Fullscreen HTML5Med Res (??MB)HTML4

Ideal Bridge Rectifier Project by Phil Prosser Background image source: https://unsplash.com/photos/a-bunch-of-speakers-that-are-sitting-on-a-table-YlSfH5XDAto for Power Amplifiers If you are making a dual-rail power supply and need high efficiency and/or low dissipation, a conventional bridge rectifier has distinct limitations. Audio power amplifiers and dual-rail power supplies are good examples of where this circuit can provide significant benefits. S emiconductor diode rectifiers have been the default choice for converting AC to DC since the 1960s. They are inexpensive, robust and relatively efficient, at least compared to what came before them (diode valves, selenium rectifiers and such). However, these days you can do a lot better. A rectifier using active devices can be much more efficient, meaning you get a higher voltage output with less power wasted as heat. That also means you don’t have to worry so much about getting the heat out of the device. We previously presented an ideal diode bridge based on the Linear Technology (now Analog Devices) LT4320 IC in December 2023 (siliconchip.au/ Article/16043). A little later, in the September 2024 issue, we published a discrete version, which sought to avoid the high cost of that IC and had some other advantages (siliconchip. au/Article/16580). The LT4320 chip controls four N-channel Mosfets, which allows for a high load current, but the maximum voltage you can rectify is limited. Also, it won’t work with a centre-tapped transformer, so if you want to generate split (positive and negative) rails, you need two of these relatively expensive devices. 66 Silicon Chip The discrete version used P-channel Mosfets on the high side to simplify the design. While not limiting the maximum voltage, such devices with high voltage and current ratings, plus the low on-resistances required for good performance, are scarce and expensive. One application where people sought to use these circuits was in the power supply of high-power audio amplifiers (eg, 200W+). The LT4320 design could be used, but as mentioned above, it would demand the use of two stacked LT4320-based rectifiers. » Low-loss rectification of centretapped transformer secondaries to produce split rails » Low voltage loss and low dissipation compared to a standard diode rectifier bridge » Four board versions: SMD with TO-252 (D2PAK) or TO-220 Mosfets; through-hole with TO-220 or TO-247 Mosfets » From 5-0-5V AC in (~±7V DC out) to 50-0-50V AC in (~±72V DC out) » Load currents up to 20A continuous, 100A+ inrush » Handles transformers up to 500VA » Works with large filter capacitor banks The later Discrete Ideal Bridge Rectifier we published has no such limitation, but to increase the voltage rating would require horribly expensive (if available) Mosfets, which made that an unattractive approach. This project gives you the best of both worlds as it can handle high voltages and a split supply without needing an expensive IC or P-channel Mosfets. Other applications where an ideal bridge rectifier comes into its own are in DC supplies where high continuous currents are drawn. In our Dual Hybrid Power Supply (February & March 2022; siliconchip.au/ Series/377) with a 5A continuous load, the bridge was dissipating about 10W, which demanded that the bridge be mounted to the heatsink. An Ideal Bridge Rectifier brings this dissipation down to just a couple of watts at most. A third application is where you are seeking to efficiently harvest energy, for example, from a wind turbine or a generator. In that case, you want to avoid the loss of the forward diode drop. The principal elements of this circuit are applicable in that use case too. To show the difference between a conventional bridge and our Ideal Bridge rectifier, compare Scope 1 & Australia's electronics magazine siliconchip.com.au Features & specifications Scope 1: rectification with a standard 40A diode bridge into Scope 2: the same setup as Scope 1 but with our Ideal a 120W load. The ground reference is the positive output rail Bridge Rectifier. The gate drive (mauve) goes 4.5V above the (blue), so the AC input (yellow) shows the voltage relative to rail, which in this case is sufficient to switch on the Mosfet. this and thus the diode voltage drop. Nearly 1V is dropped There is an 80mV offset of the AC input above the DC rail, across this large bridge even at modest loads. which gives safe switching of the Mosfet. The initial ‘glitch’ is the delay in the Mosfets being switched on – this test signal was running at 200Hz. Scope 2. These show a 10V RMS sinewave being rectified, first using a conventional 40A diode bridge, and then the Ideal Bridge. Design considerations Since the request was for an Ideal Rectifier that can be used in an audio power amplifier, let’s look at some reasonable requirements for such a power supply. The requirements can vary, so we had to consider the average case: • A typical amplifier can deliver about 200W into 8W; this sets the maximum voltages • The amplifier should be able to drive a 4W load with the maximum rail voltages; this sets the current ratings • A single board should handle the full split supply • The bridge should present much less than the 1.1V+ per ‘diode’ that you will see across a conventional bridge, consequently having lower power dissipation • We should keep the parts cost reasonable The derived requirements from the above are: • Rectified DC rails of ±60V DC, which can deliver an amplifier output of around 40V RMS. • A continuous output current of 10A for a 4W load, which we should assume is doubled if used in a stereo setup, so 20A. Normal audio has a very substantial crest factor, so while peak currents may be 10A for a 4W load, in reality the average current will be very much less siliconchip.com.au than this. Even in a subwoofer channel for a serious party, it would be surprising to see 10A continuous unless you have very heavy-duty speakers. In a practical application, this would correspond to a 40+40V AC transformer of 500VA or more, with its output rectified to feed some large (say, 20,000μF per rail) filter capacitors. One thing we need to consider is worst-case voltages. Mains voltage variation and transformer regulation can be significant. The nominal mains voltage range is 230V AC +10%, -6%, meaning a range of 216-253V AC. However, I have seen mine go well beyond this with the current trend towards many solar installations. Transformer regulation is typically 6%. This says our output supply will be at maximum: (40V AC + 10% + 6%) × Fig.1: the ideal bridge concept. It comprises four N-channel Mosfets with their gates driven by a control circuit. √2 +10% = 72V, giving us our ‘absolute maximum’ rating. The ‘nominal operating’ voltage level will be about 56V. The extra 10% at the end is a safety factor we are adding. In a conventional dual-rail power supply, the rectifier component, normally a diode but in our case a Mosfet, sees the total voltage between the positive and negative rails, so in our case it will have to handle an absolute maximum of 144V with the maximum mains supply and minimum load, plus a 10% safety margin. Our Mosfets therefore need the following ratings: • Vds > 150V • IDcont >> 50A (the peak current will be over 100A at switch-on) • Rds(on) < 10mW (trying to be reasonable) If we look for suitable P-channel Mosfets at Mouser, we see no options at all. If we allow a higher Rds(on) figure, the cheapest option is $27 for a single Mosfet. Ouch. Looking at an N-channel option, we have options starting at $4 in single quantities and $2-3 each in lots of 10. Of the 22 through-hole options available today, there are multiple usable options under $5 each. So all we need to do now is make our circuit drive four N-channel Mosfets in a bridge configuration, as shown in Fig.1. To switch on those N-channel Mosfets requires us to drive their gates (G) to a voltage above their sources (S) by about 10V (that’s enough to fully switch on the majority of Mosfets we are likely to use). Australia's electronics magazine November 2026  67 We don’t have a convenient supply rail above the output voltage, so like many Mosfet-based circuits, we will need to ‘bootstrap’ our gate drive for the upper two Mosfets, charging a capacitor during the negative half-­ cycle swing of each AC input waveform. We don’t have to do that for the lower Mosfets since we can simply pull their gates up to the 0V rail or the positive rail, as long as we clamp the voltages to something reasonable. Exceeding the Vgs rating of the Mosfet will damage it; this is typically in the range of ±10V to ±30V. A zener diode clamp will solve that easily. The majority of ideal bridge controllers are designed to rectify a single AC input to produce a DC output. Since their design is optimised for this, many of the integrated-circuit controllers switch the lower Mosfet on based on the associated AC input being at a lower voltage than the other. That means the lower Mosfets are switched on alternately with a ~50% duty cycle. This works just fine in a single-rail application, but precludes their use in a dual-rail application, where the Mosfets need to be on only when the input voltage is high enough to charge the filter capacitors. The LT4320 switching is shown in Fig.2 (from LT4320 data sheet). The green and orange dotted lines show the lower Mosfets being on for the majority of each half-cycle. What we need is the Mosfets switched on only during positive and negative excursions, like we see in Scope 3 & Scope 4, which show our prototype in operation. Our circuit to do this is shown in Fig.3. This has a fair resemblance to the simpler circuit for the Discrete Ideal Bridge Rectifier from September 2024, but exchanges the ‘pull-down’ on the top side Mosfets to a bootstrapped ‘pull-up’ that can drive the upper Mosfet gates to more than 10V above their sources. This lets us swap the upper P-channel Mosfets of the old design for N-channel types. Circuit details The two upper halves of the circuit are effectively identical (shown as mirror images). Similarly, the two lower sections are mirror images of each other, but they’re simpler because they don’t need to generate the higher-­ voltage bootstrapped supply for driving the Mosfet gates. Four named sections of the upperleft quadrant are shown in coloured, dashed boxes, so we will explain the operation of these sub-circuits. Current sink #2 sinks about 0.6mA to the negative output rail. This is required for the input comparator, which is formed from PNP transistors Q4 and Q6. With the maximum rails totalling 150V, the 0.6mA current sink will result in about 0.1W dissipation in Q9. Q4 and Q6 compare the AC1 input voltage to the positive output. Diode D6 is required to protect Q4 when the AC1 input swings negative. D3 is there to ensure we have matched voltage drops so that the comparator operates correctly. These diodes are BAV21 types as they need to have a minimum voltage rating of 150V and be reasonably fast, at least on the scale of the mains input. These diodes also need to be Scope 3: the switching waveforms for the positive rail at 11.5A RMS. A very minor switch-on waveform for the negative rail Mosfet is seen, but it has no effect. At this current, after several minutes, the Mosfets reached about 60°C. 68 Silicon Chip reasonably well-matched and at the same temperature, as the voltage drop across these directly affects our comparator function. We have placed these next to each other on the PCB, as we have with the comparator transistors, for thermal matching. The 220W resistor introduces a small offset voltage on the AC1 input side, which ensures that the AC input voltage must exceed the DC output voltage before the comparator output changes. The comparator operates by the bases of Q4 & Q6 being connected, and the emitters connected to the AC1 and DC output voltages via diodes. The comparator has a very high differential input voltage range due to the presence of D6 and D3, which stop the Q4 & Q6 base-emitter junctions from breaking down, to accommodate the full AC input voltage range. Current sink #1, which pulls current through Q4, is actually a current mirror of current sink #2. This will pull 0.6mA through D17. When the AC1 input is lower than the positive DC output, Q4 will be switched off, and this current will flow through ZD1 and D1. This will charge the 100nF capacitor to 12V, the breakdown voltage of zener diode ZD1. While the AC1 input is lower than the DC output, Mosfet Q1’s gate is pulled low by the current sink, which has plenty of time to charge the 100nF capacitor to 12V between AC cycles. As soon as the AC1 input voltage exceeds the DC output voltage, the potential of the emitter of Q4 goes above the emitter of Q6 and Q4 starts conducting. As Q4 switches on, it Scope 4: the switching waveforms for the negative rail at 11.7A RMS. It’s similar to the positive waveform; we see the positive Mosfet only on briefly, and a heavy drive to the negative Mosfet. This plot only shows one AC input to the bridge, but you can see DC ripple from the second AC input. Australia's electronics magazine siliconchip.com.au Fig.2 (left): the gate switching waveforms for the LT4320 IC show the gate drive for the lower Mosfets in dotted green and orange. These are on for much of the waveform, so if this device were used in a dual-rail application, it would short the transformer secondaries to the capacitors for the full negative cycle. Fig.3 (below): the Ideal Bridge Rectifier circuit. This will only work properly with dual/tapped secondary windings, producing positive and negative rail outputs into capacitive filters. This is ideal for power supplies in power amplifiers and some bench supplies. Note that the circuit shown here applies to the through-hole version. The SMD version uses equivalent components, but lacks the fused connections. siliconchip.com.au November 2026  69 supplies current through D8 and D17, pulling the collector of Q8 (current sink #1) high through D17. At the same time, it pulls the low side of the capacitor high through D8. The high side of the capacitor thus goes 12V above the positive output, which is connected to Q1’s gate, so Q1 switches on. If the Mosfet is switched on hard enough to bring the AC1 input within the offset voltage of the DC output (set by the 220W resistor), the comparator will reduce the drive to the Mosfet gate. So the circuit operates in a linear regime, switching on the Mosfet hard enough to reduce the 0.6V body diode drop to a few tens of millivolts. If you look at the gate drive waveforms in the scope waveforms, you will see some wiggles in the gate drive voltage at the leading edge of each pulse. This is the circuit operating in the linear region; only on the flat top of the waveform is the Mosfet switched on hard. D17 is present so that on deep negative swings of AC1 to a minimum, ZD1 and the 100nF capacitor can be pulled below the negative rail and therefore do not switch on Q1. If D17 was not present, as the AC waveform went negative, the negative end of the 100nF capacitor would break down the collector base/emitter junction of Q8. This would switch Q1 on, which would be bad. Once the AC input goes below the DC output, the Mosfet switches off, since current sink #1 pulls the Mosfet gate low via the zener diode and capacitor. The 100kW resistor in the gate drive section ensures that the Mosfet gate is always discharged when power is first applied. As you can see, the core elements of this design are pretty simple, but those diodes are there to deal with a number of subtle challenges. The Mosfet can be any suitably rated N-channel device in a TO-220 package. 100nF capacitors are sufficient for the gate drive reservoir, even for very large Mosfets. Choose Mosfets with Vgs(th) threshold voltage of 4V or more, as our negative gate drive ‘low’ is about 1.2V, and we don’t want any chance of the low-side Mosfets being partly on when they should be off. Most Mosfets use a standard pinout. Very substantial N-channel parts can be had at a good price. The IPP075N15N3GXKSA1 costs $32.30 for 10 and has an Rds(on) of just 6.2mW. resistor, and assuming an hfe of 30 for those two transistors, that happens at around 8V [(2 × 1.2mA ÷ 30) × 100kW] or about ±4V at the output (closer to ±5V when you take into account the base-emitter voltages). So 100kW is a good value for lower voltages, but with a ±75V output there will be nearly 150V across that resistor, so it will dissipate 225mW (150V2 ÷ 100kW), which is a bit too close to the ¼W resistor ratings for comfort. So bumping it up to a higher value like 220kW at the upper end of the supply voltage range is a good idea. How warm will the current source transistors, like Q15, get? With a ±75V output, Q15’s dissipation is around 150mW, which is easy for an SOT-223 package device to handle. Design notes & modifications There are four PCB designs. Two use through-hole parts and are basically identical except that one is for TO-220 package Mosfets and the other is for TO-247 package Mosfets. The advantage of TO-247 is that the packages are larger, so they can usually handle higher current and dissipation, but there are fewer TO-247 Mosfet options compared to the ubiquitous TO-220 parts. The third design uses mostly SMD parts, with some exceptions, such as the zener diodes. The Mosfets are in D2PAK (TO-252), a very common package for SMD power Mosfets, similar to TO-220 but designed for horizontal surface-mounting. There is also an SMD + TO-220 Mosfet board option. The SMD board is somewhat smaller, at 66.5 × 68mm compared to The current sources operate between the positive and negative rails. This allows the entire circuit to operate without a ground connection. However, it forces the current source to use a pull-up between the rails. The negative current source operates at a higher current than the positive one and presents the greatest challenge thermally. The 100kW resistors on the low side should be increased to 220kW for output rails above ±40V. We can’t make the value too high as it affects the minimum voltage at which the circuit will operate. There needs to be enough current through this resistor to switch Q14 and Q15 on sufficiently to deliver about 1.2mA of collector current each. With a 100kW Versions This is the SMD version of the Ideal Bridge Rectifier, using IRFS4115 N-channel Mosfets in a D2PAK (TO-252) package. 70 Silicon Chip Australia's electronics magazine siliconchip.com.au Scope 5: this plot has the oscilloscope ground connected to Scope 6: this plot has the oscilloscope ground on the the negative rail, allowing us to see the AC waveforms in positive rail; we can again see the gate drive in yellow, well purple and blue relative to this, and the gate drive in yellow. over 10V. The AC inputs in mauve and blue show that the Channel 2 (cyan) is the voltage at the load, after 60cm of AC input rises 300mV above the positive DC output. This heavy duty hookup wire. We see a full 12V gate drive here. means we have 300mV across the Mosfet, consistent with the 12.1mW Rds(on) figure for the Mosfet we used here. 76 × 82mm for both of the throughhole versions. It lacks just one thing: onboard fuses (and the corresponding connectors). You can use inline fuses, or run it without fuses; after all, a standard diode bridge doesn’t include any fuses. Instead, they are usually in the mains input and on the amplifier boards being driven. The fourth design also primarily uses SMDs, similar to the above version, except it allows through-hole TO-220 Mosfets to be used if that is desired (eg, if you can get better/ cheaper TO-220 package Mosfets or you want to add flag heatsinks). While it is identical to the TO-252 version apart from the Mosfets, some SMD components were moved to make room for them. Depending on your application, you may be able to use cheaper parts than we are suggesting. If your total rail voltage will never exceed 100V, you can use 1N4148 diodes rather than the BAV21 devices, and you could choose lower-rated Mosfets. For example, 100V-rated Mosfets with similar Rds(on) and current ratings can be found for less than $2 each. Be judicious in these choices, but there are real savings to be had depending on your use case. residue and apply a conformal coating such as Chemtools CT-ACC-400, available from Altronics. That’s especially recommended if you are going to be pushing the device to the upper end of its 150V maximum total voltage rating. Testing the prototype To properly test the Ideal Bridge Rectifier prototype, we took a 600VA, 45V + 45V AC transformer, two 33,000μF computer-grade capacitors (which are simply enormous) and ran this combination into an 8W resistive load, which is way beyond what you would see in an amplifier. We stopped when our load hit 300°C. We omitted a soft-start circuit in our tests, which you should never do with this sort of power supply. We do not recommend you do what we did, as a slip could lead to very hazardous faults. Construction First, make sure you have the right board to suit your parts: code 18109261 (76 × 82mm) for the throughhole version with TO-220 Mosfets, 18109262 (76 × 82mm) for the version Photo 1: measuring the Mosfet temperatures during operation with a 20A+ continuous load. I used a non-contact IR thermometer as it isn’t safe to put fingers near the board when powered or while the capacitors may still be charged! Safety We have followed IPC guidelines on track separation for benign environments and non-coated boards up to 150V in the PCB layout. In normal use, this will be fine. However, we recommend that once you have verified the board is working, you clean off the flux siliconchip.com.au We learned a few things from these tests. The omission of the soft-starter resulted in the mains circuit breaker tripping about one in four times we switched this on. The results are shown in Scope 5 and Scope 6. That was using the D2PAK (SMD) version of this circuit because these devices have much less thermal dissipation capacity than the through-hole devices in the way we have applied them. So it was a worstcase test. Throughout these tests, the Mosfets were dissipating about 600mW, and after about 10 minutes, they were at 45°C, as shown in Photo 1. Australia's electronics magazine November 2026  71 with larger TO-247 Mosfets, 18109263 (66.5 × 68mm) for the SMD version with TO-252 Mosfets or 18109264 (66.5 × 68mm) for the SMD version with TO-220 Mosfets. Refer to Figs.4-7 as a guide during construction. Start by mounting all the resistors. For lower-voltage applications, use 100kW resistors as marked with an asterisk on the PCBs (or in red in Figs.47). For higher-voltage applications, substitute 220kW resistors for those. Follow with all the diodes. Pay attention to their orientation. I normally endeavour to keep these all facing the same direction, but to keep the board size small, we need to optimise routing and thus component orientation. Follow with the remainder of the transistors. Note that the positive rail comparator uses MMBTA92/MPSA92 transistors, as Q5 can experience the full rail voltage (base-collector) on negative AC swings. Q9 is the same part as Q5, because we need the Vbe on Q5 and Q9 matched for the comparator to work. Do not mix these up with the BC series parts and be aware that the pinouts between the through-hole BC- and MPSA- prefix parts are reversed. On the through-hole version, we want to glue the comparator transistors together in pairs. This is about as good as we can do to ensure they remain at the same temperature. Take two pairs of the BC546 and two pairs of the MPSA42 and glue them faceto-face using super glue or shrink a length of heatshrink tubing around them. Put them straight into the PCB so you are sure they have gone into the right spot. The MJE340/350 and equivalent PZTA-prefix SMDs ensure we have good power dissipation headroom. Watch the orientation of the throughhole parts. We have marked the heatsink tab sides of these on the board and indicated the base pin so that you can check if you are in doubt. Finally, mount the capacitors, fuse clips and spade connectors. Solder the fuse clips and connectors on both sides. When you mount the Mosfets, make sure these are well soldered to both the top and bottom layers. If you expect to be drawing in excess of 10A continuously from the power supply, you might want to think about heatsinking the Mosfets with modest flag heatsinks. Be aware that the tab of these is connected to the drain pin, so these present an electric shock hazard. Thus, the Figs.4-7: use these PCB overlay diagrams as a guide during construction. Make sure you use the one that matches the board you have (note the codes, ending in 1-4). The board with TO252 Mosfets has two of them installed on the underside. Those are the only parts that mount on the bottom (it simplifies routing and helps spread out the heat load). 72 Silicon Chip Australia's electronics magazine siliconchip.com.au heatsink definitely needs to be insulated from them. Parts List – Ideal Bridge for Power Amplifiers Testing Through-hole version Setting the Ideal Bridge Rectifier to work should be undertaken at low power, to ensure things are working properly before you connect this to a large transformer and capacitor bank. We used a signal generator and 100μF filter capacitors with 1kW loads to test ours, as shown in Fig.8. We want to make sure the rectifier works before we connect it to a massive transformer and capacitor bank. The rectifier itself presents very little load to the supply, so you can use a “supply” as modest as an oscillator for testing. 1. Connect your oscilloscope’s ground to the ground point. 2. Set the oscillator to produce a sinewave at 50-200Hz and its maximum undistorted voltage. Make sure it is delivering more than 10V peak-topeak; if it can’t, you need to amplify it with a small audio amplifier. As an alternative to this, you could use a small AC plugpack with, say, a 100W 5W resistor in series with the output. 3. Connect the oscillator to the AC1 input of the Ideal Rectifier board and: a. Check with your oscilloscope that the oscillator has not been loaded down too much. b. Check the positive and negative outputs and see if you can measure a credible rectified DC voltage (notionally 1.4 times the RMS AC input). c. You should see ripple as the capacitors discharge on each half-cycle. d. Check the voltage on the cathode of zener diodes ZD1 and ZD4. ZD1 should have a waveform as shown in Scope 7, while ZD4 should have a waveform as shown in Scope 8. 4. Swap the oscillator to the AC2 input and repeat the above tests, looking at the cathodes of zener diodes ZD2 and ZD3. 5. If anything looks wrong: a. Check the diode orientations. There are quite a few, and they all need to be right. b. Check that your transistors are in the right spots, especially that the MPSA92 and BC5xx devices are where they need to be. c. Check that the TO-126/TO-225 package medium-power devices are the right way around 1 double-sided PCB coded 18109261, 76 × 82mm 6 5mm pitch PCB-mount 6.3mm vertical spade connectors (CON1-CON6) 4 M205 PCB-mount fuse clips (F1, F2) 2 M205 fuses (F1, F2; ratings to match application – 10A slow-blow suggested) 1 can of conformal coating spray (optional) [Chemtools CT-ACC-400, Altronics T3175] Semiconductors 4 IPP075N15N3GXKSA1 150V 100A N-channel Mosfets, TO-220 (Q1, Q2, Q21, Q22) ◆ 4 MPSA92 300V 500mA 625mW PNP transistors, TO-92 (Q3-Q6) ◆ 4 MJE340 300V 500mA 20W NPN transistors, TO-225 (Q7-Q10) ◆ 6 BC546 65V 100mA NPN transistors, TO-92 (Q11, Q12, Q17-Q20) 2 MJE350 300V 500mA 20W PNP transistors, TO-225 (Q14, Q15) ◆ 1 BC556 65V 100mA PNP transistor, TO-92 (Q16) 4 12V 400mW/500mW axial zener diodes, DO-35 (ZD1-ZD4) [Altronics Z0332] 18 BAV21 250V 250mA switching diodes, DO-35 (D1-D18) ◆ Capacitors 2 100nF 50/63/100V ceramic or MKT polyester Resistors (all ±1% ¼W leaded axial) 7 100kW (3 220kW and 4 100kW above ±40V) 4 22kW 2 470W 2 68W 4 1kW 2 220W Substitutions for TO-247 variation 1 double-sided PCB coded 18109262, 76 × 82mm 4 IRFP4668PBF 200V 130A N-channel Mosfets, TO-247 (Q1, Q2, Q21, Q22) Kit (SC7718, $65 + P&P): includes all the parts needed to build the TO-220 version (without the conformal coating; PCB code 18109261). siliconchip.com.au Australia's electronics magazine SMD version 1 double-sided PCB coded 18109263, 66.5 × 68mm 4 5mm pitch PCB-mount 6.3mm vertical spade connectors (CON2-CON5) 1 can of conformal coating spray (optional) [Chemtools CT-ACC-400, Altronics T3175] Semiconductors 4 NTB7D3N15MC 150V 101A N-channel Mosfets, TO-263 (Q1-Q2, Q21-Q22) ◆ 4 MMBTA92 300V 500mA 625mW PNP transistors, SOT-23 (Q3-Q6) ◆ 4 PZTA42 300V 200mA 1W NPN transistors, SOT-223 (Q7-Q10) ◆ 6 BC846 65V 100mA NPN transistors, SOT-23 (Q11, Q12, Q17-Q20) ◆ 2 PZTA92 300V 500mA 1W PNP transistors, SOT-223 (Q14, Q15) ◆ 1 BC856 65V 100mA PNP transistor, SOT-23 (Q16) ◆ 4 12V 400mW/500mW axial zener diodes, DO-35 (ZD1-ZD4) [Altronics Z0332] 18 BAV21W 250V 250mA switching diodes, SOD-123 (D1-D18) ◆ Capacitors 2 100nF 50V X7R M3216/1206-size SMD MLCC Resistors (all ±1% ¼W M3216/1206-size SMD chip type) 7 100kW (3 220kW and 4 100kW above ±40V) 4 22kW 2 470W 2 68W 4 1kW 2 220W Substitutions for TO-220 variation 1 double-sided PCB coded 18109264, 66.5 × 68mm 4 IPP075N15N3GXKSA1 150V 100A N-channel Mosfets, TO-220 (Q1, Q2, Q21, Q22) ◆ ◆ can be purchased from DigiKey or Mouser November 2026  73 Scope 7: testing the Ideal Bridge Rectifier using a 100Hz sinewave from a simple oscillator at 12V peak-to-peak, measured at ZD1’s cathode. With lower-current loads, the mauve waveform (Mosfet gate drive) will be much lower in amplitude. d. Check that the Mosfets are the right way around. 6. You should also be able to independently check the gate drive waveforms on ZD1-4. If one of these looks wrong, check the parts in that section. Using it Remember that inside the case of any high-power amplifier or similar device, there are many parts at very high voltages. When using this Ideal Bridge Rectifier, be aware that: • The tabs on the TO-220/TO-247 Mosfets are connected to their drain pins, which connect to the positive output and AC inputs to the supply. With high rail voltages present in a high-power amplifier, these are electric shock hazards. This device must be packaged inside a closed case and protected from unintended contact in normal use (eg, with an acrylic shield over the top). Scope 8: a similar test to Scope 7, this time for the negative rail, with the gate waveform measured at ZD4’s cathode. This is at a high current; when less current is drawn, the mauve trace will show a lower gate drive voltage. • Never touch the Mosfets to see if they are hot. These Mosfets connect to the filter bank, and this will remain charged after power is removed. Simply switching the power off is not sufficient to make a power supply safe. • High-power DC supplies, especially those with toroidal mains transformers and large capacitor banks, can draw very high inrush currents on startup. • The Mosfets we recommend are extremely tough, but you really should consider a soft-start circuit that puts a resistor in series with the mains side of the transformer for a second or two to reduce the inrush current, such as our SoftStarter project (April 2012 issue; siliconchip.au/Article/705). • We have included reasonable margins for mains supply variation, transformer regulation and a safety margin. Don’t push your luck on This is one of the through-hole versions of the Ideal Bridge Rectifier, using TO-220 package Mosfets for Q1, Q2, Q21 & Q22. The fuse clips are present but there are no fuses installed in these photos. increasing the input AC voltage too high. We have seen our mains voltage go over 260V AC, in excess of the mains standard in Australia, so we feel the extra 10% margin we provided is essential. • This device is intended to be mounted close between your mains transformer and filter bank. You need the filter bank connected for the bridge to operate properly, so keep that in mind. • The PCBs lack a centre-tap ground connection, but you will need to connect the centre tap of the transformer to the centre tap of the filter bank for correct operation. Note the ground connection in Fig.8. • Keep the AC input and DC output wires short, and keep the ground connection bundled with them so that the loop created from the AC input through the bridge to the filter capacitors is small. This will minimise radiated EMI. With these tips in mind, you should be able to reduce the dissipation in your rectifier and save a few volts of drop across it, for just a bit more money than an old-school diode bridge. SC Fig.8: the test setup we used. Our oscillator was set to 200Hz at the maximum undistorted output voltage, which was about 8V RMS. You could use a small audio amplifier to drive the bridge if your signal generator isn’t capable. Test each AC input one at a time and verify both the positive and negative rectified outputs. 74 Silicon Chip Australia's electronics magazine siliconchip.com.au