AEG SolarCube Pro (AS-BBL08) after the firmware update: why it drained, and how to really switch it off

On the new firmware, the AEG SolarCube Pro from our review stood off the grid for 23 hours, with the button off. The state of charge went from 49% to 49%. own measurementAEG StoragePRO app and the manufacturer's cloud, second location, without a Modbus bridge · first the plug out in Backup mode, then the button until the ring was off · 6 October 2026

In September, the same unit drained while I had switched it off in the app. HYXi, the manufacturer, confirmed two faults in the old firmware. Neither of them was in the hardware. manufacturerHYXi, in writing in the group chat with the manufacturer · 1 November 2026

This article is the written follow-up to our review of 1 October and to the update video of 3 October (in Dutch). It is about the same unit: no new unit was sent. There is also no new bench round, because the unit is at a second location and not on our techbench. Capacity and round-trip efficiency from the review were therefore not measured again. I am putting the review video back online, as it was, with a reference to this follow-up.

Key takeaways

• On the new firmware, it really switches off. With the button off and off the grid, it was still at 49% after about 23 hours. In the review I wrote that there is no real off. On the old firmware that was true. On V01.03.01.55, that off exists. own measurementAEG StoragePRO app and the manufacturer's cloud, second location, without a Modbus bridge · first the plug out in Backup mode, then the button until the ring was off · 6 October 2026
• The drain in September was down to the firmware. According to HYXi, the auxiliary power supply could stay active after switching off. Then the unit did not switch off completely, and the battery drained to undervoltage. manufacturerHYXi, in writing in the group chat with the manufacturer · 1 November 2026
• The update that broke off at 21% was also down to the firmware. A fault in the old firmware made the download fail from time to time. According to HYXi, it is a software or firmware fault, not a hardware fault in the battery management system. manufacturerHYXi, in writing in the group chat with the manufacturer · 1 November 2026
• Switch it off the wrong way and you lose power: only pulling the plug, in Backup mode, is not off. The unit then goes off-grid and draws about 37 W from the battery, also without a load. A full battery is empty that way in about 6.6 days. own measurementAEG StoragePRO app and the manufacturer's cloud, second location · Backup mode, plug out, no load · 5 October 2026 derivedfrom 5,890 Wh divided by 37 W gives 159 hours
• The test report I compared our efficiency with belongs to another product, according to HYXi. HYXi's own test of this device comes to 81.65% round-trip efficiency, at 3,000 W. We measured 81.29% in September, at about 735 W. manufacturerHYXi, own test of this device, shared in the group chat with the manufacturer · 20 October 2026 own measurementour techbench, kWh counters of the metering plug · closed cycle, about 710 to 760 W, old firmware · 12 September 2026
• The capacity was wrong in the update. Both product sheets give 6.0 kWh nominal and 90% depth of discharge, so 5.43 kWh usable. We measured 5.89 kWh. manufacturerfootnote 4 of the Dutch product sheet (via APEX) and of sheet V2.0, read again · 3 October 2026 own measurementour techbench · old firmware, the bench measurement of the review · 12 September 2026
• The safety notice about the regular SolarCube does not apply to the Pro. It is about the AS-BBL09, built by Sunpura. The Pro is the AS-BBL08, built by HYXi. third partysafety notice of the distributor on aegthuisbatterij.nl; the type number AS-BBL09 from P1meter.nl · 6 November 2026

How it was measured

• Unit: 34302261602711, the same as in the review
• Firmware: V01.03.01.55 for ARM/System and Battery, V01.03.01.30 for Control and V01.02.03.0C for Wi-Fi, from the app; in September V01.03.01.1E
• Test date: 2026-10-05 to 2026-10-06
• Test duration: about 23 hours for the sleep test, with the measurement in Backup mode just before it
• Grid setting: Backup mode; for the sleep test off the grid, with the button off
• Instrument: the AEG StoragePRO app and the manufacturer's cloud, at a second location; no Modbus bridge, and not on our techbench
• Measurement log: none; the October measurements come from the app and the cloud, not from our techbench

One unit is one unit. The figures from the review were measured on 11 and 12 September 2026, on the old firmware, with an earlier version of our test method. They are given here with their measurement date, and not on one line with the October measurements.

What happened in September

The firmware update did not come through. From 11 to 15 September, the app offered the package MS3000AC. I tried it about 20 times, and every attempt broke off at about 21%. The unit stayed on V01.03.01.1E, the firmware it arrived with. The package on offer was V01.03.01.75. On 12 and 14 September I reported it to AEG SolarCube support. own measurementAEG StoragePRO app, the update action · n/a, a firmware readout · 15 September 2026

After that, I switched it off in the app, with "Power Off", and took it off the grid. The lights and the wifi indicator went out. The RGB ring stayed on; I could not switch it off with any control. From about 16 September, the Modbus bridge was on another device. So during the whole drain, nothing was connected to the unit. own measurementthe unit itself and the AEG StoragePRO app · off via the app, off the grid · 16 September 2026

The last data point in the cloud is from 15 September at 03:36: state of charge 47%. On 24 September at 22:49 it reported "System hibernation shutdown". On 25 September at 19:39 it was back on the grid. Then it showed error code 170305110, undervoltage on the head unit, and it was locked. Pack 1 was at 0%, and pack 2 did not report. On the grid, it came out of that by itself after 4 days and 5 hours. own measurementthe manufacturer's cloud, the AEG StoragePRO app and Modbus readout · off via the app, off the grid; then locked · 30 September 2026

How fast it drained, I do not give as a figure. The rate follows only from two end points. What is certain: off the grid, in standby, it kept drawing from the battery. What the error code means and how the recovery went is on the page about error code 170305110.

Why it drained, and why the update did not come through

HYXi adjusted its explanation in steps. Below is the latest position, of 1 November 2026. The earlier positions are under the contradictions.

The drain. After switching off, the auxiliary power supply could sometimes stay active. Then the unit did not switch off completely, and the battery drained to undervoltage. HYXi wrote on 13 October: "The earlier firmware could, under certain conditions, cause the unit not to shut down completely." According to HYXi, this is fixed in a release of late July, and in the firmware that is now on our unit. manufacturerHYXi, in writing, 13 October and 1 November 2026 · 1 November 2026

The update. The older firmware had a fault that made the download fail from time to time. That is where the break-off at 21% comes from. According to HYXi, it is a software or firmware fault, not a hardware fault in the battery management system. The fix is in the release of 31 July. HYXi calls it "not a widespread issue" and "a small probability". How often it happens is nowhere given as a figure in the conversations with HYXi. manufacturerHYXi, in writing, 6 October and 1 November 2026 · 1 November 2026

Why the fix from July was not yet on the unit in September. The fault was in the firmware that had to fetch the update, and that old firmware was on the unit. So the update failed, the old firmware stayed, and the fault in switching off applied as well. That link is my reading. HYXi confirmed each fault separately, and does not say it in these words. derivedfrom the two causes HYXi confirmed separately, on 13 October and 1 November 2026

Modbus played no part. During the drain, the bridge was on another device. The successful sleep test in October also ran without a bridge. own measurementthe setup, checked · n/a, what was connected · 8 October 2026

The 23-hour test

On 5 October, the same unit went onto a new installation at a second location. There, the update succeeded in one go. After that, the app reported ARM/System V01.03.01.55, Battery V01.03.01.55, Control V01.03.01.30 and Wi-Fi V01.02.03.0C. Those are the parts of the package MS3000AC_V01.03.01.75. own measurementAEG StoragePRO app · n/a, a firmware readout · 5 October 2026

Just before the test, the unit was healthy: state of charge 50%, both packs connected, no alarms. Pack 1 had cells from 3.296 to 3.303 V, at 25.7 °C. Pack 2 had cells from 3.314 to 3.315 V, at 23.3 °C. own measurementAEG StoragePRO app and the manufacturer's cloud, second location · on the grid, just before the test · 5 October 2026 The RGB ring is a state-of-charge indicator: at 50% it is half lit. own measurementthe ring of the unit, second location · n/a, an indicator · 6 October 2026

The order in our test was: first the plug out, then the button. At 23:27 the plug came out, in Backup mode. The unit did not switch off. It went off-grid and kept the inverter running, without a load, and drew about 37 W from the battery: pack 1 about 11 W, pack 2 about 26 W. At 23:31:49 the on/off button was held down until the ring was off. own measurementAEG StoragePRO app and the manufacturer's cloud, second location, without a Modbus bridge · Backup mode, plug out, no load; then button off · 5 October 2026

At the start, the state of charge was 49%, with cells at about 3.29 to 3.31 V. After about 23 hours, it was still at 49%. The cells were 8 to 9 mV lower, and about 6 °C colder. The ring stayed off the whole time. own measurementAEG StoragePRO app and the manufacturer's cloud, second location, without a Modbus bridge · button off, off the grid · 6 October 2026 According to HYXi, that small voltage drop matches the temperature drop, and does not point to a drain. manufacturerHYXi, in writing in the chat with the manufacturer · 8 October 2026

For comparison: at 9 W, 23 hours would have cost 207 Wh, about 3.5% of 5.89 kWh. At 37 W it would have been 851 Wh, about 14%. Measured: about 0%. derivedfrom 9 W and 37 W times 23 hours, divided by the 5.89 kWh from the review

Off, in watts: the four states

Below are the states according to HYXi, next to what we measured. Our measurements of 30 September were taken on the old firmware.

So that roughly 9 W is not new: we measured it in September already. If a state lasts a whole year, 9 W is 78.8 kWh per year, and 5 W is 43.8 kWh. For comparison: the self-consumption in operation was 41 to 47 W in the review, or 359 to 412 kWh per year. derivedfrom the power in W times 8.76, for a year of 8,760 hours

Only pulling the plug is not off. In Backup mode, the unit goes off-grid when the plug comes out. It keeps the inverter running and draws about 37 W from the battery, also without a load. own measurementAEG StoragePRO app and the manufacturer's cloud, second location · Backup mode, plug out, no load · 5 October 2026 A full battery of 5.89 kWh is empty that way in about 6.6 days. derivedfrom 5,890 Wh divided by 37 W gives 159 hours

That is how it is meant to work, HYXi writes. The inverter stays idle, so that it can deliver at once when an emergency load comes. If the state of charge and the voltage drop further, the system goes into deep sleep by itself, according to HYXi, with a consumption at microampere level. Back on the grid, it charges the battery back to a safe level. Whether it goes into deep sleep by itself, we have not tested. manufacturerHYXi, in writing in the chat with the manufacturer · 6 October 2026

How to really switch it off

HYXi prescribes this order:

1. Hold the on/off button down until the RGB ring is off.
2. Then pull the plug.

We did it the other way round: first the plug, then the button. That way it also went into deep sleep. We have not tested HYXi's order ourselves. In our test, the button made the difference: without the button it kept running at about 37 W, with the button it slept. own measurementAEG StoragePRO app and the manufacturer's cloud, second location, without a Modbus bridge · Backup mode, plug out; then button off · 6 October 2026

Off via the app is not off. The ring stays lit, and according to HYXi the unit draws about 9 W. On 3 October, HYXi admitted that the app does not make the difference between "Power Off" and really off clear enough. HYXi promised a notice that you hold the physical button down to switch off completely or for storage. manufacturerHYXi, in writing in the chat with the manufacturer · 3 October 2026

Does your update hang at 21%? Then, according to HYXi, you can contact AEG and HYXi support. They have a guided upgrade procedure. manufacturerHYXi, in writing, 6 and 13 October and 1 November 2026 · 1 November 2026

An owner of a HYXi Halo reported under the update video that his update also got stuck at 21%. The update went through after he removed the P1 link from the app. On 31 October, two updates later, updating worked without removing the P1 meter. I read that as progress in the later firmware, but I have not checked it. third partycomments by a viewer under the update video of 3 October 2026, not checked · 31 October 2026

The three corrections to the review and the update

Three things I said in the review of 1 October or in the update of 3 October are not correct, or no longer correct.

1. There is a real off. In the review I wrote: "There is no real off." With the button off, the metering plug measured 9.3 W, and Modbus kept answering. Standby via the app was 9.5 W. Off the grid, that consumption came from the battery. own measurementmetering plug and Modbus readout · old firmware, button off and standby via the app · 30 September 2026 That held on the old firmware. On V01.03.01.55 there is a real off: hold the button down until the ring is off, and take it off the grid. On the old firmware, that state was out of our reach, because of the fault in switching off. derivedfrom the sleep test of 5 and 6 October 2026 and the fault in the old firmware that HYXi confirmed on 13 October

2. The test report belonged to another product. In the review and the update, I put our round-trip efficiency next to the test report RTE-2025-001, with a band of 79.96% to 87.06%. That report came via APEX. I said that our measurement fell inside that band. HYXi wrote on 20 October that the report "relates to a different product, not our Halo/AEG SolarCube", and that the test procedure does not match HYXi's: "It should therefore not be used as a reference for this unit." That comparison lapses. manufacturerHYXi, in writing in the group chat with the manufacturer · 20 October 2026

The new comparison is the one with HYXi's own test of this device. HYXi did two runs: 80.72% and 82.57%, on average 81.65% round-trip efficiency. HYXi measured on the AC side, with 30 minutes of rest between the steps. The setup was one head unit HYX-MS3000AC with one battery pack HYX-MS3000B, at 25 °C, on a grid simulator at 230 V AC and 50 Hz. The power limit in the app was set to 3,000 W. HYXi discharged at 3,000 W in Custom mode and charged at 3,000 W in Backup mode. The discharge ran until the protection stopped it at about 10% state of charge, not to 0%. manufacturerHYXi, own test of this device, shared in the group chat; the settings on 21 October 2026 · 20 October 2026

On 12 September we measured 81.29% round-trip efficiency over one closed cycle, at about 710 to 760 W, from 28% to 46% state of charge, with the kWh counters of the metering plug. That was on the old firmware. own measurementour techbench, kWh counters of the metering plug · closed cycle, about 710 to 760 W, old firmware · 12 September 2026

The difference is 0.36 percentage points. The conditions differ: 3,000 W against about 735 W, and a larger charge window at HYXi. Whether that difference is chance, I cannot say. I do see that our real-world measurement and HYXi's test are close to each other. derivedfrom HYXi's 81.65% minus our 81.29%

3. The capacity was wrong in the update. In the update I said that the Dutch sheet promises 6.0 kWh usable, and that the English one comes to 5.43 kWh. The review article says the same. That is not correct. Both sheets give 6.0 kWh nominal capacity and 90% depth of discharge, so 5.43 kWh usable. That is in footnote 4 of both. manufacturerfootnote 4 of the Dutch product sheet (via APEX) and of sheet V2.0, read again · 3 October 2026 We measured 5.89 kWh usable, more than what the sheets call usable. own measurementour techbench · old firmware, the bench measurement of the review · 12 September 2026

The error came in while copying. Our review card put the nominal capacity in the field for usable, and that was later read as a promise of the Dutch sheet. On top of that, the text layer of the Dutch sheet is missing the number 90 in the row for the depth of discharge.

What else is in the new firmware, according to HYXi

On 13 October, HYXi named what the release of 15 October was to improve. HYXi called it an "improvement scope". These are not release notes.

• A more stable update, and better handling of an update that fails
• Optimised switching off, sleep and standby consumption
• A "no charging" option in Custom mode
• Charging power adjustable from 200 to 3,000 W, in steps of 100 W manufacturerHYXi, in writing, the improvement points for the release of 15 October 2026 · 13 October 2026
• A clearer explanation of the default behaviour at 200 W manufacturerHYXi, in writing, the improvement points for the release of 15 October 2026 · 13 October 2026
• Better power control: the power on the AC side is closer to the setting
• Better Modbus/RS485 documentation, and a time-out protection

For the power control, HYXi named an internal test: about 797 W at a discharge setting of 800 W. In addition, HYXi wrote on 16 October that forced charging now follows the set AC input limit, adjustable from 0 to 3,000 W. If the limit is set to 1,000 W, it force-charges at 1,000 W, not at a separate fixed value. manufacturerHYXi, in writing in the group chat with the manufacturer, 16 and 19 October 2026 · 19 October 2026

The 23-hour test ran on 5 and 6 October, so before that release.

What we reported, and what HYXi wrote back. On 3 October, HYXi forwarded our files to its teams for the battery management system and the firmware. On 5 October at 23:41, I reported to HYXi the roughly 37 W from the battery, off the grid in Backup mode. At 00:41, HYXi wrote that R&D was preparing a new firmware release for two points: "the off-grid idle consumption and the deep-discharge/recovery behaviour". Both points came from us. manufacturerHYXi, in writing in the chat with the manufacturer · 6 October 2026 An answer within the hour leaves open that the release was already being prepared. derivedfrom the time of our report and of HYXi's answer

The list of 13 October also has "Optimized shutdown, sleep and standby-power behaviour". I like to think that our findings helped that along. The conversations with HYXi say that nowhere. And according to HYXi, the two faults themselves were already fixed in releases of late July, so before the review. Our findings were not the reason for those fixes.

On 4 October, I proposed three things to HYXi:

• going to sleep automatically after an adjustable time without charging, discharging or an emergency load
• an adjustable minimum operating threshold, so that it does not work in the low range where the efficiency is low
• an eco mode

HYXi passed the proposals on to R&D and product management. Up to and including 1 November 2026, no formal answer came. manufacturerHYXi, in writing, 13 October 2026; in the group chat no later reaction up to and including 1 November 2026 · 1 November 2026

What was not measured again after the update

On the updated unit, the packs, the sleep test and Backup mode were measured. That is above. There is no new bench round: the unit is at a second location.

On 28 October, I checked the unit remotely, via the app and the cloud. The results stayed roughly the same, and nothing got worse. I did not record figures from that check.

What is therefore derived, and not measured again:

• Usable capacity: 5.89 kWh, from the review. The cells and the packs are the same. The floor of about 10% state of charge is also in HYXi's test. Nothing on the list of improvement points touches the capacity. derivedfrom the bench measurement of 12 September 2026, HYXi's test settings and the list of 13 October 2026
• Round-trip efficiency: 81.29% at about 735 W, from the review. HYXi's own test of this device comes to 81.65% at 3,000 W. derivedfrom the bench measurement of 12 September 2026 and HYXi's test
• Self-consumption in operation: 41 to 47 W, from the review. HYXi calls the standby consumption "optimised", without a figure. A viewer reports "around 50 watts" after an update, not checked. derivedfrom the bench measurement of 12 September 2026 and the list of 13 October 2026 third partycomment by a viewer under the update video of 3 October 2026, not checked · 16 October 2026

Which AEG SolarCube do you have?

On 4 and 5 November 2026, there was news about the regular AEG SolarCube. It does not affect the Pro in this article.

Two units are called AEG SolarCube. The regular one is type AS-BBL09 and comes from Sunpura. The Pro is type AS-BBL08 and comes from HYXi. Our unit reports itself as AS-BBL08-3K. third partyP1meter.nl, article of 4 November 2026, for the type number AS-BBL09; the notice of the distributor gives no type number · 4 November 2026 own measurementModbus readout of the model string · n/a, a model string · 12 September 2026

The distributor of AEG home batteries in the Benelux published a safety notice (in Dutch). It applies only to the regular SolarCube: "Dit bericht geldt alleen voor de AEG Solarcube. Heeft u een AEG Solarcube Pro, dan hoeft u niets te doen." (This notice applies only to the AEG Solarcube. If you have an AEG Solarcube Pro, you do not have to do anything.)

What is wrong, according to the distributor. We have not checked this.

• The connector between the BMS (the unit on top of the battery) and the battery can overheat. The plastic housing can then partly melt, with light smoke and the smell of melted plastic.
• The cause is a slight dent in the bottom plate of the BMS. The connector then does not seat fully, the contact resistance rises, and the connector gets hot locally.
• On fire: "In theorie is dit mogelijk, maar een gesmolten stekker heeft nog nergens tot brand geleid." (In theory this is possible, but a melted plug has not led to a fire anywhere yet.)
• More than 1,100 repairs were carried out. After that, 0.48% of the BMS units, 5 of them, still had a melting connector. That is why the repair stops, and all BMS units are being replaced, 4,500 of them. The new BMS has a temperature sensor in the connector, and the factory has changed the production of the housing. third partysafety notice and official statement of the distributor, aegthuisbatterij.nl; the statement was created on 30 October 2026 · 6 November 2026

What an owner of a regular SolarCube must do, according to the distributor: first switch off the battery with the button on the side, then the BMS on top, and finally switch off the circuit in the meter cupboard or pull the plug. Do that even if everything works normally, until the BMS has been replaced. You do not have to request anything yourself: the distributor will contact you. The aim is to start the exchange within 2 months. The exchange is free, with one year of extra warranty on the new BMS. Compensation will be looked into after the exchange. third partysafety notice of the distributor, aegthuisbatterij.nl · 6 November 2026

The notice is about hardware of the AS-BBL09. Our findings are about the firmware of the Pro. In the review I already wrote that the brand name is not the builder. Here you see it: one brand name, two builders, and the notice affects one of them. derivedfrom the type numbers and the builders above, and the review of 1 October 2026

Where the sources contradict each other

HYXi adjusted its explanation in steps. Where two positions differ, the latest one applies. Below, per point, what HYXi wrote and when.

• The sleep logic. On 6 October, HYXi wrote that the sleep logic in V01.03.01.55 was "unchanged". On 13 October, HYXi confirmed the fault in the old firmware that kept the unit from switching off completely.
• The version with the fix for the update. On 6 October, HYXi named V01.03.01.55. On 1 November, HYXi wrote: the release of 31 July.
• The link between the two faults. On 8 October, HYXi wrote that this link was not yet proven. On 13 October, HYXi confirmed the fault in switching off for the old firmware. The link remains my reading.
• Hardware or software. In my technical summary of 29 October, I myself wrote "older hardware in the BMS". HYXi corrected that on 1 November: software, not hardware.

Frequently asked questions

Related articles

• The review of the AEG SolarCube Pro for the full test of September, with the discharge curve and the closed cycle
• Error code 170305110 on the AEG SolarCube for the undervoltage lock and how it came out of it
• The Modbus registers of the AEG SolarCube Pro for the register map

Transparency

This article is the written follow-up to the review. It is about the same unit, sent free of charge by APEX Power Supplies. Nothing was paid for it, and nobody saw the script or the retest beforehand.

HYXi did see my technical summary beforehand, on 29 October 2026, and corrected two points in it on 1 November. HYXi did not see the script or the video. There are no agreements about what I say about it. This article has no affiliate links and no discount code.

Every figure in this piece carries its origin in the line. eigen meting (own measurement) carries the setting, the instrument and the day; fabrikant (manufacturer) the document or the message and the day; derde (third party) the source; afgeleid (derived) what it follows from. A message from HYXi is a statement by the manufacturer: we have not checked it, unless a measurement stands next to it. A report by a viewer is derde, and not checked.

Sources