Nikkai QM82 800W DC/AC Inverter (also marketed as Genius, SkyTronic and PowerUp)
1. Overview This is a report of the repair of an inverter purchased on ebay, having been described as “faulty the inverter suddenly stopped working. It just died”. Upon opening the case, it was obvious that the unit had major areas of burning to the PCB, one chip had exploded, and there were several burned resistors. Dismantling the unit revealed that there had been a couple of replacement components fitted sometime, whether before or after the fire was not clear. A search on the net showed that there was little or no information available, but a plea on the newsgroup sci.electronics.repair newsgroup sci.electronics.repair got the component values for those which were too burned to identify, identify, and the type for the exploded e xploded chip. Because of the extent of the fault, I decided that a circuit diagram (schematic) was required, so the next few days were spent carefully tracing the circuit and drawing it up.
2. Circuit description The inverter converts 12V DC from a vehicle or leisure lead-acid battery to 230-240V AC suitable for a range of mains-powered appliances in Europe. It uses a two-stage converter to keep efficiency up (quoted at 85 - 90%) and parts weight down (3.15Kgs). The first stage is a 12V SMPS driving the transformer primaries at high frequency, and the second stage takes the high voltage from the transformers, rectifies it and chops that at 50Hz to drive the mains voltage output. An additional low-power secondary stage drives the 50Hz chopper and pulse-shaper through a 12V regulator.
2.1 Low-voltage stage
The 12V battery voltage is supplied through a set of parallel-coupled 30A fuses directly to the centre-tapped primaries of the six series-connected high-frequency transformers. The start and finish of the primary windings are connected in parallel to six RFP50N power MosFETs MosFETs per side (again, in parallel - to increase the current handling and decrease the RDSON). The 12V supply is also connected, via the front-panel on/off switch, to the KA3525A switch-mode switch-mode control chip and to the cooling fan. The SMPS oscillator runs at about 4.5kHz and directly drives the output MosFETs MosFETs (via the usual gate-blocking resistors) from the push-pull outputs on pins 11 and 14. Error control is provided by an LM393 op-amp, which is driven from the high-voltage side via a TLP734 opto-isolator. Ten 2200µF electrolytic smoothing capacitors are placed on the 12V battery supply to iron out any stray AC component. All of the timing capacitors are low-tolerance ceramic chip – there seems to be no requirement for precision here!
2.2 High-voltage stage
BE WARNED: This stage runs at approximately 300V DC – and accidentally touching a live part could be lethal. There are a couple of big capacitors which can store this voltage for up to an hour or more – so be very careful! If you are unsure about anything, don't touch it, and call a qualified engineer to sort it out!!
Here's where it all starts to get a big more interesting! The output of the high-frequency transformers is bridge-rectified by four fast diodes (D1 – D4) and smoothed by two 330 µF capacitors. The resulting 300V DC is applied directly to a bridge switching output circuit comprising four high-power MosFETs. MosFETs. These are switched to provide a 50Hz AC waveform at 230240V rms which is LC filtered by L1 and C3. Output current is measured as a voltage drop across the source resistors (R29, R74 & R75) – at full load, the source terminals te rminals of the lower two MosFETs MosFETs will be at around 1V above ground. This voltage is applied to an LM393 op-amp to trigger the overload shutdown.
2.3 50Hz pulse generator AC from a low-voltage secondary winding on the transformers is rectified by four diodes (D5 – D8) and connected to a 7812 linear voltage regulator to provide the 12V supply for the 50Hz oscillator and driver components. Running the low-frequency circuit in this way ensures that the output only runs when the input SMPS is running effectively effec tively.. U4 is a standard 556 dual oscillator/timer arranged to provide a 50Hz pulse at the base of Q1. The frequency is adjustable with VR1. This pulse is applied to CMOS logic gates to generate the gate drives for the output MosFETS. The output pulse shape is adjusted by a bleed through R33 from the high-voltage DC to maintain the correct AC output voltage.
3. Fault investigation 3.1 Visual before dismantling Taking the lid off the case, I saw several areas of black carbon, concentrated around the four largest power devices and around four smaller power transistors. There were several burned resistors, and one IC which had exploded. The fuses were intact. An electrolytic capacitor had burst and an adjacent power component had broken apart. There were several burned bits of PCB track. After dismantling the unit, which involves removing all the screws securing the heatsinks to the case and sliding the PCB out of the aluminium housing, a closer examination of the PCB revealed that one chip had been replaced before and that there were a couple more breaks to the copper print on the underside. At this stage, a circuit diagram was needed, and since searches on the 'net were fruitless, I settled down to draw one out. Several days passed.
3.2 Finding the burnt-out parts A plea on the newsgroup ( sci.electronics.repair ) sci.electronics.repair ) for someone who was prepared to take the cover off another inverter and report back on the part numbers and resistor values was successful – I now had somewhere to start! Searches on the 'net gained data sheets for all of the semiconductors (except the four smaller power transistors – which are drivers for the output MosFETs) and the circuit diagram began to take shape. The result can be seen on this web site in two JPG files. Once the diagrams were to hand I could proceed to actually fixing the inverter.
3.3 Replacing parts and testing My initial thoughts about the failure sequence were that since the 12V smoothing cap had burst, and the 7812 regulator had cracked, this might be to origin of the whole thing. It's often true that electrolytics, subject to high temperatures at close to their maximum ripple current, can fail catastrophically and take associated components with them. In this case, a failure of the 12V regulator seemed to have resulted, and the 50Hz pulse generator circuits had suffered – including one CMOS chip that had literally exploded. Having a failed driver, the output MosFETs MosFETs could have been switched on and shunted the 300V to ground – a lethal result for those devices! I started with the SMPS controller (U3), since that had already been replaced and it was driven directly from the 12V battery. battery. Coupling up the scope and a temporary 12V power supply directly to U3, I could see the output waveform happily driving the MosFET gates and the cooling fan was whirring away. Applying the 12V supply to the MosFETs MosFETs started the voltage transformation for a few seconds, then a protection circuit activated and shut down the controller. Repeated short bursts of 12V to the SMPS controller got the HV capacitors (C28, C29) nicely charged to about 300 volts, so I knew the transformers were okay. Similarly acceptable results were measured at the input of Q14 for 12V regulated supply. I then replaced the 7812 regulator (Q14) and fitted a new, higher-rated smoothing capacitor at C29. Further testing with the switching controller confirmed that the regulated 12V DC supply was working properly. Next, by using the 12V power supply, I checked the 556 timer (U4) – which wasn't doing much at at all. This tended to confirm my original suspicion that the failure of the 12V regulator had mistreated all the connected semiconductors. Replacing the 556 brought that to life, and replacing the various CMOS pulse-shaping chips (U1 & U2 on the sub-board, plus U8 & U9) got all that section working. I now had satisfactory signals available at the base of the driver transistors Q17 - Q20. While waiting for some parts to arrive, I tested a few ancillary sections of the circuit. The lowvoltage alarm around U2A was okay and sounded the buzzer at about 10.5 volts. The low-voltage shut-down at U2B was also working, and stopped the SMPS at about 10 volts input. With a 100W 240V light bulb soldered across the 300V DC output, to provide some sort of load and keep the voltage within acceptable limits, I checked the temperature sensor – by simply sticking a soldering iron into the mounting hole of the sensor tab! This had no effect, but the thermistor tested okay, as did the driver transistor Q16. I replaced the opto-isolator, U6, and the temperature circuit worked properly.
3.4 Those driver transistors Some investigation was required because the driver transistors were all completely dead and carried short-code identification. Two people helped by looking at their inverters and found that they had similarly identified transistors, but a lot of searching in the net revealed no information at all. The driver signals from the CMOS were a square-wave at about 5V peak to peak. At start-up the voltages were high on Q19 and Q20, and low on Q17 and Q18. Looking at the circuit, it seemed that Q19 and Q20 should be NPN devices, and Q17 and Q18 should be PNP devices to ensure that the output MosFETs MosFETs (Q21, Q24, Q25 and Q28) were switched off initially – until the 50Hz switching oscillator had started up and could provide acceptable signals. Since the dead components were housed in TO-126 cases, it seemed that they were expected to handle appreciable power, so an investigation into possible devices was made. Of course, until the whole circuit was running, it was not possible to measure the voltage on the higher voltage terminals, although the low voltage terminals were at the same potential as the sources of Q21 and Q25 – nearly ground. Since the gates of the IRFP450 MosFETs are limited to 20V (absolute maximum) the BD135 & BD136 (45V collector to emitter) seemed to be suitable. After fitting those BD135 and BD136 transistors, I powered up the inverter and found to my horror that they failed more or less instantly. A question to Bob (this site owner!) revealed that the collectors of at least two of the devices were handling nearly 200V, 200V, so it was back to the Towers International Transistor Selector to find some that were more suitable. The specifications for the MJE 340 and MJE350 seemed okay, so these were chosen. After installing these components, I checked the drive to the output MosFETs MosFETs on the 'scope and found that the signals were variable - one FET wasn't driven at all, and one had a very unacceptable signal. I discovered that a couple of 1N4148 transient protection diodes were failed, and after replacing these, the signals looked much better - a nice square wave of reasonable amplitude. However, it seemed that the output pairs were being driven 180 degrees out of phase. I went back to the design of the output stage and discovered that actually, the drivers were better if they were all NPN devices. A further look at the transistor tables suggested that the high-voltage drivers were ok as MJE340, and that BD135 would work for the low side drivers. Now the signals on the scope looked much better, and the output MosFETs were driven in the correc t phase. I still was not sure about connecting the output devices directly to the 290V DC, so I rigged up a variable supply to drive the MosFETs, MosFETs, coupled the 'scope to the AC output and switched on. Ramping the input voltage up to 60V produced a nice AC output, and putting a standard 240V light bulb across the output got it to glow nicely. A good good result so far!
3.5 Finally..... There comes a time when all the analysis and substitution has to be put to the test. Connecting the 290V DC to the output MosFETS via a variable resistor, I found that the output was working, and ramping the resistor to zero ohms, a nice AC output was obtained. I removed the resistor and remade the proper connection, and switched on again. Plugging in a 240V lamp, it illuminated, and adjusting the Voltage Voltage pre-set adjusted the output to a steady 230V – even when I connected several more lamps to increase the load.
