Silicon ChipCompact Fluorescent Lamp Driver - July 2000 SILICON CHIP
  1. Outer Front Cover
  2. Contents
  3. Publisher's Letter: 42V electrical systems in cars
  4. Feature: Say Bye-Bye To Your 12V Car Battery by Julian Edgar
  5. Project: A Home Satellite TV System by Garry Cratt & Ross Tester
  6. Project: A Moving Message Display by Atilla Aknar & Ross Tester
  7. Project: Compact Fluorescent Lamp Driver by John Clarke
  8. Project: El-Cheapo Musicians' Lead Tester by John Clarke
  9. Order Form
  10. Project: Li'l Powerhouse Switchmode Power Supply; Pt.2 by Peter Smith & Leo Simpson
  11. Review: Motech MT-4080A LCR Meter by Leo Simpson
  12. Product Showcase
  13. Review: ADEM Compac II Security System by Ross Tester
  14. Vintage Radio: The AWA P1 portable 11-inch B/W TV set by Rodney Champness
  15. Book Store
  16. Back Issues
  17. Notes & Errata
  18. Product Showcase
  19. Market Centre
  20. Outer Back Cover

This is only a preview of the July 2000 issue of Silicon Chip.

You can view 37 of the 96 pages in the full issue, including the advertisments.

For full access, purchase the issue for $10.00 or subscribe for access to the latest issues.

Items relevant to "A Moving Message Display":
  • DOS software for the Moving Message Display (Free)
Items relevant to "Compact Fluorescent Lamp Driver":
  • 12V CFL Driver PCB pattern (PDF download) [11107001] (Free)
  • 12V CFL Driver panel artwork (PDF download) (Free)
Items relevant to "El-Cheapo Musicians' Lead Tester":
  • El-Cheap Musicians' Lead Tester panel artwork (PDF download) (Free)
Items relevant to "Li'l Powerhouse Switchmode Power Supply; Pt.2":
  • Li'l PowerHouse Power Supply PCB pattern (PDF download) [04106001] (Free)
  • Li'l PowerHouse Power Supply panel artwork (PDF download) (Free)
Articles in this series:
  • Li'l Powerhouse Switchmode Power Supply; Pt.1 (June 2000)
  • Li'l Powerhouse Switchmode Power Supply; Pt.1 (June 2000)
  • Li'l Powerhouse Switchmode Power Supply; Pt.2 (July 2000)
  • Li'l Powerhouse Switchmode Power Supply; Pt.2 (July 2000)

Purchase a printed copy of this issue for $10.00.

Compact Fluorescent Lamp Driver Although our photo shows a single CFL in a loose bayonet fitting, the unit is designed to be wired as a permanent installation with fixed bayonet sockets. Compact fluorescent lamps (CFLs) are far more efficient than their incandescent counterparts. This inverter circuit is ideal for driving 240VAC CFLs from a 12V battery, up to a total load of about 40W. It can be used anywhere you require good lighting when there is no mains power, or as part of a solar-powered lighting system. By JOHN CLARKE O BTAINING GOOD LIGHTING from battery power has never been easy unless fluorescent lamps are used. This is because fluores­cent lamps produce far more light output than incandescent lamps for a given power input. In practice, this means that you can burn your lights for longer before the battery goes flat – up to five or six times longer, in fact! This “Compact Fluorescent Driver” is built into a sturdy metal case and is specifically designed to drive 30  Silicon Chip CFLs. It deliv­ers a 240V (340V peak) waveform that’s approximately sinusoidal in shape, which means that it’s capable of driving CFLs with a high power factor, as well as earlier designs with low power factor. Note that, in general, this circuit is not designed to power other mains-operated equipment. Background Fluorescent lighting which operates from 12VDC is not new and there are many commercial 12V fluorescent lights available which use the “long-tube” style of lamp. Of course, the tube itself doesn’t operate from 12VDC but rather via a DC-DC convert­er circuit built into the fitting. In fact, SILICON CHIP published a high-efficiency inverter for 18W and 36W fluorescent tubes in the November 1993 issue and this was designed to fit inside the lamp batten. Because of their large size and light dispersion character­istics, this type of lamp is mainly used for mounting on the ceiling of a caravan or for emergency lighting. However, they are somewhat less than ideal for camping or outdoor use because of their bulk and rather awkward shape. Compact fluorescent lamps on the other hand provide light­ing similar to a gas-powered camping lantern and are ideal for mounting on a picnic table, on the ground or even in a caravan ceiling. They will fit into standard mains-style bayonet or Edison screw (ES) fittings and consequently are easy to replace. They are typically rated at 9-25W which means that you can run up to four lamps using this circuit, depending on their rating. Physically, a CFL consists of a small folded fluorescent lamp attached to a base which contains the driver (inverter) circuit. The internal driver circuit for the tube is basically an “electronic ballast”. This typically produces a high-frequency drive waveform and includes inbuilt current limiting for the tube. The high frequency prevents flicker and also improves the light output. What this all means is that the CFL’s internal driver circuit assumes that it will be supplied with a 50Hz sinewave at 240VAC, which is close to 340V peak. Any or all of these features of the supply may be used by the electronic ballast to drive the fluorescent tube. Any external circuit which drives the CFL from a different supply (such as 12VDC) must take these requirements into consideration. Fig.1: a typical power supply circuit as used in many older compact fluorescent lamps. Its main disadvantage is the fact that it draws current over only a small part of each mains cycle. Note that the tube is powered from 340V DC. Fig.2: this circuit is used in the newer CFLs and offers a much im­proved power factor compared to the circuit shown in Fig.1. The tube is no longer powered from 340V DC but from the full-wave rectified mains waveform. Basic circuits Fig.1 shows a typical power supply circuit as used for some CFL electronic ballasts. It uses a low value resistor in series with a full-wave bridge rectifier. Filter capacitor C1 charges up to the peak voltage of the 240VAC waveform to give a 340VDC supply which is then applied to the electronic ballast circuit. The ballast circuit in turn drives the fluorescent tube. Based on this power supply, you would expect that CFLs can also be driven from a 340VDC supply. After all, if 340VDC is applied to the Active (A) and Neutral (N) terminals, the voltage across C1 would still be 340VDC (neglecting the voltage drops across the diodes in the bridge rectifier) and the electronic ballast would be none the wiser. This may be true for those CFLs that use this particular power supply but not for all CFLs. Some CFLs derive their supply in a different manner and cannot be directly powered from 340V DC, as we shall see. Power factor correction A big disadvantage of the circuit shown in Fig.1 is that it only draws current from the 240VAC mains near the crest of the waveform, where the voltage is at or near its peak of about 340V. This occurs because capacitor C1 is topped up to the peak mains voltage and the rectifier diodes do not conduct below this peak voltage. So this is a rather crude supply because current is only drawn for a brief interval during each mains half-cycle. What’s more, the current pulses will be quite high in value. This leads to poor power utilisation and results in considerable power losses, which reduces the efficiency. A better approach is to use a circuit that draws current over the greater WARNING! The output voltage produced by this CFL Driver circuit is potentially lethal! Do not build it unless you are experienced and know what you are doing. In particular, make sure that you do not touch the output leads and check that they are securely connected to an ap­proved mains lighting socket in a fixed installation before connecting 12V DC power. Treat these wires as though they are at mains potential. Finally, be sure to keep your hands away from the PC board components associated with the output terminals, diodes D7-D10 and the 470kΩ and 1MΩ resistors in the feedback path. part of the waveform and this is what CFL de­signers are now beginning to do. The newer electronic ballasts effectively draw current over most of the mains waveform, thus reducing the peak current and also improving the power factor. This also creates less of a problem for the supply utilities. By the way, the term “power factor” in this case refers to the extent to which the sinusoidal mains waveform is utilised. A low power factor of around 0.5 means that current is only drawn over a part of the mains waveform while a high power factor (greater than 0.95) means that the waveform is almost fully utilised. Fig.2 shows how these “improved” electronic ballasts derive their supply from the mains to improve the power factor. The circuit looks similar to Fig.1, with C1 charging to 340VDC as before. However, there is a major difference and that is that the fluorescent tube is no longer simply powered from 340VDC but from the full-wave rectified mains waveform. The supply for the electronic ballast is still derived from the 340VDC across C1 but this is now isolated from the rectifier output using diode D1. An LC filter on the mains input prevents the high-frequency switching noise produced by the electronic ballast from being fed back into the mains supply. JULY 2000  31 Fig.3 (left): this scope shot shows the mains waveform, along with the current waveform for a non-power factor corrected CFL. Note that the current waveform is quite “peaky”. The flattening at the top of the mains waveform is not caused by the circuit but is present in many industrial areas due to switchmode power supplies in PCs and gas discharge lighting. Fig.4 at right shows the mains wave-form and the corresponding current waveform from a power factor corrected CFL. Note how the current is far less “peaky” than before. Using this circuit, the current drawn by the fluorescent tube is much more sinusoidal in shape compared to that from Fig.1 (see Figs.3 & 4). The peak current is substantial­ ly reduced and the current peak lasts much longer. By the way, you might expect that this circuit would pro­duce a flickering effect similar to that produced by standard fluorescent long-tube lamps which are driven from the mains. This is because the current through the lamp is varying at a 100Hz rate. In fact, the light output is modulated by a small amount but the effect is not noticeable due to the use of longpersist­ence phosphors and because the tube is driven at a high frequency by the electronic ballast. Now for the million-dollar question: What would happen if we were to drive a CFL which uses the circuit of Fig.2 from a 340VDC supply? Well, initially, not much. The fluor­ escent tube would be quite bright but would otherwise appear to be operating correct­ly. In practice, however, it would be severely over-driven. That’s because it is being driven from 340VDC rather than a 340V peak fullwave rectified waveform which has a DC equivalent (RMS) of 240V. As a result, the life expectancy of the CFL would be severely compromised. 32  Silicon Chip So how do we produce a circuit which will comfortably drive all types of CFLs? The obvious answer is to use a sinewave in­verter which produces a 50Hz 240VAC waveform. In this way, all CFLs would be powered correctly. However, this type of inverter is fairly complicated and requires a fair number of power devices to produce a clean 240V sinewave. Fortunately, a pure sinewave inverter is overkill and we don’t need to do this. The alternative approach is to take into account the fact that CFLs actually fullwave rectify the mains waveform before doing anything else. If we use an inverter that provides a fullwave rectified sinewave output rather than a genuine sinewave, the CFL circuit would be none the wiser. And that’s just what we have done here. Of course, only two diodes in the CFL’s fullwave bridge rectifier are now used, since we now have a pulsating DC supply rather than AC, but this is of no consequence. Block diagram Fig.5 shows the general arrangement of our CFL driver circuit. It’s powered from a 12V battery which supplies a switch­mode controller IC, the push-pull outputs of the controller in turn driving transformer T1 via buffer stages and Mosfets Q1 & Q2. In operation, the Mosfets switch the primary windings of transformer T1 in an alternate fashion at a high frequency and the resulting waveform is stepped up to a higher voltage in the secondary winding. The secondary output of T1 is then full-wave rectified and filtered to produce pulsating DC with a peak vol­tage of about 340V. The exact voltage at the output is controlled via the feed­back from the A+ terminal to the switchmode controller, in this case via a “half sinewave shaper” circuit. In a conventional switchmode circuit, this feedback is simply a voltage divider which is set to provide the required DC output vol­tage. It adjusts the pulse width applied to the transformer so that the output voltage is maintained regardless of variations in load current or input voltage. In this circuit, however, we have to produce a half sinew­ ave shape. This is done by rapidly switching in different voltage divider resistors in sequence to simulate the half sinewave shape. This job is performed by the “shaper” circuit. Circuit details Refer now to Fig.6 for the final circuit details. It uses just six lowcost ICs, two Mosfets, a transformer and a handful of transistors, diodes, resistors and capacitors. At the heart of the circuit is a TL494 pulse width modula­tion (PWM) controller (IC1). It contains a sawtooth oscillator, two error amplifiers and a PWM comparator. Also crammed onto the chip are a “dead-time” control comparator, a 5V reference and output control options for push-pull or single-ended operation. The RC oscillator components at pins 5 & 6 set the operat­ing frequency to about 50kHz. The PWM outputs from the error amplifiers appear at pins 9 & 10 (E1 & E2) and drive paralleled buffer stages IC2d-IC2f and IC2a-IC2c respectively. In turn, these drive Mosfets Q1 & Q2. Q1 & Q2 drive the centre-tapped primary winding of trans­former T1 in push-pull mode; ie, when Q1 is on, Q2 is off and vice versa. As shown, the centre tap of the transformer connects to the +12V rail, while each side of the primary winding is connected to the drain of its corresponding Mosfet. When Q1 is on, 12V is applied across the top half of the primary winding. Because of transformer action, the lower half of the primary winding also has 12V impressed across it which means that Q2’s drain is at 24V. Similarly, when Q2 is on, the bottom of the transformer primary goes to 0V and the top goes to 24V. The resulting 24V peak-to-peak waveform on the primary is then stepped up by the secondary winding. High speed diodes D7-D10 rectify the resulting AC output from the secondary and this is then filtered using two paralleled 0.1µF 250VAC capacitors. Note that Mosfets Q1 & Q2 are protected from over-voltage excursions on the drains using 16V zener diodes ZD1 & ZD2, togeth­ er with series diodes D1 & D2. The series diodes prevent each of the zener diodes from conducting when its associated Mosfet is switched on. In addition, any reverse voltages that would otherwise be applied to Parts List 1 PC board, code 11107001, 143 x 112mm (302 holes) 1 diecast box, 171 x 121 x 55mm 1 front-panel label, 167 x 117mm 1 M205 fuseholder 1 M205 5A fuse 1 25-28mm diameter iron powdered toroid (L1) (Jaycar LO-1244 or similar) 1 E30 transformer assembly (T1) 1 SPDT 10A toggle switch with integral LED and resistor (S1) 2 M3 x 10mm screws 2 M3 nuts 2 M3 flat washers 1 M4 sized solder lug 2 transistor insulating bushes 2 TO-220 silicone insulating washers 2 cordgrip grommets 3 100mm cable ties 1 200mm cable tie to secure ferrite cores on T1 1 2m length of heavy duty automotive figure-8 wire 1 2m length of 240VAC 7.5A figure-8 wire 1 12m length of 0.25mm ENCU wire 1 1200mm length of 1mm ENCU wire 1 500mm length of 0.8mm tinned copper wire 4 PC stakes Semiconductors 1 TL494 switchmode controller (IC1) 1 4050 hex buffer (IC2) 1 7555, LMC555CN, TLC555, CMOS timer (IC3) 1 4029 4-bit counter (IC4) 2 4051 8-channel analog multiplexers (IC5, IC6) 2 MTP3055 60V Mosfets (Q1, Q2) 2 BC547 NPN transistors (Q3, Q4) 2 16V 1W zener diodes (ZD1, ZD2) 5 1N914, 1N4148 diodes (D1-D4, D6) 1 1N4004 1A diode (D5) 4 1N4936 500V high-speed diodes (D7-D10) Capacitors 2 4700µF 16VW PC electrolytic 2 10µF 16VW PC electrolytic 2 0.1µF 250VAC MKT X-Class 5 0.1µF MKT polyester 1 .039µF MKT polyester 1 .001µF MKT polyester 1 560pF ceramic 2 220pF ceramic Resistors (0.25W, 1%) 2 1MΩ 1 12kΩ 1 470kΩ 4 10kΩ 1 270kΩ 2 4.7kΩ 1 75kΩ 1 3.3kΩ 1 47kΩ 1 3kΩ 1 33kΩ 2 2.2kΩ 1 27kΩ 2 1kΩ 1 24kΩ 1 470Ω 4 22kΩ 2 10Ω Miscellaneous CFLs, bayonet or ES lamp holders. Fig.5: the CFL Driver uses a switchmode controller to drive Mosfets Q1 & Q2. These in turn drive centre-tapped transformer T1 which steps up the voltage across the primary. The transformer output is then rectified and fed to the CFL. The half sinewave shaper circuit in the feedback path ensures that the output waveform approximates a sinewave. JULY 2000  33 MAIN FEATURES • • • • • • Suitable for driving compact fluorescent lamps (CFLs). Can drive loads up to 40W for CFLs with a power factor of 0.95. Can drive loads up to 33W for CFLs with a low power factor. Output voltage (and thus lamp brilliance) remains constant for 11-14.4V DC input. Reverse polarity protection. Built-in electric shock protection between high voltage output and battery terminals. the gates of Q1 & Q2 due to capacitive effects are shunted to ground via diodes D3 & D4. Feedback The feedback signal for the PWM controller (IC1) is derived from the high-voltage output at the A+ terminal. This is sampled using a voltage divider consisting of series 470kΩ and 1MΩ resis­tors and a resistance value switched in by the 16-step half sinewave shaper circuit. The resulting feedback signal is then applied to the pin 16 input of IC1. Pin 16 is the non-inverting input of one of the internal error amplifiers in IC1. A 1MΩ feedback resistor between pins 3 & 15 and the 4.7kΩ resistor between pins 3 and 14 (VREF = 5V) sets the gain of this error amplifier to 213. Also included in the negative feedback loop is a 1kΩ resistor and series 0.1µF capaci­tor and these set the low frequency rolloff for the error ampli­fier. In operation, IC1 continually adjusts the pulse width drive to the Mosfets so that the voltage on pin 16 is maintained at 5V. The duty cycle and thus the output voltage on the A+ terminal at any instant depends on the resistor values switched in by the sinewave shaper circuit to form the bottom leg of the voltage divider in the feedback path. For example, if a 22kΩ resistor is switched in, the ratio is 22kΩ divided by (470kΩ + 1MΩ + 22kΩ), or .0147. As a result, the A+ output will be at 5V/.0147 = 340V. Lower output volt­ages are selected by switching in higher value resistors. Half-sinewave generator In operation, the shaper circuit sequentially switches in various resistor values to give an approximate half 34  Silicon Chip sinewave at the A+ output. IC3, IC4, IC5 & IC6 make up the shaper circuit. IC3 is a CMOS 7555 timer which produces a 1.6kHz square wave at its pin 3 output, as set by the RC timing components on pins 2 & 6. This signal is applied to the clock input of IC4, a 4029B 4-bit coun­ter, via a 2.2kΩ resistor. This resistor and its associated 220pF capacitor increase the risetime of the pin 3 output of IC3 to suit the operation of the counter. The Q1-Q2 outputs of IC4 are applied to the A, B & C inputs respectively of both IC5 & IC6. These ICs are basically single-pole 8-way switches, with the position of the switch selected by the count value on the A, B & C inputs. As IC4 counts up from 0 to 7, the Y0-Y7 outputs of IC5 & IC6 are each selected in succession and so different resistor values are sequentially connected to the common terminal at pin 3 (and thus to pin 16 of IC1). As a result, the divider ratio is constantly being altered and this means that the feedback voltage also alters each time a different resistor is selected. As shown on Fig.6, the Y0-Y7 outputs of IC5 and IC6 are connected together in reverse order; ie, Y0 of IC5 goes to Y7 of IC6, Y1 goes to Y6, Y2 goes to Y5 and so on. The reason for this is that we use IC5 to progressively select lower-value resistors (starting at 270kΩ) for the rising part of the output waveform and then use IC6 to select the resistors in reverse order for the falling part of the waveform. In this way, IC5 and IC6 use the same set of resistors. They just use them in reverse order to each other! The Q3 output from IC4 is used to decide whether IC5 or IC6 is selected. This output connects directly to the inhibit input (INH, pin 6) of IC5 and also drives transistor Q3 via a 10kΩ resistor. Transistor Q3 functions as an inverter and controls the inhibit input of IC6. In practice, the inhibit input must be low for the IC to be selected. As a result, IC5 is selected while IC4 counts from 0-8, while IC6 is selected for the 8-16 count, after which the cycle repeats. Dead-time So how do we stop the circuit from producing glitches in the output each time IC5 or IC6 selects a different voltage divider resistor? The answer to this is transistor Q4 which is connected between the dead-time (DT) input of IC1 (pin 4) and the VREF terminal (pin 14). This transistor is driven by the pin 3 output of IC3 via a 3.3kΩ resistor and a 220pF capacitor. Each time pin 3 goes high, Q4’s base goes high for about 2µs (as set by the 22kΩ resistor to ground) and so Q4 briefly turns on and connects the DT input to +5V (ie, to VREF). This effectively shuts the PWM controller down for 2µs on each clock pulse, which is ample time for IC5 or IC6 to select the next resistor value. When Q4 switches off at the end of the 2µs period, a 4.7kΩ resistor pulls the DT input low and the PWM controller begins operating again. High voltage protection The high voltage output at the A+ terminal is potentially lethal since it produces 240V RMS and can provide well over 150mA of current. For this reason, it is important that you don’t simultaneously come into contact with the A+ and N terminals. Any contact between a battery terminal and the N terminal will not cause a shock since the N terminal is tied to ground. However, the A+ terminal could cause an electric shock if you connect yourself between it and a battery terminal. As a safeguard, we have added a leakage-to-ground detector circuit Fig.6 (right): the final circuit uses IC1 to drive Mosfets Q1 & Q2 via parallel buffer stages . IC3-IC6 form the half sinewave shaper circuit. It constantly changes the feedback so that IC1 varies its PWM output to produce a half sinewave shape. JULY 2000  35 Fig.7: install the parts on the PC board and complete the wiring as shown here. Fig.8: this diagram shows the mounting details for Mosfets Q1 & Q2. Use your multi-meter to check that the device tabs are correctly isolated from the case. 36  Silicon Chip which will switch off the PWM controller if the current between the A+ terminal and one of the battery terminals exceeds 224µA. Let’s see how this circuit works. As shown, the N terminal is tied to ground via a 1kΩ resis­tor and a parallel 0.1µF capacitor which is used as a filter. Normally, this terminal will be at ground unless there is leakage between the A+ terminal and ground (or battery +). If there is leakage, the voltage across the 1kΩ resistor rises by 100mV for every 100µA of leakage current. This voltage is monitored by the pin 1 input of IC1 which is the non-inverting input to the second error amplifier. The in­verting input at pin 2 is connected to a voltage divider across the 5V reference and sits at 224mV. If the voltage at pin 1 reaches this 224mV limit, the PWM controller shuts down the high voltage step-up operation and limits the current to 224µA. Power Power for the circuit is derived from a 12V DC source (eg, a battery). This is applied to the ICs via reverse-polarity protection diode D5 and to the centre tap of transformer T1 via inductor L1. Two 4700µF capacitors decouple the supply for the transformer and are bypassed with a 0.1µF capacitor. Reverse polarity protection for the Mosfets is provided by fuse F1. If the supply is connected the wrong way around, the internal drain-source protection diodes in the Mosfets conduct heavily and the fuse blows before any damage occurs. Building it The CFL Driver circuit is built on a PC board coded 11107001 and measuring 143 x 112mm. This fits inside a The PC board fits neatly into a standard metal diecast case which also serves as a heatsink for Q1 and Q2. Be sure to use 240VAC-rated cable for the output lead and make sure that this has been correctly terminated before applying power. standard diecast case measuring 171 x 121 x 55mm. Alternatively, the PC board could be fitted into a plastic case with 6021-type flag heatsinks (29.5 x 25 x 12.5mm) used for each Mosfet. The PC board includes solder mounting points for these heatsinks if the die­cast case is not used. Begin construction by checking the PC board for shorts between tracks and for any breaks in the copper pattern. Also, check to ensure that the hole sizes are correct. You will need 1mm holes for the transformer pins, diode D5 and the zener diodes. Next, check that the PC board fits neatly into the case. The PC pattern (Fig.11) shows the profile required. In particular, the half-moon “cutouts” (to clear the central mounting posts) and the small rectangular cutouts (to clear internal ribs) may need filing to shape so that the board fits. It is also necessary to round the corners of the board as shown, to clear the corner posts of the case. Fig.7 shows the wiring details. Begin the PC board assembly by installing the links and resistors. Table 2 shows the resistor colour codes but you should also use a digital multi­meter to check each value, just to be sure. The ICs can be mounted next, taking care with their orien­tation. Make sure also that each IC is placed in its correct position. Now for the capacitors. The electrolytic types are polar­ised and must be oriented with the polarity shown. The MKT and ceramic types usually include a value code and these can be deciphered using Table 1. Table 1: Capacitor Codes       Value IEC code EIA code 0.1µF  100n  104 .039µF   39n  393 .001µF   1n0  102 560pF   560p   561 220pF   220p   221 Fig.9: here are the winding details for transformer T1. The secondary is wound on first, with each successive layer covered with insulating tape. The primary is then bifilar wound (ie, two wires at once) over the secondary. JULY 2000  37 What’s Inside A Compact Fluorescent Lamp? While CFLs are a throw away item once the fluorescent tube has burnt out, they have relatively complex circuit, as shown in the photo and Fig.10. This is a typical circuit for an electronic ballast without the power supply (ie, rectifier diodes, filter capacitor, etc). The circuit operates in two separate modes, one to start the tube and the second mode for normal running. There are two Mosfets (Q1 & Q2), transformer T1 and a number of associated components which make up an oscillator. The fluorescent tube is driven via inductor L1 and winding N1 of the transformer. T1 also drives the gates of Q1 & Q2 via windings N2 & N3 which are connected in antiphase. Tube starting When power is first applied, the .022µF capacitor connected to Diac 1 charges via the 560kΩ resistor. When the voltage reach­es about 30V, the Diac fires (breaks down) and discharges the ca­pacitor voltage into the gate of Q2. Zener diode ZD2 protects the gate from over-voltage. Mosfet Q2 is now switched on and current flows from the positive supply via the .047µF capacitor, the fluorescent tube top filament, the .0033µF capacitor, the second tube filament, inductor L1 and transformer T1’s N1 winding. This current flow in N1 then applies gate drive to Q1 via N2 and switches off gate drive to Q2 via N3 due to the antiphase connection of this wind­ing. If oscillation doesn’t occur, the process starts all over again with the .022µF capacitor charging again to fire the Diac to turn on Q2. When oscillation does occur, Mosfets Q1 and Q2 rapidly switch on and off in alternate fashion. The frequency of operation is set by the combined inductance of L1 and the N1 winding, together with the .0033µF capacitor across the tube. The startup circuit comprising the .022µF capacitor and the Diac is now prevented from operating by diode D1. This diode discharges the 38  Silicon Chip Fig.10: typical circuit for a CFL electronic ballast, minus the power supply components. It’s basically an oscillator circuit that operates in two different modes – one for starting and the other for normal running. .022µF capacitor every time Q2 is switched on. The oscillator current now flows through the filaments of the fluorescent tube and allows the normal mercury discharge to take place. This means that the fluorescent tube will light up. When this happens, the .0033µF capacitor is effectively shunted by the mercury discharge and the voltage across the tube is now at about 100V peak. Normal running The frequency of oscillation is now determined by the prop­erties of the core used for transformer T1. As the current builds up in winding N1, the core begins to saturate. When this happens, the flux in the core stops changing and gate drive to Q1 or Q2 ceases. The flux now collapses to drive the opposite Mosfet and this process continues to maintain oscillation. The current through the tube is limited by the current at which T1’s core saturates and by L1’s inductance. The two 10Ω resistors, together with zener diodes ZD1 & ZD2, limit the gate drive to Q1 & Q2, while the .0022µF capacitor at the cathode of D1 forms a snubber network to suppress commuta­tion in the opposing Mosfet at switch on. This considerably reduces the switching losses in each Mos­ fet. The 330kΩ resistor in parallel with this capacitor keeps diode D1 reverse biased at start-up. Finally, the .047µF capacitor in series with one of the tube filaments ensures that the tube is driven by AC. This prev­ents mercury migration to the tube ends which would cause black­ ening and short­en the tube life. The diodes, zener diodes and transistors can now all be installed, followed by Mosfets Q1 & Q2. The latter should be mounted at full lead length, with only 1-2mm of each pin protrud­ing below the PC board to allow for soldering. This enables their metal tabs to be bolted to the side of the case later on. Inductor L1 can now be wound and installed. It comprises a 25-28mm iron powdered toroid with 20 turns of 1mm enamelled copper wire wound around it. The wiring diagram and photographs show how it is wound. Keep each turn tight around the toroid and space the windings evenly. Clean and tin the ends of the winding before mounting it on the PC board. After mounting, the toroid is secured using two plastic cable ties which are fed through adjacent holes in the PC board. Winding the transformer Fig.9 shows the winding details for transformer T1. Begin by soldering one end of a 12-metre length of 0.25mm enamelled copper wire to pin 4, then wind the turns on neatly side-by-side. Wrap a layer of insulating tape around each layer as it is com­pleted before winding on the next layer. After completing 200 turns, terminate the wire at pin 7 and secure the windings with another layer of insulating tape. The centre-tapped primary wind­ ings are wound together (ie, bifilar) Fig.11: check your PC board against this full-size etching pattern before mounting any of the parts. Table 2: Resistor Colour Codes  No.   2   1   1   1   1   1   1   1   4   1   4   2   1   1   2   2   1   2 Value 1MΩ 470kΩ 270kΩ 75kΩ 47kΩ 33kΩ 27kΩ 24kΩ 22kΩ 12kΩ 10kΩ 4.7kΩ 3.3kΩ 3kΩ 2.2kΩ 1kΩ 470Ω 10Ω 4-Band Code (1%) brown black green brown yellow violet yellow brown red violet yellow brown violet green orange brown yellow violet orange brown orange orange orange brown red violet orange brown red yellow orange brown red red orange brown brown red orange brown brown black orange brown yellow violet red brown orange orange red brown orange black red brown red red red brown brown black red brown yellow violet brown brown brown black black brown 5-Band Code (1%) brown black black yellow brown yellow violet black orange brown red violet black orange brown violet green black red brown yellow violet black red brown orange orange black red brown red violet black red brown red yellow black red brown red red black red brown brown red black red brown brown black black red brown yellow violet black brown brown orange orange black brown brown orange black black brown brown red red black brown brown brown black black brown brown yellow violet black black brown brown black black gold brown JULY 2000  39 Fig.12: this is the front panel, reproduced here two-thirds actual size. A full-size reproduction can be obtained by scanning it at 150% on a flatbed scanner or by enlarging it on a photostat machine. using 1mm enamelled copper wire terminated on pins 1 & 2 and finishing on pins 10 and 9 respectively, as shown. Make sure that there are 6-turns for each winding. Finish with a layer of insulating tape. The transformer can now be completed by inserting each core half of the transformer into the bobbin and clamping them with clips or a 200mm cable tie. This done, the transformer can be mounted on the PC board with pin 1 orientated as shown on Fig.7. Finally, complete the board assembly by installing four PC stakes at the supply input and A+ and N terminals. Final assembly Now for the final assembly. First, temporarily fit the PC board into the case and mark out the mounting holes for Mosfets Q1 & Q2. This done, remove the board and drill the holes for these, taking care to remove any metal swarf with an oversize drill. You also need to drill holes in the case ends for the input and output cordgrip grommets and for the fuse­ holder. A hole is also required in the lid for the power switch. The on/off switch mounts on the case lid, adjacent to the 12V DC supply cable and the fuse. An integral LED acts as a power on/off indicator. 40  Silicon Chip Once all the holes have been drilled, secure the PC board to the corner pillars of the case using the supplied screws. Note that a solder lug must be placed under one of these screws – this solders to an adjacent PC stake and is used to earth the negative supply rail to the case (see Fig.7). The two Mosfets can now be bolted to the side of the case. First, check that the mounting areas are perfectly smooth and free of metal swarf, then mount each device using a TO-220 insu­ lating kit as shown in Fig.8. After each device is mounted, use a multi­meter to check that its metal tab is electrical­ly isolated from the case. If the meter indicates a short, the device will have to be removed and the cause of the problem determined. Finally, wire up the connections to the fuse, switch and PC board as shown using automotive wire for the 12V side and 240VAC rated cable for the output. This 240VAC output cable can then be connected into one or more bayonet or ES lamp holders. Make sure that the output cable is actually connected to a socket, since the wires should be treated as you would any mains outlet. The voltage produced could prove fatal if you are care­less enough to connect yourself across the output leads while the unit is running. Testing Before doing anything, check that the output leads have been correctly terminated. This done, connect a 12V DC supply (rated at 1A or more) and check that the switch LED lights when the switch is “on”. If the LED doesn’t light, check that you have installed the 5A fuse in the fuseholder. Now check the supply rails to the ICs. There should be 11.5V on pin 12 of IC1, pin 1 of IC2, pin 8 of IC3 and pin 16 of IC4, IC5 & IC6. Next, carefully check the output voltage across the PC stakes on the board, using a multimeter set to measure up to 340V DC. Assuming no load is connected, the meter should indicate a value close to 340V DC (not 240V) due to the storage effect of the capacitors across the output. For the final test, you will need a 12V lead-acid battery capable of supplying several amps. Plug in a load such as a 15W 240V filament lamp or CFL and check that the output voltage SC is now around 240V DC.