of the flip-flop’s two inputs, which are a combination of the comparator and the oscillator, the transistor is either driven into its saturation state or cut-off state.
Figure 4-2 : M34063A Switching Regulator
In a boost configuration the circuit is generally wired-up as can be seen in Figure 4-10. As stated before our intention is to take a 5V DC input voltage and produce a DC output voltage of about 24V. By doing this the potential difference of the backlight circuit will be 19V higher than the prototype PCB’s input voltage and hence we need to operate the regulator in a boost configuration.
Why are we not trying to produce an output voltage of 29.4V to match the maximum allowed potential difference of the backlight? Well, in general it is never a good idea to drive any electronic device at its maximum rating as by doing so we run the risk of shorten the life expectancy of the device.
Also, if transient voltages and other effects are taken into account then the input voltage to the backlight circuitry could be momentarily taken past its maximum value when they occur. So in this instance I have tried to achieve a backlight voltage of roughly 80% of its maximum allowable value. This should provide an adequate brightness without running the risk of damaging the device.
It is not an uncommon technique when choosing energy storing passive components like inductors and capacitors to choose them such that they are used to between 66% and 75% of their maximum rated value.
4.2.2 Electrical Characteristics
Before we dive into the application of the MC34063A we will investigate some of the electrical characteristics of its active functions presented on the third page of the MC34063A’s datasheet. The electrical characteristics are categorised functionally with the functions being oscillator, output switch and comparator.
Figure 4-3: The electrical characteristics of the MC34063A’s internal oscillator.
Internally, the oscillator which typically oscillates at 33 KHz, (1), consists of a current source and current sink that charges and discharges an external timing capacitor, CT. Typically the charge current is 35uA and the discharge current is 220uA. Now since the charge (discharge) time is inversely proportional to the charge (discharge) current it follows that the ratio of charge time to discharge time is approximately 13:2 (2). Alternatively, this is described as the on-time, ton, being 6.5 times longer than the off-time, toff.
Figure 4-4: MC34063A timing curves taken from the datasheet. The exact ratio of on-time to off-time can be determined by choosing a timing capacitor and determining the ratio from the graph above.
An example plot of oscillator timing capacitor versus the output switch on-off time is presented in the datasheet on page 4 and is repeated in Figure 4-4 above. It can be seen when a 1.5nF timing capacitor, CT, is used an on-time, ton, duration of about 38us will result. The corresponding off-time is then about 5us. With these on-off times the corresponding ton/toff ratio will be approximately 7.6, which is on the outer maximum limits of the ratio quoted in the datasheet.
Figure 4-5: In the example plot of oscillator voltage against time shown, the on-time is roughly about 25us and the off-time is 3.3us. The charge to discharge ratio can then be considered to be about 7.6.
An example of the timing capacitor waveform is also provided in the datasheet on page 4 (Figure 4-5) where it can be seen that when a timing capacitor of 1nF is used along with an input voltage, VCC, of 5.0V then the on-time to off-time ratio approximately 7.6.
22.214.171.124 The Darlington Transistor
A Darlington configuration is used in the implementation of the MC34063A’s switch, rather than a single transistor, to increase the device's current throughput. Some useful characteristics of the switch are shown in Figure 4-6. The important thing to remember about the operation of transistors in general is that they are normally operated in two basic modes of operation.
Generally, a transistor is either used to amplify current in instrumentation type applications or, as is the case with our boost regulator, it is used as a switch to control the current through a load. By switching the current we are inadvertently controlling the voltage across the load.
Figure 4-6: The electrical characteristics of the MC34063A’s output switch.
When operated as a switch a transistor is operated in the transistor’s saturation, (1), and cut-off regions, (2), shown in Figure 4-7. When operated in the saturation region the voltage at the collector is only marginally larger than the emitter voltage and the base current is biased such that the maximum collector current flows. When biased under these conditions the switch is fully switch ON. In the case of the MC34063A the typical collector-emitter saturation voltage is 0.45V [(2) in Figure 4-6].
Figure 4-7: Saturation and Cut-off regions of a typical transistor used as a switch.
On the other hand when there is no base current, no collect current flows through the transistor and the collector voltage is at its maximum and the transistor is considered to be fully switched OFF.
Consider, for example, the collector-emitter saturation characteristic curves of the popular 2N3906 transistor see in Figure 4-8. Then, if a collector saturation current of 30mA, (2), is required to drive a load, the base current can be calculated by using the forced gain, or beta, ß (1), as IB = IC x (1/ß) or 3mA. Now, supposing we want to control a switch using a FPGA pin set at a VCCIO of 3.3V then a base resistor of 1.1K Ohms will be required.
Figure 4-8: Collector-Emitter Saturation Voltage Vs Collector Current taken from the 2N3906 device's datasheet.
So if the current drawn is at a maximum, when the transistor is operated such that it is fully ON, do we need a heat sink? Well, we should never really need a heat sink on our MC34063A device considering that the power dissipation, Pt, due to the transistor when it is switched ON, is PT = 30mA x 0.07V or 210 x 10-5 Watts or 2.1mW! Likewise, when the transistor is OFF the collector current is in the micro-amps and the power consumption is negligible.
126.96.36.199 The Comparator
The comparator is used to compare an output voltage with a threshold voltage, VTH, of 1.25V (Figure 4-9). The output voltage is usually scaled down to be marginally larger than 1.25V by using two external resistors, R1 and R2 , in a voltage divider configuration.
When the scaled value of the output voltage falls below 1.25V the comparator’s output value is asserted to notify the MC34063’s internal logic that action must be taken to adjust the fall in the output voltage. We utilise the gain of the comparator at the expense of contributing to the voltage ripple of the output voltage.
Figure 4-9: The electrical characteristics of the MC34063A’s comparator.
Armed with this information we should now be ready to tackle the requirement of configuring the MC34063A in a step-up regulation mode to drive the LEDs that function to provide the backlight of the TFT-LCD module.
In this section the component values are chosen based on the boost switcher configuration of the MC34063A.
4.3.1 Design Consideration
Before we start we should bear the following design consideration in mind:
The output current and the switch current should not be equal.
The maximum available load current is always less than the current rating of the switch transistor. (You could use an external transistor to supply more current. This configuration is not required in this exercise.)
The maximum power available for conversion is equal to the input voltage multiplied by the maximum average input current. We cannot create energy out of nothing.
Since the output voltage is higher than the input voltage it follows that the output current must be lower than the input current.
The MC34063A voltage regulator is operated as a step-up regulator by implementing the circuit shown in Figure 4-10. By operating the internal switch in one of two modes it is forced to be either on or off creating the three major current loops shown in the diagram. Two of these (Loops 1 and 2 in Figure 4-10) loops are associated with the switch being ON while the third is as a result of the switch being OFF.
Figure 4-10: The MC34063A in a boost circuit configuration.
When the switch is “ON” a changing current is drawn from the input voltage, Vin, that is forced through the inductor, L1. Now, according to the Biot-Savart Law when a time varying current flows through an inductor a magnetic field is developed that is proportional to the current that produces it. In this case the current is the switch charge current.
An inductor, which is essentially a multi-turn core device, is used to concentrate the magnetic field lines at its centre and hence maximize the lines of changing magnetic flux cutting through it. As long as the rate of change of current is not zero the magnetic field produced will increasingly store energy. Hence, in the same way that an electric field is used to store energy in a capacitor when used as a voltage reservoir, the magnetic field created in an inductor is also used to store energy that is used as a current reservoir.
Now, according to Faraday’s Law any time-varying magnetic flux through an inductor induces an electromotive force that tends to oppose the motion of the current that produces it. Hence, a voltage is induced that is of opposite polarity to the input voltage, Vin. As a consequence of all this the diode, D1, is reversed biased relative to the output capacitor and the load current is supplied solely by the capacitor, CO, as can be seen in Figure 4-11.
Figure 4-11: Current loops of a typical switching regulator.
When the voltage across the capacitor, CO, falls below the threshold voltage determined by the voltage divider resistors, R1 and R2, the switch turns off. When this happens the magnetic field collapses causing the diode to become forward biased and the capacitor, CO, charges up to the inductor voltage VL, which by now is considerably larger than the input value VIN or in our case 5.0V.
At the same time as the capacitor is charging, current from the inductor is also supplied to the backlight LED’s. This is fundamentally how all step-up or boost inverters work. However, as the output voltage is larger than the input voltage it follows conversely that the input current must be larger than the output current. We cannot create energy out of nothing!
4.3.2 Determining the Passive Component Values
Supposing we wish to drive the backlight in the TFT-LCD module under the following conditions:
Table 1: Output power supply requirements.
If the minimum frequency of oscillation, fMIN = 24KHz (see the datasheet) and if we consider the lowest voltage, that the 5.0V power rail (into the prototype board, VCC = 5.0) can fall to, to be, VIN(MIN) = 4.5V, with an allowable voltage ripple of 0.5% of the output voltage then, Vripple = (0.5% x VOUT) = 25mVpp. The design parameters for our step-up regulator are summarised in the Table 2.
Table 2: Boost regulator design parameters.
Given these conditions and the set of equations on page 10 of the datasheet we can precede in determining the values of the inductor, L1 resistors R1,R2, ROSC and RCDR the capacitors COUT, C1,C2 and CT using the equations provided in the datasheet and repeated below.
Figure 4-12: The datasheet provides a list of formulas that can be used to determine the passive component values.
Firstly, we determine the ratio of the switch conduction time, tON to the diode conduction time, tOFF.
Then we calculate the cycle time of the LC network.
From the result of equations (1) and (4) calculate the off time using equation (7) below.
Then determine the on time, ton, using equation (10)
From the maximum on time, ton(max), determine the value of the timing capacitor, CT by using equation (13).
For the timing capacitor we should choose the standard capacitor value of 1500pf.
Next determine the peak switch current, IPK(SWITCH), using equation (17).
From there we determine the minimum value of the inductance, LMIN, using equation (20).
For the inductor we should choose the common inductor value of 470uH.
Next we determine the value of the current limiting resistor ROSC by firstly calculating the peak switch current.
The peak switch current IPK(SWITCH) is re-calculated using the minimum inductance value, LMIN, = 235.67uH,
Then using this value and the value of the external oscillator resistor, ROSC is calculated.
Next we approximate the minimum value of an ideal output filter capacitor.
According to the last equation in Figure 4-12 we should choose CO to be nine times this value or 630uF. I think that this value is too conservative and instead I will use a value of 220uF rated at 50V.
The ripple voltage contribution due to the gain of the comparator taken from the application note is given by using the equation:
If a 220uF capacitor with an ESR of 0.10 Ohm is chosen then the ripple voltage due to this capacitance value is given by:
Since the chosen capacitor is considered to have an Equivalent Series Resistance (ESR) of 0.1 Ohms then the ripple voltage due to the ESR of the capacitor is given by:
Therefore, the total voltage ripple is 28.8mV + 7.88mV + 59.6mV or 96.28mV with more than half of this being the result of the ESR of the capacitor. In general tantalum capacitors have a lower ESR compared to the aluminium electrolytic ones however, the aluminium electrolytic ones are cheaper especially in small order quantities.
The nominal output voltage is programmed by the R1, R2 resistor divider network. If a standard resistor value of 2.2K Ohms is selected for R1 then an insignificant amount of current of about 1.25/2.2K or 568uA is consumed. If a value of R1 = 2.2K is used in the resistor divider network then the value of R2 is determined by:
The output switch transistor is driven into saturation with a forced gain, Bf, of 20 and therefore the base current Ib is given by:
To calculate the value of the drive collector resistor, RCDR, it is necessary to analyse the diagram in Figure 4-13:
Figure 4-13: MC34063A collector current equation.
The collector current, IC, is approximately equal to the emitter current, Ie and according to Kirchoffs current law, Ie = Ib + I100.
The current, I100, through the 100 ohm base resistor is given by:
Therefore the value of the collector drive resistor is given by
Choose a standard resistor value of 110 Ohms.
4.4 Bill of Materials
A tabulated list of our chosen components is presented in Table 3. The table shows that the material cost of adding a more capable backlight circuitry is considerably small especially when compared to the human effort involved in implementing it.
As mentioned previously I have intentionally chosen components that are readily available from online electronic component suppliers which I think is the safest route when requiring small order quantities for prototyping. If I had used some of the recommended devices in the datasheet I might have been able to achieve a higher level of performance but this would have been risky as some of the parts might not have been readily available.
Table 3: Bill Of Materials.
From the discussions of the previous section it is clear that choosing a capacitor with a low ESR is important if we wish to minimize the effects of voltage ripple on the output voltage induced by the capacitor. A technique used to reduce the effect of ESR is to use two similar capacitors in parallel which would effectively halve the ESR.
When choosing an inductor it is important to remember that cost can be a major discriminator on the type of inductor used. Toroid inductors can be used in designs where high performance is a major influencing factor as any magnetic flux generated is completely contained within their magnetic core. This helps to reduce the EMI and noise associated with inductors in general. On the other hand if cost is the primary concern then the less expensive bobbin core inductors can be used.
However, as the core of bobbin inductors do not completely confine the magnetic flux they generate a lot of radiated noise that can affect nearby circuits especially those that are sensitive.
4.4.1 Diode - MBRS140
I have chosen to use a standard surface mounted diode that can provide a maximum average rectified forward current of 1.0A. Again, this could be considered to be overkill for the project as the maximum amount of current drawn by the backlight LEDs is quoted as 24mA.
Figure 4-14: The maximum ratings of the MBRS140 diode.
However, the part is so cheap and readily available that I do not think that it is really worth considering anything else. The maximum instantaneous forward current of the device is 0.6V, the number that we have used in equation (2) above to calculate the on-time to the off-time ratio.
Figure 4-15: The electrical characteristics of the MBRS140 diode.
4.5 PSPICE Simulation (using LTSpice IV)
In this section I have produced some of the results of simulating the design using LTSpice IV, a PSPICE variant available free of charge from the Linear Technology website. By simulating the boost regulator circuit containing the MC34063A and the analytically calculated component values we can verify the behaviour of the backlight circuit.
Primarily, the simulation allows us to determine whether we have achieved our main objective of creating a boost regulator design that should convert a 5V input supply voltage into an output supply voltage of approximately 24V and a steady current supply of at least 24 milliamps.
Also, a visual inspection of the results produced by the PSPICE simulation should allow us to determine whether there will be a need to improve the performance or efficiency of the design. It might be deemed necessary to improve the performance to reduce the switching noise for example. Of course, any actions we take are entirely dependent on the accuracy of the PSPICE models used and the PSPICE application package.
The spice model of the MC34063A used in this simulation has been obtained from the OnSemi website. The schematic diagram in Figure 4-16 shows the layout of the circuit used in the simulation. All of the component values used have been derived previously in Section 4.3 above.
Figure 4-16: The MC34063A simulation circuit.
The circled numbered items could form the basis of test points. The critical voltages of interest are labelled 1-4 and are the (1), The input voltage, (2) The output voltage, (3) The voltage output feedback voltage and (4) The switch collector voltage.
Figure 4-17: The output voltage shown in two periods, the start-up period and the quiescent period. In this particular example the steady output voltage is achieved after 90ms.
The simulation plot presented in Figure 4-17 shows that the output voltage, VOUT, of 24V has been successfully achieved. The plot shows that the output voltage is stable in the quiescent period as opposed to the start-up period where the voltage is building up to the required level.
The duration of the start-up time is dependent on the output capacitor, Co. The 220uF output capacitor used in this case has contributed to a rise time duration of around 90 ms seconds.
This section provides an analysis of the simulation waveform results compared to the analytically predicted results.
188.8.131.52 Voltage across the Diode (and the current through it)
The theoretically expected voltage across the diode, D1, is shown in Figure 4-18 where the voltages to the left (red) and to the right (green) of the diode and hence, the voltage drop across it is depicted. It can be deduced from the graph in the top left hand corner of this figure that the diode is forward biased when the output voltage, VOUT is less than the voltage at the pin 1 of the MC34063A, the switch collector voltage, VSAT.
Figure 4-18: The Diode, D1, is forward biased when the output voltage is less than saturation voltage.
When the output voltage, VOUT, falls below the inductor voltage the diode is reversed biased and a diode current, ID, does not flow through the inductor. The result of simulating the voltage across the diode and the current through it is presented in Figure 4-19. As expected a current (brown curve) flows through the diode when it is forward biased due to the output voltage, VOUT, falling below the saturation voltage, VSAT.
Figure 4-19: A plot of the output voltage, Vout, the saturation voltage, Vsat, and the current through the inductor, Id.
184.108.40.206 Inductor Current
The current through the inductor is shown in Figure 4-19. This figure shows the inductor current in both the start-up and quiescent periods. The scale of the plot does not allow us to fully appreciate the inductor current in relation to the charge and discharge current flowing through it.
Figure 4-20: PSPICE simulation of the current through the inductor in the MC34063A boost configuration.
However, theoretically we know that the plot should look like the curve in Figure 4-21. During the on-time period there should be a positive rate of change of current and a negative one during the off-time period. The analytically determined ratio of the discharge to charge current calculated previously in Section 3 is approx 4.96:1
Figure 4-21: Theoretical output of the inductor current. The maximum current through the inductor is Ipk. When this level of current flows though the inductor the switch switches off and the current decays as the magnetic field collapses.
If we zoom into the PSPICE simulation and take two snapshots in time of the inductor current, one during the start-up period and the other during the quiescent period then we derive the plots shown in Figures 4-22 and 4-23 respectively.
Figure 4-22: The inductor current during the start-up period.
If we calculate the ratio of on-time to off-time or (B-A)/(C-B) in the plot above then the ratio is approximately 44.7/9.4 or 4.8:1 which is very close to our analytically determined value of 4.96:1. This should give us a high degree of confidence that our circuit is driving the MC34063A in the expected manner.
The appearance of the inductor current curve in the quiescent period appears to be slightly different from the one in the start-up period. In the quiescent period there are periods when no current flows through the inductor at all. I am not quite sure what phenomenon this represents; this could be the period of time when the output current to the backlight is supplied entirely by the output capacitor, CO. It could be an inaccuracy in the MC34063A PSIPCE model or the PSPICE application program itself.
It might be worth taking an actual measurement when the prototype is built to determine whether this waveform of inductor current during the quiescent period is repeated.
Figure 4-23: The inductor current during the quiescent period.
220.127.116.11 Output Voltage Ripple
Figure 4-24, below, shows the theoretically expected waveform of the ripple voltage while Figure 4-25 shows the output voltage transients when a more detailed snapshot of the output voltage, VOUT, is taken.
Figure 4-24: Theoretically predicted capacitor, co, ripple voltage.
Figure 4-25: PSPICE simulation of the output voltage. The spikes are due to the output voltage ripple.
As discussed in section 3? The ripple voltage is primarily due to the Equivalent Series Resistance, ESR, of the output capacitor. The effect of ESR can be reduced by using two similar capacitors in parallel. The peak-to-peak ripple voltage in Figure 4-26 is approximately 50mV. This does not really compare to our analytically predicted value of about 96mV.
The discrepancy in the values may be due to the fact that the capacitor used in the simulation model has an ESR of 0.07 Ohms rather than the 0.1 Ohms value used during the analytical calculations. It could also be due to the fact that the PSPICE model might only have modelled the effect of the capacitor’s ESR when determining the ripple voltage and not the ripple effect due to the MC34063A. Further investigation, at a later date, might be necessary to get the bottom of the discrepancy.
Figure 4-26: Output voltage transients at a higher resolution.
18.104.22.168 Capacitor Co, Current
The waveform of the ripple current through the output capacitor is shown in Figure 4-28 and looks remarkably similar to the theoretically predicted waveform in Figure 4-27. A fantastic result!
Figure 4-27: Theoretically predicted capacitor, Co, current waveform.
Figure 4-28: PSPICE simulation of the capacitor, Co, current.
4.6 PCB Layout Considerations
Using a switching regulator usually means that the PCB layout engineer has to deal with voltage transients arising from the switching circuitry. This means that extreme care must be taken in how component devices are placed relative to each other and relative to other components on the prototype board. In particular attention must be paid to the effects of ground bounce and current loops.
4.6.1 Ground Bounce Analysis
Accounting for an imperfect ground is critical in successfully laying out any switching regulator power supply design. We can see why by examining the circuit in Figure 4-29 where the ground connection to the output capacitor and a digital IC1 are through a connected PCB trace signal. Assume that the length of this ground trace from the output capacitor CO to ground is approximately 25.4mm.
Figure 4-29 : Improperly laid out circuit used to demonstrate the effects of ground bounce.
Naively, in Figure 4-29 not being aware of any PCB layout techniques means that the impedance of the 25.4mm trace has not been taken into consideration. Now, as there is a switching current flowing through the capacitor, CO it means that not only will there be a voltage drop across the resistance of the PCB trace but there will also be an induced voltage due to the changing current through the trace inductance. This induced voltage will vary in time with the changing current.
Figure 4-30: Formulas used to determine the impedance of a PCB trace.
By using the formulas in Figure 4-30 with a PCB trace length of 25.4mm (L) a trace resistance of approximately 719.67m Ohms and an inductance of approximately 25.11nH will result when the PCB trace is about 0.8mm wide (d) and 36um thick (e).
Figure 4-31: Modelling the effect of ground bounce in PSPICE of the MC34063A switching regulator circuit in a step-up configuration.
The effect of the PCB trace impedance can be modelled in PSPICE as part of the switching regulator design by using the circuit in Figure 4-31. The difference in potential between the 25.4mm PCB trace and the actual ground can be measured by using a comparator as seen in the figure.
Figure 4-32: The ground bounce waveform as a result of the 25.4mm PCB trace in the circuit configuration above.
The result of the comparison of the ground and PCB trace potential is shown in Figure 4-32 where in this particular example the potential difference is between 375mV and 800mV. If this difference in ground potential couples onto the output of IC1 then depending on the frequency and phase of the coupled voltage undershoot, overshoot or ringing can occur on the voltage seen by the other device, IC2.
Figure 4-33: The star ground technique used in low-frequency circuit design. Because the two ICs have grounds that are at the same potential the risk of overshoot and undershoot are greatly reduced.
A technique to avoid ground bounce from ruining a good design, in relatively low frequency circuits, is to layout the design using the so-called star ground technique as can be seen Figure 4-33.
4.6.2 Current Loops
Previously, in section 3 above, it was noted that a time varying current flowing through an inductor creates a magnetic field. This magnetic field is proportional to the current, according to the Biot-Savart law. When laying out a PCB it is important to minimize the size of any current loops to reduce the magnetic field associated with them. Minimization is beneficial since the magnitude of the magnetic flux is proportional to the size of the current loops.
Minimizing the size of the current loops can help reduce cross coupling between PCB signal traces. It can also help to reduce the coupling between the lines of magnetic flux and nearby components as well as nearby electronic equipment. The overall aim of minimizing the current loops is to reduce the amount Electromagnetic Interference (EMI) emanating from the switching circuitry.
Some techniques used to minimize EMI and improve the efficiency of the circuit include careful placement of components and using short and wide traces to reduce the inductance of the PCB traces as can be seen in Figures 4-34 and 4-35.
Figure 4 -34: Careful placement of components can minimize the lines of magnetic flux emanating from large current loops and hence the associated EMI.
4.6.3 Evaluation Board Layout
It is also worth looking at the layout techniques used on the evaluation boards of similar boost switching regulators to appreciate the efficient layout of the components. Figure 4-35 for example is the layout of the NCP3064 boost demo board taken from the OnSemi website. Notice the different thicknesses of the PCB traces and how this impacts on the trace impedance (remember that a wider PCB trace has less inductance!).
Figure 4-35: NCP3064 boost regulator demo evaluation board taken from the NCP3064 evaluation board datasheet.
Well, after this slight detour the new design of the backlight circuit is definitely better than the 5V potential difference that I described previously in Part III of the series. However, it has come at the cost of at least three weeks of extra work! This means that I have only partially completed the PCB design.
In any design project the design engineer is solely responsible for meeting deadlines. If this project was funded by a group of non-technical financiers the design engineer would have to make the decision of between showing the financiers a fully working design that partially drives the backlight delivered on time or a fully working design and backlight delivered a few weeks late.
It is always a tough decision to make but from my experience I would try and deliver something on time rather than nothing at all. If this means a partially working design then so be it, as long as your financiers are happy there will always be time to tackle the more complicated aspects of the design later on.
However, as this project is only a hobbyist one, the knowledge gained from the extra time spent has been well worth it. A compromise solution during the initial design phase could be to drive the backlight from an evaluation board.
In general the products of the major international companies last longer than the cheaper replicas, because they are better designed. However, better designed usually translates into better trained engineers who earn more money. This is one of the reasons why branded products are more expensive.
One of the nice things about hobby projects is that you can take as long as you like without the commercial pressure! The next part in the series will provide the details of the PCB design.