Fullscreen HTML5Med Res (??MB)HTML4

Salvage It! By KEN KRANZ This is a rather different Salvage It: We’re not trying to recycle a complete device but instead, just one part of it. We’re looking at Switch-Mode Power Supplies . . . and specifically, the CommonMode Choke at the mains input. There’s a lot you can do with one of these handy components. T hese days, most electronic equipment has a switchmode (or switching) power supply. They’re cheaper than transformer-based (linear) supplies and you can obtain more “grunt” from a given space. And while they’re pretty reliable, they do occasionally fail (sometimes spectacularly!). There’s not much worth safely salvaging in a dead switchmode supply – the chances are at least some of the semis (if not most/all) have failed and, given the supply’s relatively high temperature operation, you wouldn’t want to place too much faith in any electrolytics (check them out – we’d bet London to a brick that many would show signs of distress – especially swelling on top). 50kHz Apart from the smoke which escaped from DRIVE (1) 200W Mains Inverter, Februaryl 1994* (2) Courtesy Light Delay, October 2014 * some components may now be difficult to obtain SPICE simulations may be downloaded from siliconchip.com.au 90  Silicon Chip FLOATING DRIVE SECTION K D2 BAT54 Q1 IRF1405 A D1 BAT54 10nF D 150W K C1 20nF SILICON CHIP Floating Gate Drive circuits Two projects spring to mind which had floating gate drives, employing the type of circuitry in this feature. They provide good background and reinforce the theory explained here: the supply above (and there was a lot of it!) at least some of the electrolytics are showing they’ve failed the battle of the bulge. But there is one component which is worth saving, if only because it is so useful in other ways. That component is the common-mode choke (CMC). It’s rare that the CMC will have failed (and that is easily checked) so it’s well worthwhile to remove it before junking the rest of the SMS. G S A V1 L1 30mH R1 22W L2 30mH K 10nF D3 BAT54 A K 22k 10pF 10nF D4 BAT54 A Fig.1: An LTSPICE simulation of the CMC used to provide a floating gate drive for a MOSFET. In this simulation, the voltage source (V1) at the left simulates a PIC micro with a 4.9V supply. siliconchip.com.au What’s a CMC? The attraction of CMCs is the fact that two windings are on a closed ferrite core, often with very good high-voltage insulation between the windings and low capacitance between the windings. The chokes are not designed to be used as transformers but if the choke and application are selected carefully the results can be very good indeed. First some recommendations: keep the core flux density below 1500 Gauss and limit the frequency to <75kHz. Below 20kHz most cores seem to be OK at around 2000 Gauss. To calculate the flux density use the following simple formula: Bmax = 108 E   KANf Bmax = maximum flux density in Gauss. E = voltage across coil. N = number of turns of the coil. A = effective area of the coil in cm2. f = frequency in Hz. K = 4.44 for sinewave (RMS). K = 4 for squarewave (peak). Often the number of turns can be counted or estimated (without destroying the coil). The small black 30mH CMC tested had 93 turns on each winding (one winding was un- Actual volage drop across the 22 resistor. wound for this information, it was worth the sacrifice as it cost less than $1.50 on ebay). Leakage inductance is normally higher than for a customwound transformer but often it is more than satisfactory for the task. The low capacitance between windings is often a big win. It was decided to test the circuit at high power; as I do not own a 300W 10resistor a simulation was carried out, again using SPICE. A 10 250W resistor and a 50V, 5A power supply were simulated, with the input signal to the CMC a 20ms burst of 50kHz (FET-on) followed by 20ms of no drive (FET-off). The FET turn-on switching loss was <3mJ, the FET turn-off switching loss was <10mJ, the power FET’s loss when ON (static) was <200mW. MISS THIS ONE? This simulation is of a 3ms on-pulse, actually a 3 millisecond burst of 50kHz from the PIC, the waveform measurement was taken at the gate-drive test point. Published in Dec 2012 2.5GHz 12-digit Frequency Counter with add-on GPS accuracy And here’s a scope trace of the actual waveform. It has more ripple than the simulation; this could be removed, at the expense of a longer turn-off time, by increasing the value of C2 (10pF). In the simulation, it effectively does nothing, the gate capacitance of the FET does the job – it is on the schematic to show where to add capacitance if desired. The ripple is well above the FET’s gate turn on voltage so it’s not actually a problem. siliconchip.com.au Wow! 10Hz – >2.5GHz in two ranges; 1ms - 999,999s with a 12-digit LED display. It’s a world beater and it’s the perfect addition to any serious hobbyist’s bench – or the professional engineer, technician, in fact anyone who is into electronics! You’ll find it one of the handiest pieces of test gear you could ever own and you can build it yourself. All the hard-to-get bits (PCBs, micros, LEDs, panels, etc) are available from the SILICON CHIP Online Shop. You’ll find the construction details at http://siliconchip.com.au/project/2.5ghz PCBs, micro etc available from On-Line Shop January 2015  91 RL TP1 R2: 50W* OUTPUT CMC R3 1W L1 30mH *SIGNAL GENERATOR OUTPUT RESISTANCE L2 30mH R1 1k Fig.2: using a 30mH CMC as a transformer. USING A COMMON MODE CHOKE AS A TRANSFORMER INPUT 37.3kHz SQ WAVE L3 7.5mH R2 0.031W L4 34mH OUTPUT R1 7.2W L5 7.5mH R3 0.031W R2, R3 = WINDING RESISTANCE L4 = LEAKAGE INDUCTANCE R1 = 20W LOAD AT 12 VOLTS Fig.3: this time the input is a square wave at 37.7kHz but SIMULATION OF THE SAME CIRCUIT the CMC is much smaller. The slow speed of the switching is fine for many applications – it actually keeps the RFI down. How much drive current does this require ? For the following test C1 was changed to 100nF to increase the gate voltage and the drive was a continuous 50.25kHz square wave from the PIC. The voltage drop was measured across the current sense resistor R1 in Fig.1. The peak voltage was 104mV, peak current 4.7mA (0.104/22) and RMS voltage 52.2mV, so the current required to hold the high-side switch on is very reasonable. Common mode chokes are a good choice for pulse triggering of SCRs, quite high trigger currents can be obtained with suitable chokes. The actual measurements above were taken using a square wave drive from a PIC 16F1783. This has a super-cool PSMC (Programmable Switch Mode Controller Module) with no less than 10 modes of operation: • Single phase • Complementary single phase • Push-pull • Push-pull H-bridge • Complementary push-pull H-bridge • Pulse skipping • Variable frequency fixed duty cycle • Complementary variable frequency fixed duty cycle • ECCP compatible modes - Full bridge - Full bridge reverse • 3-phase 6-step PWM The chip looked so interesting I made up some test boards. Other uses for CMCs If the rules regarding flux density are followed and the source impedance is low enough, CMC’s can be used at low frequencies with a reasonable bandwidth, normally obtained at millivolt levels. The primary inductance combined with the signal source impedance forms a high-pass filter. Making the source impedance low reduces this effect. The leakage inductance combined with the load on the secondary forms a low pass-filter. A higher value load resistor can be an advantage. As a rule of thumb when using a CMC as a signal transformer, I keep the source impedance under 1/10th of the primary inductive reactance Xl, calculated for the minimum frequency that is expected to be used. Most cores at low frequency seem to be OK with a maximum flux density of 2000 Gauss. Using the 30mH CMC a circuit was set up as shown below. At 50Hz sine wave, the maximum input voltage would be 53mV (1974 Gauss). The 50resistor was used as it is built into my signal generator. With the 50:1voltage divider the maximum input (AC in) is 2.55VRMS. Some tests were carried out with the signal generator set such that the input into the transformer was a 68.8mV square You can get enough power out to run a 10W halogen globe, as seen here. The input into the transformer was a 68.8mV square wave at 1.0kHz from the signal generator. The scope screen at left shows the 1kHz output, with a very respectable rise time (right). A gain recovery amplifier would normally be required. 92  Silicon Chip siliconchip.com.au Sine Wave Tests My signal generator started to clip when driven to 53.3mV RMS, so testing was carried out at 56mV peak into the transformer (across the 1 resistor). wave at 1000Hz, with the results shown opposite. What about higher power? A CMC was removed from the mains input filter of a large switched mode power supply (the one shown overleaf) that had destroyed itself, along with many major components. The core details measured were: Core OD .......................................................... 22.4mm Core ID............................................................... 14mm Core Height ...................................................... 8.3mm Core cross sectional area .................................35 cm2 Measured Inductance ........................................7.5mH Measured Leakage inductance ............................35H Number of turns ....................................................... 29 DC resistance.................................................... .031 50 Hz output into 1000. A 20W halogen lamp was set up as a secondary load and a 37.3kHz square wave used as an input to the primary. After running for half an hour the CMC was warm to touch. The calculated flux density was 2116 Gauss. It can be seen that common mode chokes can be very handy when used for applications other than their intended purpose. The construction used for the mains input types gives superb high voltage isolation. Note: I adjusted the simulation for a 10W halogen lamp; the reduced effect of the leakage inductance allowed the input voltage to be reduced to 15.1V peak input for 12VRMS out. The output was a nicely rounded square wave. The flux density was just over 1500 gauss. I re-ran at 25kHz, 13.9V peak input for 12VRMS at the load, with the calculated flux density 2090 gauss. SC Radio, Television & Hobbies: ONLY 0 the COMPLETE 0 $ 2 6 0 P&P archive on DVD + $1 1kHz output into 1000. 100kHz output into 1000. It can be seen apart from the low level and low-Z input the results are very handy for less than $1.50. It should be possible to run some low frequency Manchester code through these devices when low capacitance galvanic isolation is required. Even audio could be worth a try – some opamps can deliver the current required. 1k input impedance can be very low noise for many opamps. siliconchip.com.au • Every issue individually archived, by month and year • Complete with index for each year • A must-have for everyone interested in electronics This remarkable collection of PDFs covers every issue of R & H, as it was known from the beginning (April 1939 – price sixpence!) right through to the final edition of R, TV & H in March 1965, before it disappeared forever with the change of name to Electronics Australia. For the first time ever, complete and in one handy DVD, every article and every issue is covered. If you're an old timer (or even young timer!) into vintage radio, it doesn't get much more vintage than this. If you're a student of history, this archive gives an extraordinary insight into the amazing breakthroughs made in radio and electronics technology following the war years. And speaking of the war years, R & H had some of the best propaganda imaginable! Even if you're just an electronics dabbler, there's something here to interest you. NB: Requires a computer with DVD reader to view – will not work on a standard audio/video DVD player Use the handy order form included in this issue January 2015  93