weber

Interesting update. Thanks Calvin. No, I didn't take him up on the offer to send me some cells, as we already had the charge/discharge curves from Arvio (that showed they were LTO) and I couldn't do any chemical or microscopic analysis. It's interesting that they now admit they are LTO. But I expect they are still poorly-manufactured LTOs and I expect their BMS will still be rubbish. I'm sorry I have no additional evidence.

Users of Coulomb and Weber's patched firmware 72.00a beta for the Axpert King 1 or PIP-MK, please note: The rarely-used feature "Dynamic Current Control" does not work with solar charging. It only works with AC charging. The SCC continues to use the EEPROM value of maximum charge current. I have updated the documentation accordingly. Many thanks to @Calvin, for reporting this. Thanks also to Calvin for reporting that all other features work as advertised in his 3-parallel-machines setup, as we did not have parallel Kings for testing.

Hi @Marius2001. No, those are not the latest, nor are they compatible. Sorry. You can only use patched display (MCU) firmware with patched main (DSP) firmware having the same patch letter and released together. The latest is here: https://forums.aeva.asn.au/viewtopic.php?p=97712#p97712 If we release future patched versions, you will be able to find them by going to the index post (the first post) of that thread and looking in the Firmware section. https://forums.aeva.asn.au/viewtopic.php?t=5955

I just noticed that we never announced in this thread, the September 2023 release of our fully patched firmware for the Axpert King, kindly sponsored by Power Forum member @Youda. See https://forums.aeva.asn.au/viewtopic.php?p=97712#p97712 It says "beta", but it's thoroughly tested by now and no problems have been reported. [Edit: Calvin has since reported that Dynamic Current Control does not work with solar charging in this beta patched firmware. It only works with AC charging. The SCC continues to use the EEPROM value of maximum charge current. Calvin reports that all other features work as advertised in his 3-parallel-machines setup.]

I assume I assume you're happy that the upper limits on all 5 user voltage settings are not too restrictive for 18S LFP. They look OK to me. There are 3 hidden voltage settings that can't be changed by the user. We have set them based on 15S and 16S in the LFP flavour of our patched firmware. But they will probably be OK for 17S and 18S too. You can read about them here: https://forums.aeva.asn.au/viewtopic.php?f=64&t=5955&p=76194#LFP Time, indeed, is the real issue. The electrolytic capacitors on both the battery bus and the 400 V bus will age more rapidly because they will be running closer to their rated voltages. They will spend most of their time around 60 V instead of 53 V. It's almost certain that these inverters were not designed with that in mind. These capacitors, particularly those on the battery bus, have always been a weak point of these inverters. And when their internal resistance rises enough that they can't control the voltage spikes that the MOSFETs are subjected to, the MOSFETs will also be 7 V closer to their rated voltage.

Anyone thinking they might take a risk on these so-called "super-capacitors" on the basis that LTO is a long-lived battery chemistry, should forget it. That's only the case if the LTOs are properly manufactured and are protected and balanced by a proper BMS. The BMS in this device is rubbish. The photos of the fire should have been enough to convince us of that. But some more information has recently come to light, thanks to a brave volunteer guinea-pig. I received the following email 6 weeks ago. I asked the author to please post the information himself directly, but because he hasn't yet done so, and I'm concerned that people might still be thinking it's worth the risk, I post the email below, unedited. "Arvio" is the Australian agent for these devices. The executive summary: It lasted about 2 years. The warranty was not honoured. Disassembly found bulging cells and leaking electrolyte.

We have already released patched firmware for the King which fixes the premature float bug (and nothing else). That's presumably what Lionel is referring to below. It can be found here: https://forums.aeva.asn.au/viewtopic.php?p=76178#p76178 I assume you meant to write "PIP-5048MK". Yes, Coulomb recently obtained one, and we plan to start porting more of our PIP-5048MS patches to it in a few weeks time.

See Coulomb's excellent posts via the first 3 links here: https://forums.aeva.asn.au/viewtopic.php?t=4332#commands

No. You should not let it go to 3.7 V. Use a headlight bulb to bring it down to where the others are.

Yes. If you need to set the absorb voltage lower than your usual float voltage then you should set absorb and float the same. You typically won't be allowed to set absorb voltage lower than float voltage, so you'll probably have to set the float voltage down first.

Only one cell needs to be above 3.4 V for it to be worthwhile doing balancing. It doesn't matter if the others are greater or less than 3.4 V. And you should stop (or the BMS stops you anyway) when any cell goes above 3.6 V. So you need to extend the time between the first cell going over 3.4 V and when it goes over 3.6 V. To do that, you need to reduce the charging current. Ideally, you would reduce it to the same as (or a little less than) the current that you are able to bypass around a cell by connecting a headlight bulb across it, which is typically 2 A for one filament. With the 5 kVA Axpert inverters, if you are able to charge from the utility, you can use the lowest utility charge current setting [11] of 2 A. This is ideal. If you are not able to charge from utility, you will need to set the absorb voltage [26] down to whatever will reduce the current to about 2 A by the time the first cell exceeds 3.4 V. This could be as low as n × 3.36 V where n is the number of cells. As you progress with balancing, you will need to keep raising setting [26] to keep the charge current close to 2 A. And if a cell does go over 3.6 V and so you have to stop charging, (or it stops automatically), then you leave the headlight bulb on that cell until its voltage goes below that of the lowest cell, then you start the charge cycle over again, and keep repeating this until you get all cells between 3.45 V and 3.6 V at the same time, and within 0.05 V of each other. Hopefully, after you've done that once, you'll be able to set your charge voltage and current settings back to normal, and the BMS will be able to keep them balanced from then on. I note for other readers, that these voltages only apply to LFP cells. Also, Ojeysky, I wonder if you have the correct Daly BMS — i.e. whether you have a model designed specifically for LFP cells and not other lithium-ion cells. Other lithium ion cells have a more linear voltage vs SoC curve, where LFP has some very flat plateaux. With a linear curve, the BMS can balance at any SoC, and can therefore get by with a few tens of milliamps of bypass current. This will not work with LFPs. With LFPs, such a scheme will actually unbalance the cells due to inaccuracies in voltage measurement. With the LFP cells you can only balance when they are near 100% charge (or less conveniently, when they are below 30% charge) and because the time available is so much shorter, the bypass current must be so much greater, typically half an amp or more.

This is not mere theory. Coulomb and I do this with every new battery (set of cells). The typical 55 W headlight bulb draws about 4 A at 13.8 V, so you might expect it would only draw a quarter of the current at a quarter of the voltage, so about 1 A at 3.4 V. But in fact it will draw about 2 A at 3.4 A. This is because the filament has a much lower resistance when it is only glowing a dull red compared to when it is white hot. A fan will behave in the opposite way. i.e. it will draw much less than a quarter of the current at a quarter of the voltage. I doubt that the fan will draw enough current from a single cell to be useful in balancing the cells in a reasonable time. But it can't hurt to try it. If you only do this manual balancing after you stop charging, then the cell you connect the load to, and in fact all the cells, will drop below 3.4 V almost immediately, so you will only have a very short time of balancing. You would have to repeat this many many times.

One way is to use one or more car headlight bulbs with alligator-clip leads, and a multimeter. With the battery on charge, and close to full charge, clip the bulb(s) on to the cell(s) with the highest voltage(s), provided that voltage is greater than 3.4 V (for LFP cells), to burn off some charge and let the other cells catch up. You can leave the bulb on a cell until its voltage drops below that of the lowest voltage cell. You can also work in the other direction. If you have an adjustable-voltage current-limited power supply (a lab power supply) you can set it to 3.6 V, and when the whole battery is being charged and some cells are over 3.4 V, connect it to the lowest voltage cell until it goes above the voltage of the highest voltage cell. If there happens to be two to four high cells next to each other, you can clip one bulb across the lot to burn off charge faster. You should put each bulb in a porcelain cup or mug to stop it blinding you and melting things. This is a @Coulomb innovation. 😀 Likewise if you have a number of low cells next to each other, you can adjust the lab power supply to (n × 3.4 V) + 0.2 V. But this is more dangerous. You must monitor the individual cells often to ensure none goes over 3.6 V. 3.4 V is an absolute minimum for balancing. The higher you go, the more accurate the balancing will be, but you shouldn't let any cell go over 3.6 V. I note that connecting LFP cells in parallel will not balance them unless you charge them to more than 3.4 V while they are connected in parallel.

If this inverter accepts the same serial commands as its big brother, then I think it will reset its charging algorithm if you send it a command to change the float voltage or absorb voltage. You only need to change it by 0.1 V and then you could change it back. You could automate this, using Node-RED on a Raspberry Pi. The commands are PBFT and PCVV. http://forums.aeva.asn.au/uploads/293/HS_MS_MSX_RS232_Protocol_20140822_after_current_upgrade.pdf

I thought I was a bit more nuanced than that. 🤣 I did give at least one suggestion for making it work. https://powerforum.co.za/topic/4614-axpert-settings-for-lifepo4/page/2/?tab=comments#comment-70618