UPS-1000 Uninterruptible Power Supply from Novatek-Electro

My history with uninterruptible power supplies goes back to the late 1990s and early 2000s. And not as a user, but more on the manufacturing side. That’s why I was doubly interested in seeing what’s being made today and remembering how we used to do it. So this review will probably have a touch of nostalgia.



There was a time in our country when so-called “rolling blackouts” were a thing. Where I live, the power was cut off at least twice a day, for two hours each time. I wasn’t exactly thrilled about sitting at home without electricity, but things were even worse in the suburbs. Money was tight, and the selection of commercially made devices was pretty dismal. You could get either an expensive APC unit or something from an unknown brand, but they usually had one thing in common: they couldn’t run on battery for long, simply overheating.
In any case, I first had to build myself an inverter (we already had a standalone 12 V system at home, by the way), then add automatic switching between mains and battery. An acquaintance of mine was interested in the idea. He was good at writing microcontroller programs (back then, in assembly), so we decided to develop our own UPS.
The main differences from what was on the market:
1. It could run on battery indefinitely; in other words, runtime was limited only by battery capacity.
2. It could charge the battery at no less than 5 A, so it could recharge it as much as possible while the power was on. Typical UPS units charged at 1 A.
3. It had a “cold start” function, so you could turn on the UPS even when there was no mains voltage.

Over time, this evolved into a modular design consisting of a power board, processor module, charger module, and display board. The processor and display modules were the same across all product lines. We started with 300 W models, then made 600 W models (12 and 24 V, 5 or 8 A charging), followed by 1,200–1,500 W models (24 V, 10 A charging). We also had models for heating boilers and voltage stabilizers. I remember that people in the suburbs were especially grateful. We even had regular customers who collected orders in their area and then bought from us. It was quite a “fun” time.

Out of curiosity, I tried to find at least one photo. I remembered taking one with a digital camera I’d bought back then, right at the start of the 2000s, and I actually found it. I also found the materials we’d developed at the time, with descriptions, manuals, warranty cards, and even the name of the company that mainly handled sales: NPF Ekodan. But there’s only one photo.


We also developed various timers and protection devices. We even made an eight-channel battery tester with galvanic isolation between all channels, capable of charging and discharging batteries and measuring their capacity—lead-acid, nickel-based, and even lithium-ion—in 2003. All the information was displayed on our own dot-matrix display. As for microcontrollers, we used either Microchip PICs or Atmel’s 90S1200. We went with the latter and built everything around it, including an ADC, I think 8-bit. And the transistors were IRFZ44 (46), or the newly released IRF3205, and... that was it.

All right, folks, I won’t bore you with too much more. I may remember a few more things to compare as I go, but for now, let’s get to the review.

So, I received a box of impressive size and even more impressive weight: over 11 kg. And that’s despite the fact that there are no batteries included; the UPS isn’t even designed to have them installed.
No complaints about the packaging. Everything is well thought out, even how to get the unit out of the box.


The unit was also wrapped in film for extra protection. The package includes only the UPS and the manual. You could suggest including a set of fuses, but in practice there’s no point: even if they blow, the included spares will probably have been lost by then.


The manual contains all the necessary information on the technical specifications, charging details, settings, runtime calculations, and so on. It also states that the warranty period is 3 years.


For a closer look, you can use the electronic version. I’ll periodically include screenshots of the key points, starting with the technical specifications.


The manufacturer describes this UPS as a line-interactive, continuous-duty unit with a sine-wave output when running on battery.
I don’t entirely agree, though, because a true online UPS always runs its inverter, which is why these units are classified as online. In this case, the specifications state a transfer time of no more than 5 ms.
Line-interactive means that it can interact with the mains. Put simply, it has a voltage stabilizer inside and, before switching to battery, tries to compensate by raising or lowering the output voltage through transformer tap switching. In normal operation, the transformer powers the charger and switches taps to act as a stabilizer; when running on battery, it steps up the voltage after the inverter. That's why the phrase “continuous operation” sounds a little odd to me in this context.
Maybe I'll write about the different UPS topologies at some point, how they differ and which one is best for particular situations.


The exterior is more than classic: a large metal box with an indicator panel. Though I suppose the panel is what makes it less “classic,” since UPS units with lots of indicators are becoming increasingly rare.


There are ventilation slots on the sides, positioned toward the front to improve cooling from the airflow.


The front panel is large, with lots of indicators—probably even too many. In my opinion, this would be better served by a seven-segment display or a small 2004-type LCD instead of discrete LED indicators. That would let you display all the information, including the option to set the battery capacity and calculate the estimated backup time.
There’s just one control button, on the left. Fortunately, there’s not much to configure, and you only have to do it once.


Front panel description.
• 1–5 BATTERY: battery charge level indicators;
• 6–10 OUTPUT: UPS load indicators;
• 11: charging current indicator (15 A) and battery type indicator;
• 12: charging current indicator (12 A) and battery type indicator;
• 13: charging current indicator (8 A) and battery type indicator;
• 14: UPS status indicator (steady light (CHARGE) – battery charging; flashing
(DISCHARGE) – battery discharging);
• 15 LINE: AC input status indicator (steady light – mains supply is normal;
flashing – mains voltage is outside the 150–280 V range; off – AC mains supply
is absent);
• 16 FAULT: UPS fault indicator, also indicates a short circuit at the output;
• 17 ON/OFF: UPS on/off button, also used to set the battery type and charging current


The back is much more modest: a fan, a socket for connecting the load, a pair of fuses, and wire terminals.

A quick note about the UPS model markings. There are three versions: UPS1000, UPS-1000L, and UPS-1000LP. At first I couldn't figure out how they differed. It turns out the difference is the socket type: UPS-1000L has the German standard, while UPS-1000LP has the French standard. The one in the photo is simply UPS1000. I think the manufacturer should make the differences in the model numbers clearer, or mention them closer to the beginning of the description.

The fan is made by Sunon. I'm glad they didn't cut corners here.



The cables are fixed, so you could complain that the input cable can't be replaced. But it looks to be quite high quality, and extra plug connections don't improve reliability.
The battery wires have a cross-section of 10 mm², which is perfectly adequate for a maximum current of around 80 A. They have copper terminals at the ends. The cable is made by Tumen and is very stiff, though that isn't critical here. It may have seemed stiff to me because I'd just been working with a highly flexible, large-gauge cable with silicone insulation—but that costs a lot more, and there's no need for one here.


There are four rubber feet on the bottom, along with a label showing the device name, firmware version, and serial number. One of the screws on the left is countersunk, and you can also see traces of a paper seal. I assume that since the unit was sent for review, they simply didn't bother sealing it, because I would have taken it apart anyway. I'll share my theory about why it had been opened later.
Its weight and rubber feet make it quite difficult to slide across the table. Sometimes that even got in my way while moving things around.


Undo six screws and remove the cover to get inside.


The layout is very spacious, with plenty of empty room inside the case. In principle, with a bit of ingenuity, you could probably even fit a battery made from four 25–30 Ah LiFePO4 cells in here.


The PCB takes up the entire upper section, and it seems to me there are more components here than on products from the well-known APC brand.
The board has a handwritten marking, V11, while the sticker said V12.1. I suspect the firmware was updated before it was shipped.


The control section: microcontroller and voltage regulators for the auxiliary circuits.


Control is handled by an STM32F070 microcontroller. To the right, you can see a pair of logic chips and an op-amp.


There are also:
1. Another pair of MCP6004 quad op-amps
2. An IR2301 high- and low-side gate driver
3. Single-channel IRS21271 driver with integrated current monitoring.
4. Dual op-amp LM258


Overall view of the power section


I liked how well thought out the design is. To remove the board, you just undo the four screws securing the power wires to the heatsinks, then the four screws holding the board itself, and finally disconnect a few connectors.
This gives a clear view of one of the device's key components: the power transformer. The primary winding appears to be made from a copper busbar, and since there is only one, this is clearly a bridge inverter topology.

Inverters usually use one of two approaches:
1. Step up the low voltage to a high DC voltage using a switching converter, then use a transistor bridge on the high-voltage side.
2. Use a low-frequency transformer to step the low voltage directly up to the required high AC voltage.

The first design is more compact and cheaper, but personally, I think the low-frequency transformer design is more reliable and handles sudden load changes better. The greater “inertia” of the low-frequency transformer smooths out the surge currents through the inverter transistors.


Subjectively, the transformer seems generously sized for the power rating. For comparison, here’s a photo of an APC UPS that also has a sine-wave output, the same rated power, and a line-interactive design—in other words, essentially an identical unit, except it has a pair of batteries inside. The weights tell a similar story: 13.6 kg (with batteries) for the APC and 11.4 kg (without batteries) for the Novatek.




Top of the board


The heatsinks are aligned with the airflow, which improves cooling.
I mentioned above that this is a bridge topology, which implies four sets of power transistors, but there are five sets here. The fifth is for charging. The charger transistors are mounted on one of the small heatsinks, while the inverter transistors are arranged as one set on each of the two small heatsinks and two sets on the large one.
This arrangement is actually an optimization: the upper-leg transistors are mounted on the large heatsink, and the lower-leg transistors on the two smaller ones. This meant the designers didn't need to electrically isolate the transistors from the heatsinks, which also improves cooling and, consequently, reliability.


1, 2. Charger transistors: 1 × IRLB8314 and 2 × IRLB3813; the ASC712 current sensor is nearby.
3, 4. Lower-leg sets, each consisting of two IRLB3813 transistors and an STPS30L30CT diode assembly.


The battery-line fuses are visible nearby: three 40 A fuses connected in parallel. The fuses are plugged into holders, which are in turn soldered to the board.
You can also see the power-rail capacitor bank: 6 × 3300 µF, 16 V.


The bridge's upper-leg section. The temperature sensor is visible on the right, inserted into a hole in the heatsink.
Two sets are used here, each with 2 × IRLB8314 and 1 × STPS30L30CT.
At first, it seemed a little odd that the upper and lower legs use different transistors. My guess is that this may be because the upper transistors are controlled by drivers that monitor current by measuring the voltage drop across the transistors.


Part of the control section. As you can see above, the board around the transistors is not only double-sided, but also has additional copper from soldered areas.


Also installed here:
1. Four HRS4H-S relays, three of which most likely switch the windings in stabilizer mode, while the fourth switches between mains and battery. A 430 V varistor is visible nearby.
2. The load current sensor is also located here.


For comparison, here’s the circuit board from the APC equivalent shown above.



A brief digression.
I can't help thinking of my own UPS, which I've had for a long time and which I got second-hand. It has a larger case, but inside there's room for five 12 V batteries. It's also an online UPS, and that's partly why I stopped using it a while ago: not only did it need new batteries every three years, it also consumed around 50–80 W for its own operation. With the cost of replacing the batteries included, that came to about 60–70 dollars a year.
It has a few other quirks, too. For example, the load power indicator doesn't go up to 100%, but to 140%—the markings are 100, 120, and 140. Bypass mode switches on automatically when you turn off the UPS using the switch on the front panel. This UPS easily powered my desktop PC, a 19-inch CRT monitor, an external modem, speakers, and a laser printer. The transition to online mode when you switch it on is especially interesting. Connect an oscilloscope to the output: you see the distorted mains sine wave. Switch it on, and a few seconds later a relay inside clicks and the sine wave becomes clean.
If anyone knows what kind of UPS this is, I’d be glad to get links or any information.

Sorry about the photo quality. The UPS is in a cabinet, it’s heavy, and taking it out is a hassle. The photo below is actually five years old.


I really like its design, both electrically and mechanically: large heatsinks running the full length of the case, air intake at the front, two fans at the back, and lots of interference filters. By the way, its power rating is the same as the unit under review.


There’s another unusual feature: two COM ports, one male and one female. When I was figuring it out, I found that it most likely supports daisy-chaining or simultaneous signal output from several UPS units.

And here I think I should make a comment about Novatek. Sure, a UPS is great, but how did they manage to leave out outputs for even the standard signals—“mains failure” and “battery low”?
The lack of software doesn’t bother me at all. UPS units aren’t always used with computers (although modern devices are computers to one degree or another), but standard “dry contacts” can be very useful, especially since this option costs next to nothing. In principle, you could even tap the signals from the front-panel LEDs yourself, but that would be a bit... hacky.

Back in the early 2000s, we started building computer connectivity into our UPS units. It was incredibly primitive, but used standard signals, and worked perfectly well at the time. As far as I understand, Windows 7–10 can no longer do that.



After putting it back together, I decided to at least check that the UPS worked before running the tests. I connected a 12 V battery with a quick-and-dirty setup, even though you shouldn’t do that, and also found out that the UPS won’t start without a battery.
With a battery connected, you can start the UPS without mains power, so it supports “cold start.” My old UPS doesn’t have this feature.
One more thing: the LEDs are very bright. Since 4–5 of them are usually lit, and at least one is also blinking, they can get annoying at times.
There’s also a low-frequency hum while it’s running. It isn’t very loud, but it’s definitely noticeable in a quiet room. My old UPS didn’t hum, but its fans were noisy, so I had to replace them with quieter ones.

These indicators are usually lit during operation (some in the photo are illuminated by the flash):
1. Line — running on mains power
2. Discharge — lit while charging, then goes out and blinks when running on battery.
3. Charge — blinks to indicate the type of battery connected. I’d remove it altogether; I don’t see the point of indicating something that’s usually configured once and left that way.
4. One LED indicating the charge percentage
5. One LED indicating the load percentage, if it’s above 20%




A note on connecting the battery. The UPS can draw a lot of current from the battery, so there’s a good chance you’ll have several batteries connected in parallel. There’s a particular way to connect them: diagonally, as shown in the manual. This ensures that the current through each battery is the same, provided the batteries themselves have identical specifications. By the way, other brands’ UPS units designed for external battery modules usually connect them in a “daisy chain,” which isn’t quite right: the battery closest to the UPS will wear out faster.

The manual also says not to connect different batteries, though there are plenty of caveats, and under certain conditions it’s perfectly possible. But using identical batteries makes maintenance easier, since they wear out at the same rate.


For the tests, I used a LiFePO4 battery made from four cells, connected with short wires and bolt-on terminals.


A brief digression. Above, I mentioned that the UPS hums while running—or, more precisely, that the transformer hums. I think many of you are familiar with the usual 50 Hz hum of a BJT power supply. It’s the same here, but there’s a catch. Apart from the hum itself, which you quickly get used to and eventually stop noticing (though, as I said, you’ll still hear it in an apartment), there’s another sound. While running, the UPS makes clicking sounds. They can come in pairs, like 1.1...1.1......1.1...1.1..., or as single clicks, and those are more annoying than the hum. And they don’t just happen while the battery is charging or in standby; they’re still there even if you disconnect the UPS from mains and switch it off using the front-panel button. As the battery voltage drops, the clicking frequency changes slightly, and after a while the clicks stop. I’d say they happen periodically, but it would be more accurate to say that they periodically don’t happen.

In the photos above, you saw a fan. Its operation depends on the temperature, so it’s usually off in standby and typically turns on while the battery is charging or the inverter is running. That's a plus.

Since the UPS has no software and no built-in battery, I’ve had to cut the tests down a little. Even so, they took quite a while.

First, I checked the current draw when the UPS is off (from the mains and using the button), and when the inverter is running with no load connected. The electronics draw about 9–10 mA in standby, while the inverter draws 2.5 A with no load.


Charging current and power consumption from the mains.

This section will be a long one. In my view, the charging mode is at least as critical as the conversion mode: incorrect charging wears out the battery very quickly, and sometimes the battery can cost as much as the UPS itself, if not more.

The UPS supports three charging current settings: 8, 12, and 15 A. That’s a huge plus. The APC equivalent I mentioned above, with its 0.9 A, can only watch from the sidelines. But these charging rates come with limitations: forget about using a pair of 7–9 Ah batteries connected in parallel. As I’ve mentioned before, batteries like these have a maximum charging current of around 2.5–3.5 A, so even two of them would be overloaded. This UPS is designed for high-capacity batteries.

Selecting the charging current is very simple: connect the batteries and plug the UPS into the mains. The default charging current is 8 A. To select another setting, press and hold the button while the UPS is on until it starts cycling through the settings—8, 12, and 15 A. Release the button at the desired setting. The UPS will switch off; then turn it back on.

1, 2. 8 A mode; maximum power draw from the mains was 163 W (at the end of charging).
I tested it with a LiFePO4 battery, whose end-of-charge voltage is lower, so the maximum power draw from the mains is also slightly lower. In AGM mode, the maximum was 170 W.
3, 4. 12 A mode
5, 6. 15 A mode
7, 8. After the battery is fully charged, a small trickle current is maintained. The UPS itself draws about 13–14 W from the mains, which is very good, since high power consumption was one of the reasons I stopped using online UPSs.

The charging currents are slightly higher than specified, but that’s not a big deal. The UPS’s own power consumption is low, which is an advantage. I can confidently give it five stars here.


The charging voltage maintained across the battery and the charger’s operating algorithm are just as important as the charging current. Most often, charging uses the CC/CV mode, where CV is the battery voltage in float mode.
Incidentally, batteries usually have two operating modes: cyclic and float. Their voltage ranges are typically listed both in the documentation and on the battery case.
This UPS supports three battery types: AGM, Gel, and LiFePO4. In fact, AGM and gel batteries use the same VRLA technology, which also includes so-called multi-gel (MG) batteries. These batteries differ in their operating voltages, both in cyclic and float modes.
There’s a lot of confusion around VRLA batteries in general. Some manufacturers specify one thing, others something completely different. So before testing, I searched online and picked four batteries as examples to illustrate their operating modes. The selection was simple: ideally, the battery had its type marked on it and a reasonably good photo available. Here’s what I found:
1. Azbist, ASAGM121000M8. Float mode: 13.7–13.9 V (13.6–13.8 V; two different values are printed on the case). Cyclic mode: 14.6–14.8 V
2. CSB, GP12650. Float mode: 13.5–13.8 V. Cyclic mode: 14.4–15 V
3. LogicPower, LPM-GL 12 65. Float mode: 13.5–13.8 V. Cyclic mode: 14.5–14.9 V
4. Delta, GEL-12-75. Float mode: 13.38–13.61 V. Cyclic mode: 13.8–14.1 V

The first two use AGM technology, the second is a gel battery, and the third is listed as a combined AGM+Gel type, although I initially found it listed as a gel battery.
The voltage figures show that the float-mode range is around 13.5–13.8 V for almost all of them. Only the last type has a lower range: 13.38–13.61 V. The cyclic charging voltage is 14.4–15 V, again with the last type being different at 13.8–14.1 V.



The manufacturer lists two charging algorithms in the description. The one for AGM and Gel is the same, and looks quite reasonable: first, the battery is charged using cyclic charging parameters, then switched to float mode. In my view, this is the most appropriate algorithm.



A similar charging mode is generally recommended for LiFePO4 batteries, and it’s described quite well in the datasheet for the LT-LFP220-300 batteries. I also ran a similar test in another review, where I checked how charging time and delivered capacity depend on the float-mode voltage.



The engineers at Novatek-Electro took a slightly different approach here: they simply charge the battery to the float voltage and then maintain that voltage at a constant level. At least, that’s exactly what the graph in the manual shows.
It also says the battery is charged to 14.5 V, which works out to 3.625 V per cell with four cells. That’s high for LiFePO4 and puts it into cyclic mode; with a protection board, it can also cause false trips.


Here’s what actually happens.
The tests were simple: first, I partially discharged the battery, then started charging and watched for the peak voltage. After that, I waited for the Charge indicator to go out and started recording the maximum and minimum values. The thing is, charging doesn’t stop at that point. Instead, it proceeds in steps: one step up, two down. The voltage would drop to some value, rise a little, then drop again, this time slightly lower than on the previous step. And this goes on for a very long time.

The results are divided into three groups, corresponding to the three charging modes.
1. AGM: charge termination voltage of 14.67 V; minimum over three hours, 14.06 V; average, 14.12 V
2. Gel: charge terminates at 14.26 V; minimum over three hours, 14.02 V; average, 14.09 V
3. LiFePO4: charge terminates at 14.26 V; minimum over 12 minutes, 13.91 V; average, 13.92 V. I ran this test first, using a slightly different method: I hadn’t set the monitoring interval to three hours yet, and simply started recording the max/min values after roughly the same three-hour wait. But you can see that the voltage had stabilized, since it fluctuated by just 27 mV over 12 minutes.

One thing I noticed was that in AGM mode, the interval between reaching the maximum voltage and the Charge LED going out was noticeably longer than in the other two modes. Also, even after three hours, the voltage was still dropping, albeit in very small steps, and I simply didn’t wait any longer. It may have dropped further. I never did see the charger shut off completely.


Later, I ran another experiment: I turned off the UPS with its button and unplugged it. It turned out that the tests above were simply limited by the fact that the battery voltage was dropping very slowly. In LiFePO4 mode, this means the battery was being topped up, but only with very short pulses. During charging, the voltage changed roughly like this: +20 mV, -22 mV, +21 mV, -19 mV, +22 mV, -19 mV. But in the end, the battery had simply reached the required capacity, and the UPS was only compensating for self-discharge.
The photo below shows the voltage after charging was stopped, after 16 minutes, and after about an hour and a half.
By the way, when I switched off my old UPS, it automatically switched to bypass, but the charger kept running. So this time, just to be safe, I disconnected it completely.


To test my theories, I went a step further: I dug a box of used batteries out of the closet, connected them in parallel, and decided to try it that way.
I noticed that the voltage was swinging across a wide range, from 12.47 to 14.74 V (AGM mode was selected), and doing so very quickly. The reason became clear after a couple of hours: most likely, one of the batteries had an internal short in one of its cells. It’s clearly visible in the thermal image, so I had to disconnect that battery.

I’m sure someone will ask why I’m using batteries from some obscure brand. No, they’re all Panasonic. One of them is at the bottom; they were left over after a scheduled replacement of six batteries in one UPS, and the box had been lying around for ages.


After that, everything went normally. The UPS charged the battery, held it at full voltage for about an hour, cycled it up and down a little, then stopped charging completely. During that time, though, it kept topping it up with short pulses, which you can see in the measured voltage and the recorded minimum.

So, why did I measure all this, take photos, and write about it?
The thing is, this charger isn’t the typical “battery boiler” found in the vast majority of UPS units, including some decidedly non-budget models from well-known brands. That’s a good thing, because few manufacturers bother with this. They need to sell their battery modules, after all—and how are they supposed to sell them if the UPS takes good care of the batteries?

To put it very simply, the charging process looks something like this:
1. Charge the battery to 14.2–14.6 V, depending on the mode, and hold it at that voltage for a while. During the first charge, the voltage may briefly (literally for a couple of seconds) rise to 14.7–14.8 V, but that’s not critical.
2. Reduce the charging current almost to zero so the voltage starts to drop.
3. After a while, turn charging back on, and repeat this several times.
4. Stop charging completely, topping up the battery a little only from time to time.
5. If the battery discharges below some critical threshold, charge it back to full voltage—in other words, return to step 1.

In my opinion, this is a well-designed algorithm: it includes cyclic charging to get the maximum usable capacity out of the battery, a little exercise, and no “boiling” function.
The only thing that gives me pause is using it with LiFePO4 batteries, since they are used with protection boards and need to be charged gently so the built-in balancer has time to even out the cell voltages. If charging is too aggressive, the voltage on the weakest cell will exceed the critical limit and the protection board will shut down the entire battery.
But I’ll test the LiFePO4 functions separately, later on. I’ve already ordered the batteries.


This test is fairly uncommon, but useful for users of heating boilers. I checked whether both input and output wires are disconnected, or just one. I’ve come across UPS units that disconnected both the live and neutral wires when switching to battery power, which can cause problems with gas boilers and trip their protection.
Everything is fine here, although there’s a small detail I didn’t quite understand: over the course of about a minute, the resistance changed smoothly and in waves from 0 to 0.3 Ω and back.


Output voltage.
1. Mains power, no load
2. Battery power, no load.
3. Charged battery, 420 W load.

I also checked how it worked with a discharged battery, when the charge indicator was already red and the load was still 400–420 W. The output voltage dropped to 216 V, which is perfectly normal.


The charge indicator raised a few questions.
1, 2. After a small battery discharge and disconnecting the load, the indicator showed 75–100%, although the actual charge was closer to 90–95%.
3. When I connected a load, the charge indicator immediately dropped to just 25%.
4. Voltage at the UPS terminals with no load connected (only the UPS's own power consumption).
5. Voltage at the UPS terminals with a 420 W load
6. Voltage at the battery terminals with the same load.



And here I’ll come back to the difficulties of measuring the charge level of LiFePO4 batteries.
On the left is the discharge graph for a lead-acid battery with 12 mΩ resistance at 23 A; on the right is the LiFePO4 graph at 40 A (lower curve), with battery resistance of 6–7 mΩ. So we can say the batteries are under roughly the same load.
But the LiFePO4 voltage clearly drops more slowly. During the first 10% of discharge, the lead-acid battery is at 11.9 V, while the LiFePO4 is already at 12.8 V (which roughly matches the measurements). So it was strange to see the indicator show 25% charge when the battery was charged.
Ideally, with LiFePO4 you should set the battery capacity and reset it after every full charge so the Ah or Wh counter works correctly. But that’s complicated, and I’d simply check the indicator’s calibration.

Also, the UPS has cables about a meter long, so the total length is 2 m. With a cross-section of 10 mm² and a current of about 40 A, the voltage drop will be 3.3 mΩ × 40 A = 0.132 V. In other words, at that current the UPS was receiving 12.37 V, not 12.5 V, so the indicator could have been off—and the protection board would add to the drop as well.


By the way, about current draw and efficiency: with a 420 W resistive load (a plain old iron), the current in the battery circuit was 42.16 A. At 12.5 V across the terminals, that gives an input power of 527 W and an efficiency of 80%.
According to the UPS indicator, it was loaded to 60%, which makes sense given its rated power of 600 W.

One useful feature worth noting: when switching to battery power, the UPS beeps twice and then stays quiet. That’s quite sensible for UPS units designed to run for hours. It’s also worth mentioning that this UPS has no “green” mode and works perfectly well with no load, without shutting itself off automatically.


Just to be sure, I repeated the test in a more illustrative format at 200 W and 420 W loads. The efficiency was again around 78–79%, which isn’t very high, though it comfortably exceeds the stated 75%. The low supply voltage is a factor; UPS units running at 24 V or higher usually do better.
Interestingly, APC claims 95–97% efficiency for a similar UPS with a 24 V battery under the same conditions, and something tells me that figure is a bit optimistic.


As for power, the UPS has indicators for 20, 40, 60, 80, and 100%. None light up with no load or a very small one; with a 200 W active load it shows 40%, with 420 W it shows 60%, and with 620 W it shows 100%.
This reminded me of the UPS units we had back home. I mentioned above that they were rated at 300 W, 600 W, and so on—that wasn’t a typo; their ratings were specified for active loads.


Testing the battery low-voltage cutoff threshold. The battery is rather weak and its voltage dropped quickly, but I noticed that the battery indicator quickly turned red even though the battery voltage was still at least 11.3 V. In the first photo, the readings were taken four minutes after the test began, but in reality the indicator entered the red zone much earlier.
I was pleased that even with the battery voltage at 10.6 V, the output was still 229 V.
The UPS shut down when the battery voltage reached 10.48 V. I think it was about 10.6 V right at the UPS input.


Checking the operating voltage ranges.
Here I tested operation across the input voltage range, from the lowest possible voltage up to 250 V. The upper limit was set by the capabilities of the Variac.
The UPS has two step-down and two step-up stages. It can operate with the voltage dropping to about 150 V (under a light load), after which it switches to battery power. It switches back to mains at an input voltage of 160–170 V. I couldn't pin down an exact relationship, but it seemed that the longer it ran on battery, the later it switched back to mains.

When the input voltage rises, the UPS starts stepping it down at around 248–250 V. Below that, it operates in bypass mode. Based on my estimates, it should switch to the next stage at an input voltage of around 280 V; if the voltage rises further, it will switch to battery power.



Input voltage rising from 130 to 250 V; the output stayed within 209–245 V.


The same thing in motion.


Input voltage dropping from 250 V to the threshold for switching to battery power. The output voltage stayed within 206–230 V. The actual range was a little wider; the logger records more slowly than the display updates.


And a clearer view.


And now the UPS output waveforms, along with its response in different operating modes.
1. Mains power, stabilizer off. Yes, this is what my mains supply looks like—not pretty.
2. Running on battery with no load: clearly much better.
3. Connected a 420 W resistive load; not much changed.
4. Here, a switching power supply without power factor correction is connected, at 420 W. The waveform has become trapezoidal, PF = 0.72–0.75.
5. Same type of load, but reduced to 200 W. If it improved at all, it was only slightly.
6. Combined load: 200 W resistive + 200 W switching power supply with PF = 0.72–0.75.

Of course, the waveform with a switching power supply without active power factor correction as the load was predictable, especially with the load at about 70% of the maximum. Still, it looked pretty bad. I was much more interested in how it would respond to a reactive load, but when I started looking around the house for something to create one, I suddenly realized I had three options, and none of them would work:
1. A small fan—too low-powered.
2. The kitchen range hood. It uses a triac controller, so it's unclear how it would behave during transients, and the validity of the test would be questionable anyway.
3. The refrigerator. It works fine under normal conditions, but the compressor startup would overload the UPS.

I'm still looking for something else, but that will have to wait for part two of the review.


Transients when switching from mains to battery. Here and below, I use two types of load: a 200 W incandescent bulb and a switching power supply with a 200 W load and no power factor correction.
The manual specifies a switching time of up to 5 ms, but I measured longer. I ran each test several times, but the results varied every time.
1, 2. Resistive load. Switching doesn't always take 5 ms; for example, the left trace shows 12 ms, and the UPS didn't settle into normal operation immediately after switching. On the right, it settled correctly, but the switching time was still 7 ms.
3, 4. Switching power supply. In the first case, it was about the same 12–13 ms, and it settled correctly. In the second, the switching time was the same, but the waveform as it settled into operation was strange: normal at first, then a voltage dip and recovery.


Switching back, from battery to mains.
1, 2. Resistive load; switching took about 7 ms.
3, 4. Switching power supply. You could say it looks very good: the waveform (if you can call it that) wasn't interrupted.

Things were noticeably better in this mode because the UPS initiated the switch itself and could time it to the right moment. Even so, with a resistive load it slightly exceeded the specified 5 ms.


Transients in voltage regulation mode.
1, 2. Input voltage drops, so the UPS switches to boost mode. The left trace shows a 200 W resistive load; the right, a switching power supply.
3, 4. Switching from boost regulation back to normal operation. The transition looks really ugly, and this kind of wobble appeared in every test. The traces shown here and above are average results. Here, too, the active load is on the left and the UPS on the right.


Curious, I ran one more test, this time with only the active load. This shows the transition from normal mode to step-down mode and back. The input is 250 V on the left; the UPS steps the voltage down, and the reverse process is shown on the right.
You can also see the transient when the UPS switches to passing the input voltage through to the output.

In my opinion, transients like these are not a good thing. This may be due to the quality of the particular relay, or perhaps to the lack of output interference-suppression filters.


Now for the hardest part: the conclusions.
What I liked.
The UPS can work with LiFePO4 batteries and has a high charging current of up to 15 A, making it suitable for use with high-capacity batteries and allowing it to restore their capacity quickly after discharge. Put simply, after an hour of running from the battery at full load, the UPS can replenish the capacity used during that hour in 4 hours. The APC UPS mentioned above would take 4–8 times longer.
The battery charging algorithm is correct—just like the ads say, with no forced “boiling.” On top of that, the external battery doesn’t heat up from the UPS itself.
Heat. I didn’t run any tests here, but I can say that I heard the fan running only once during the entire time: while the battery was charging for an extended period at 15 A. The fan also didn’t start during 15 minutes of operation at about 35% output power.
The voltage stabilizer has four steps, not two as is often the case, so it can operate without switching to battery power over a range of 150–280 V!
The stated operating temperature range is from -5 to +40 degrees.
Build. There’s a great Russian word for it: “oak-like,” in a good way. I got the impression that this UPS is actually about one and a half times more powerful than the manufacturer claims, especially judging by the size of the transformer. As they say, heavy is good; heavy means reliable.
There’s a built-in buzzer. I heard it only in two situations: when switching from mains to battery (two beeps), and just before shutdown due to battery discharge.
There’s no “green mode” with its various smart power-saving shutdowns.

What I didn’t like.
The switch to battery power takes too long. I measured about 7–12 ms, compared with the stated 5 ms. I’d hoped for better, especially since I liked everything else I’ve mentioned so much, but the oscilloscope traces are quite conclusive.
There’s a transient when switching from voltage boost/buck mode to direct pass-through. Additional filters or snubber circuits across the relay contacts may be needed.
Efficiency is higher than stated, but I’d still like it to be better.
Noise. It may be within the stated limits, but the odd continuous clicking is somewhat irritating. It’s there even if you turn the UPS off completely and leave only the battery connected. Still, given that the UPS is unlikely to be kept in a bedroom, I’m probably nitpicking. I later found out that the fan was making the clicks; for some reason, it’s being supplied with a small amount of power.

What I think should be added or changed.
1. Indicators and controls. The flashing LEDs were irritating, especially when the battery reached its lower threshold while running on battery power: three LEDs flashed at once, each at a different rate. It was like a Christmas tree. Why not use two-color LEDs to indicate the operating modes? Indicators usually flash to attract attention—for example, in the event of a fault or mains failure—not during normal operation. I could suggest using seven-segment displays, of course, but that might be overkill.
2. External signals. How can a UPS have no outputs for reporting its status? I understand that not everyone needs this, but I think enough users would refuse to buy it for that reason alone. Dedicated software isn’t essential either (though it would be nice), but surely they could at least provide dry contacts.
3. A 24 V version would be useful, I think, since running currents of 50–60–70 A through wires for extended periods isn’t very efficient.
4. Support for LiFePO4 batteries is genuinely great, and that alone gets 5 points. But there’s a catch: these batteries require a protection board, and such boards don’t always reset themselves after an over-discharge cutoff. That could leave us with a nonfunctional UPS and a discharged battery, since the UPS won’t start without a battery. Perhaps they should develop their own protection board. It wouldn’t even need power components; it would be enough to tell the UPS that the battery is discharged, as with branded power tools.

My subjective overall assessment.
I used this UPS for about two weeks, running it in different modes, charging batteries and powering loads. Strangely enough, during all that time I had only one complaint: the noise (humming and clicking), since it was sitting a few meters away from me.
Otherwise, it reminds me of something made by a defense-industry company and repurposed for civilian use: everything is overbuilt, and it looks like it should work reliably in freezing temperatures somewhere in a utility room.

Objectively, the most significant issue is the slow mains-to-battery switchover, which doesn’t meet the stated specifications.
It would be very helpful to hear from the developer and manufacturer’s representative. Perhaps the problem is “on my end.”

That’s all from me. I hope this was useful, and as always, I’m happy to answer any questions on the topic.