If you look at the post number, you will realize that this post actually got started a long time back, but didn’t see the light of day for a while. The main reason for that was because I didn’t want to spoil the surprise that I was doing a RoadTest of the Tektronix PA1000. Now that the review has been published, now I can unveil this …
Readers of the RoadTest will realize just how problematic mains electricity is for IEC 62301 Ed.2 standby power testing. The requirements are very stringent, including that the voltage and frequency be of the nominal value for the region (230v, 50Hz for Australia) and remain within a 1% window. This means the voltage should remain between 227.7v and 232.3v throughout a test, and the frequency should remain between 49.5Hz to 50.5Hz. The frequency is easy, as such large frequency excursions on the grid often mean catastrophic blackouts are going to be happening soon, but the voltage is not so easy …
As excerpted from my RoadTest, the voltage histogram for >24 hours logging at my house gave me this. There’s a very high probability that the voltage is somewhere between 243 to 247v, roughly, which is way too high. This can be fixed with a variable autotransformer, also known as a Variac. That would “scale down” the voltages, however, that doesn’t fix the range issue. The voltage can be seen varying a span of around 10v, which is way too much. You might be able to get a run here and there (15-minute periods) where it stays stable enough, but often, it will not.
The other requirement includes the Total Harmonic Content of the first 13 harmonics be less than 2%, which occasionally gets exceeded due to industrial loads nearby. The crest factor is also a bit low at 1.39-ish, due to peak-flattening (lots of switchmode supplies).
In order to get a chance to do proper IEC 62301 standards test reports, the easiest way is to use a synthetic mains source. Unfortunately, high end laboratory grade power synthesizers cost an arm and a leg, so I decided to “cheap out” and give a low cost pure-sine-wave inverter a try.
This particular unit doesn’t have a visible branding on the front, and is rated as a 300 watt (600 watt surge) pure sine wave inverter. Such units are commonly used by campers, field technicians and by people who want to use a mains appliance in a car without worrying about buying an automotive-adapter (e.g. those running a laptop using its originally supplied adapter).
Such units have been reducing in price quite a lot over the years, and initially, only modified sine wave units could be purchased at reasonable prices. Modified sine wave inverters put out a “rectangular” wave, which isn’t anything like a sine wave (more on this later) and is not suitable for standby testing. It can also cause connected appliances to be stressed by the harmonics and fast-rise of the waveform, and cause them to buzz and hum. Sensitive electronics are best suited for Pure Sine Wave units.
The synthesis of a pure sine wave is most commonly done through pulse-width-modulation (PWM) methods, which switches the voltage at many multiples of the actual output frequency, and restores a sine wave out of “pulses” by smoothing it through a filter (commonly inductor-capacitor based). It is possible to generate fairly good outputs this way, although the harmonic content could be troublesome if the filter is poorly designed!
On the back of the box, there is some basic specifications – as far as I know, this unit is a 12v input with 220-240v AC output. It claims an efficiency of >90% and >85% full load efficiency with over-discharge warning at 10.5 +/- 0.5v, and switch-off at 10 +/- 0.5v. It claims overloading, short circuit and thermal protections, which are pretty common claims for inverters.
The actual applicable specs are ticked on the bottom. Note that it does claim to have a 240v output and a 50Hz +/- 5% output.
Internally, the unit is a rectangular brick, with an aluminium body and steel plate ends. The top is emblazoned with a label, with a model number of HIP-300, which suggests this inverter was made by Ningbo Honghui Electrical Appliance, although this exact model isn’t visible in their list.
The underside of the unit seems to have a serial number but is otherwise bare.
The front has two LEDs, one indicating power on status, and the other indicating error. One socket is provided, as is a slightly-flimsy feeling power switch. A vent grille is visible, and a large inductor wound on a toroid is visible through the grille. This may form part of the output filtering circuitry.
The rear of the unit has a fan output with grille for cooling, with the screws slightly rusted, and two binding posts for power connection. Unfortunately, these weren’t of the banana-combination type which would have made it easy for me to use banana to banana cables to hook it up to my power supply.
There are screws on both sides, where it is ribbed, which implies that the casing of the inverter is used as a heatsink, which is a very common arrangement.
The unit is supplied with a manual, and a bundle that includes two cables – one is a cigarette lighter to spades (unfused) with thinner wire and a warning to limit draw to 150w maximum, whereas the other is a thicker clips to spades cable which allows for you to directly connect it to a battery.
Taking this thing apart was quite simple, mainly because the rectangular aluminium shell around the inverter is not a single piece. To take it apart, you can remove the four screws on the power-plug end, tilt the panel away from you and slide off the top cover entirely.
The internals consist of one PCB which slides into the rails in the bottom chassis, and a vertical PCB which is soldered to that one (which is likely the controller itself, and is changed when the unit is manufactured as a modified sine wave or true sine wave model).
Visible on the left is the binding post inputs, and the fairly powerful fan. The input is routed to a pair of electrolytic capacitors, a high 35A rating soldered fuse-clips and a reverse polarity “crowbar” diode (which should blow the fuse given a source with low enough internal resistance).
There is a pair of transistors at the bottom, and two pairs of transistors at the top. I am thinking the bottom pair handles the DC chopping into high voltage DC, through the main transformer core, which is then stored in the large capacitor which is approximately near the middle of the screen with the Pass QC3 label on it.The capacitor doesn’t seem to be of any reputable brand, it seems.
The top two pairs handle chopping that DC into AC with PWM, with the two transistors to the left possibly being the drive transistors that drive the two pairs. This is then filtered through the polyester caps and large inductor, and again filtered through the isolation transformer (next to the capacitor) before becoming the output.
On the bottom PCB, there are two visible ICs, an ST LM324N quad op-amp and an AC08AA. There are also three precision trim-pots and a buzzer for warning. It is quite possible that the trimmers control the output voltage, and a variac might not be needed, but I wasn’t going to play with it and potentially destroy it. Near the vertical PCB, an opto-isolator is visible too for feedback.
The vertical PCB houses a PIC16F716 microcontroller, and a pair of International Rectifier IR2110 MOSFET/IGBT drivers. There’s a third HA17393 comparator IC as well. The top right corner seems to show a 3.9v zener diode, which may provide a voltage for comparison.
One of the units I received, also had a cracked ceramic capacitor behind the vertical PCB. The unit still seemed to work, so I kept it, but it;s a sign of very rough handling or poor construction techniques.
Specifically of interest is that the ground pin is connected to the inverter chassis. I suppose “you have to connect it somewhere”, and it’s better than connecting it to the battery negative (which could make a whole car “live”), but if anything does go wrong with your appliance, touching the inverter shell is probably not a wise idea.
Will it Work?
From an initial glance, this unit has a few features which could make it problematic as a power source for standby energy testing. For one, the output voltage is claimed to be higher than the 230v nominal harmonized line voltage of Australia, so some voltage adjustment might be necessary. The tolerance of the frequency at 5% is much bigger than the 1% required by the standard. Likewise, the quality of the power highly depends on the quality of the filtering on the output. There are many variables which all need to align for this to work!
As the cost of the unit was only about AU$50, it was a pretty good deal for a pure-sine-wave unit even if it wasn’t suitable for standby testing, so I took the plunge. In fact, I took the plunge twice, as I decided to get two units, and one of the units came with a rattling transformer core, and diminished efficiency, but was still meeting specifications. I suppose you get unlucky sometimes … especially when postage is concerned.
The first thing to do was to power the inverter up. As I would like to use it nearly continuously, using it on a battery wasn’t a good option. Instead, I hooked it up directly to a high-powered (20A) Manson benchtop switch-mode power supply. By using it to provide AC to DC, and having the inverter then change the DC back into AC, we achieve a level of isolation, which should mean that mains variations shouldn’t affect the output voltage at all, or as much.
It also improves safety, as the output of the inverter is current limited, and the total power output can never exceed the amount of power the Manson power supply can deliver, which is only 276w (minus a little bit for efficiency of converter). Overall, a large inverter isn’t necessary for standby power consumption checking as we want to measure loads which “sip” roughly 1w or less!
So, how does the output look like?
The waveform is a true sine wave, no doubt about that. Looking at the harmonic graphs, there isn’t too much energy in harmonics, although there is more energy in the upper harmonics than in real mains electricity and this comes down to the fact that this is PWM generated (rich in high frequency components that need to be filtered out). You can compare this to a modified sine wave inverter below (ringing is due to the test instrument having limited harmonic data):
The raw numbers which are relevant to IEC 62301:
|Parameter||Modified Sine Wave||True Sine Wave|
It’s pretty interesting, because it shows that the modified sine wave inverter I have is pretty lousy at keeping voltage regulated as well. Anyhow, the true sine wave inverter doesn’t look like a 240v model, but it is a little high for 230v, so a variac is needed. I actually built a 1A variac based on an open frame unit so as to trim the voltage. The frequency, is good as I would expect – as it’s likely to be generated by microcontroller based on the crystal input which generally only varies a few hundred ppm. The crest factor is pretty close to ideal (1.4142) and well within range.
Using it for standby power consumption runs worked fine provided the unit was left to warm up for 10-20 minutes before starting. It seems that as it warms up, the output voltage can make ~2-3% jumps but it’s all fine once it’s reached temperature. The fan itself doesn’t seem to be thermostatically controlled, and is instead activated dependent on load. For small loads below about 10w, it’s not uncommon to see the fan never spin up, but the transformer inside does get quite hot. There hasn’t been any failure to date though, so I suppose it’s okay.
Under heavy loads (120w+) the fan starts to spin quite loudly and it seems effective as I’ve run a computer off of this setup for hours with no trouble either.
Definitely a cheap fix to a potentially expensive problem, chalk this one up to a “hack”.
Despite my initial reservations as to how good a low cost pure-sine wave inverter could be at being a synthesized mains source for IEC 62301 Ed.2 standby power testing, my experience seems to have proven me wrong. Using the inverter connected back-to-back with a benchtop power supply produced an isolated output which was relatively insensitive to mains variations. It worked a treat, provided the output voltage was trimmed with a variac and the unit was left to warm up initially. A cheap hack, sure, but I don’t think the manufacturer of the unit intended it for this use anyway. At least it allowed me to evaluate many appliances under IEC 62301 for my own interests and give the Tektronix PA1000 a good run.