@IDEngineer
I had a cheap Aldi 2.8KVA/2.5KW 3000rpm AVR gasoline emergengy genset a few years back which I'd hoped would supplement my APC SmartUPS2000 backed protected supply. Unfortunately, the 2KVA UPS reacted very badly despite configuring it to its "Dirty Mains" settings (a smaller APC SmartUPS700 could tolerate the genset's "Dirty Supply" but this wastrel weakling was due to be given the sack - it consumed 20W to maintain just itself even with the 24v battery pack disconnected, a high price to pay for a lousy 450W max continuous inverter output rating).
I'm afraid to say I had bought into the lie of "poor, noisy, distorted and unstable frequency output" of these AVR gensets so it took quite a while for the penny to drop as to the actual issue. I had examined the output waveform using an ancient 1940s boat anchor scope (its 1v calibration source rather conveniently was just a low voltage output secondary winding on its mains transformer which avoided the need for connecting another low voltage output transformer by which to safely see the mains voltage waveform).
I had previously compared the 50Hz 240v mains supply against the inverter output of the SmartUPS2000 revealing the now classically distorted utility supply (a little flat topped with a small amount of slope towards the zero crossing point on both positive and negative peaks which I now know to be an artefact of the AC coupling in this boat anchor 'scope - a modern DSO only shows this when AC coupling is deliberately selected otherwise it's a dead flat top form of distortion in DC coupled mode).
When I disconnected the mains input to the UPS to examine its output waveform, the wave trace turned into a shining example of the classic sine wave (with just a low level of 5KHz or so PWM ripple riding on the trace). When I repeated this comparison between mains and the genset output waveforms, the genset, funnily enough looked almost identical to the UPS output. There was a subharmonic 25Hz component due to the use of a 4 stroke single cylinder gasoline engine directly driving the generator head, complete with a high frequency 'slot ripple' similar in frequency and amplitude to the PWM ripple of the UPS's inverter neither of which effects (including the 53 to 48Hz variation with loading) were the slightest bit troublesome to the UPS.
Curious about the possible effect of the genset's slot noise ripple, I wired a spare 4.7uF fluorescent lamp ballast PFC capacitor across the genset's output in an attempt to attenuate the slot noise component only to be greeted with the 240vac output rocketing north of 270vac!
It seems you don't need much capacitive loading on these AVR gensets to render the AVR completely helpless against the self excitation effect induced by this leading current loading.
Close examination of the circuit diagram I had managed to track down for the SmartUPS2000, revealed the presence of a pair of 4.7uF caps in the mains input circuit which under mains fail conditions got disconnected from the incoming supply when switched to battery power, neatly explaining the horrible interaction between genset and UPS whereby it would keep cycling between generator and battery power until the genset was disconnected and mains power plugged in to give the battery pack a chance to recover before becoming completely depleted.
Although I had considered using a 275v autotransformer tapped for 240v output with a 4.7uF on the genset side to force it into saturated output mode to provide a more stable 240v feed, I was concerned that the increased field current would shorten the life of the generator head so I sold it on, ending my plans for a more robust emergency back up supply until I could get hold of an inverter genset at a less eye watering price than those Honda units.
Now I don't believe this behaviour with modest capacitive loads and AVR gensets is unique to the particular brand that had let me down so badly, so it seems to be a glaring omission in the list of reasons for problems that arise in their use for "Sensitive Electronic Loads", hence my tale of woe to put the record straight.
I did eventually find myself a real bargain in inverter genset technology two or three years after that unfortunate episode, almost two years ago now. It took the form of a Parkside PGI1200B2 1200W peak (30 seconds timeout) 1000W continuous "Suitcase" generator purchased from Lidl for a mere £99.95 (about 10% of the price of the cheapest Honda inverter suitcase genset). I'd have preferred a higher output rating but, at that price, beggars can't be choosers, besides which, I was curious enough to make the modest 100 quid investment to test my theory that only the inverter genset designs were up to this particular task (free of the overvolting effect of capacitive loading) to unequivocally answer this question.
Long story short (the first two had show stopping, yet as I later discovered, trivially easy to fix, stock faults), this inverter genset
did solve the problems I'd had with my first generator purchase. Also, with virtually all of the lamps in the house now being LED types, the 1KW (995W actually) is actually quite sufficient to maintain the IT kit
and keep
every light in the house burning.
I did try out three examples of the rather shitty Workzone 2KVA inverter gensets being sold in Aldi stores about six months later but they all proved to have an unacceptable response to changes in loading (almost completely stalling with a step increase of loading even when preloaded to 50% before applying a further 40% of loading).
Those little Parkside generators showed hardly any drop in engine speed when their loading went from 0% to 100% even when switched into 'Eco mode"(after a few minutes of warm-up - I did manage to stall one by not allowing the extra couple of minutes warm-up time before switching to 'eco' mode).
Those Workzone units had no eco mode switch option - it ran permanently in 'eco' mode meaning you'd have no choice but to wait the several minutes warm-up time before taking the risk of applying a load - unlike the Parkside unit which could be loaded within 15 seconds or so of startup in 'normal mode' before making the switch into eco mode, plus the ignition kill switch was integrated into the fuel shut off tap, preventing the carb float bowl from conveniently being run dry, unlike the Parkside's separate fuel shut off tap and ignition kill switch arrangement which allows such convenience.
The only irritating omission with the Parkside being the absence of a fuel lift pump priming lever to save the need to crank it with the pull starter with ignition off leisurely enough to save your strength before attempting an actual starting yank in the hope that it starts "first go" without it snatching back.
Incidentally, I did toy around with the idea of installing an electric start circuit to drive the PM alternator as a BLDC starter. I even got so far as to invest 16 quid in a 60v rated 300W electric scooter BLDC control unit with both Hall effect sensor and sensorless control options. I could get it to spin at about 480rpm using a 48v battery pack but only by removing the spark plug. Even blocking the spark plug hole with my finger whilst it was running was sufficient to stall it, proving that sensorless operation wasn't up to the task of handling such erratic mechanical loading.
This task would require hall effect sensor input physically synchronised to the rotor position or a shaft encoder that could simulate such sensor inputs with better accuracy and consistency than actual hall sensors embedded into BLDC motors during their manufacture. Such sensors do exist and are even affordable (around 15 dollars US complete with disk magnet to attach on the end of the motor shaft). However, I'd have to determine whether the PM alternator is a 6 or 7 pole pair design (18 or 21 stator poles) and deal with how best to attach the magnet and mount the detector board. This was a lot more time investment than the purely electronic project of a sensorless setup I'd first envisioned and hoped would suffice so I called a halt to the whole project at that point.
Using a 60v rated BLDC controller with a BLDC that could be run up in very short order to generate some 280 volts rms back EMF would call for additional protection circuitry (isolating relays) to save frying the controller, making the project a whole lot more complicated than I was prepared to invest any further time into.
All I have to show for my efforts so far is that the concept of making the PM alternator pull double duty as both a generator and a 'starter motor' looks to be a practical reality given enough time and attention to all the nitpicking details involved.
Should I ever decide to try and take this project into any further development, a microphone by the spark plug hole to generate a timing datum point signal by which to count pulses from one of the stator coil connections on a DSO display by cranking with the pull cord will answer the pole pair count question and, if there's enough flux leakage on the outer surface of the flywheel, I might be able to lash up a set of 3 hall sensors salvaged from my retrieved PC AT and ATX cooling fan collection to mount just clear of said flywheel to generate the required rotor locked source of commutation pulses for the BLDC controller.
Only if this overcomes the stalling issue with the spark plug fitted can I then look to solving the issue of disconnecting the controller swiftly enough on a successful firing of the engine to eliminate the over-voltage burnout hazard and then deal with how best to supply the 48 to 56 volts required to drive the whole shebang should the project ever reach this penultimate stage.
As interesting an exercise in DIYing an electric start add-on to a cheap as chips inverter generator as this would be, I might just 'cheat' and throw a wad of cash at the problem by buying a good quality higher output electric start inverter genset and have done with the whole "Blood Sweat and Tears" commitment which could just as easily become a folly as a success.
When I was searching the internet to see whether anyone else had had a similar stroke of genius in using the PM alternator as a BLDC starter motor in inverter gensets, I came across a 20 year old patent application by Black and Decker on a variation of my own add-on starter circuit. In this case, they cheated a little by using a modified alternator with low voltage taps to facilitate the direct application of the 12 or 14.4 volt starter battery voltage. Their interest seemed to be in the way of expanding the market for their 14.4v cordless tool battery packs.
I suspect this no-brainer electric start option in even small cheap inverter gensets has been held back by Black and Decker's patent application. Once that patent has expired (less than 5 years to go?) we may well see the market flooded with cheapish inverter gensets in the 1 to 3 KW range all offering an overpriced electric start upgrade option for the cost of a cheap 7 or 12AH 12v SLA and a license code to enable this built into the inverter module function, possibly with a luxury 12.8v 4 cell LFP battery pack option for those who'd like to test run their gensets once a week over the next four or five decades in between battery changes.
Anyhow, I digress (and then some!)
John