Coulomb

Wow. 1.9 / 1.2 is almost 1.6:1. That's a very wide voltage range. Not many 24 V loads would be comfortable with 37 or 38 V feeding them either. Most searches seem to talk about 1.45 to 1.5 V maximum, but perhaps flooded NiCd is different. It will take an unusual charger to accommodate these cells. Edit : Battery University mentions 1.55 V per cell, which would be a more achievable 31 V for 20 cells in series. Are you sure of this 1.9 V per cell figure?

Yes. sorry, it was a crazy day yesterday, and busy today. That is indeed the "release" version of the patched 72.70 firmware, that we're calling 72.70c. There are some "Easter Eggs" to be discovered in the new firmware too.

No external device; the Axperts have relays inside that do the switching, and the inverter synchronises to the grid, so it's a pretty clean changeover. The Axpert will switch according to its own firmware based on battery voltage, but the ICC/Victron combination overrides this and makes it happen at more appropriate times, based on real SOC, as Plonkster mentioned.

Not really. This setting is compressed to about 4 bits when sent in a CAN bus message to paralleled or 3-phase machines to synchronise settings. So then you'd have to reduce the range, and users would disagree about what values they would want to keep. I sometimes wish they used half volt steps myself.

No bug reports so far. There is one minor change we want to make to the LiFe version. So release version of the patch should be out soon.

Yes.

My only suggestion is to check the relays. I suspect that the 3 kVA models are a little different, but some sort of relay fault could still cause this. For example, if there is a changeover contact set (normally open and normally closed), or pair of changeover contacts, then look for shorted contacts (normally open to corresponding normally closed). If you need to replace a relay, try extra hard to get a good quality replacement.

My colleague Weber and I have been working on a new version of the patched firmware. It's still based on 72.70. since that's the latest firmware that we have files for. For various reasons, he's been doing the majority of the work on this patch. It's been an amazing job: some 5 weeks of work, over 60 individual patches, and there is a complete new charge control mode called dynamic current limit control. The latter is complex enough that it has its own manual. For users with lead acid batteries, the dynamic current limit probably won't be of much interest, but for users of paralleled systems there is a bug fix that should improve charging. There is a quicker transition from line (utility, bypass) mode to battery mode, but I'm not happy with that part of things yet, and I am still investigating. But transitions are under a minute now, and it seems to me that they were much slower than that in both standard firmware and patched firmware before now. Finally, it's no longer necessary to update the SCC firmware, which was always a bit more of a hassle for users than the DSP firmware. It's beta software, so please try it if you're keen. But if you're cautious or not very confident about changing firmware, hold back for the "production" release, which will be called 72.70c, and will hopefully only be a week or so away. Details are on this AEVA page: http://forums.aeva.asn.au/viewtopic.php?p=65938 . [ Edit: It's now released. See the release post instead. ] [ Edit: removed the word "Beta" from the start of the topic title. ]

Two in series is fine if they are 72 cell, but these are hard to find these days except in the very large and heavy 300 W panels. For 60 and the now very rare 54 cell panels, 3S is best. Assuming Axpert MKS [ edit: and 5 kVA] as per the topic title. It's easy to figure out how many cells a panel has; count the rows and columns. 10 rows x 6 columns = 60 cells. You can usually tell from a photo, when planning a layout.

The other problem is a severe selection bias. Those who had trouble post about it, looking for solutions. Those that have no problems forget about it, and don't tend to post about it. But yes, I'd love to know the failure rate too, as long as it was reasonably accurate.

No pressure, huh? Unfortunately, repairing Axperts is more of an art than a science. My colleage "Weber" and I have fixed one or two, but we've also repaired some and had them fail nearly straight away again. When the MOSFETs blow, it's very common that some of the gate driver components also fail. Same with the IGBTs. Sometimes a failed component is obvious: it's blackened, and it might have a leg blown off. Sometimes it's more subtle: the epoxy is cracked, and when you look at it carefully you notice it. But sometimes it's really subtle: you have to look very carefully, and then you almost convince yourself that you can see a hairline crack. So really you have to test the three sets of usual suspects: the 16 MOSFETs, the 4 IGBTs at the high voltage end of the DC-DC (opposite end to the inverter IGBTs), and the 4 inverter IGBTs themselves. If they're not short circuited and you can measure about 0.43 V drain to source or collector to emitter reverse biased, and no leakage gate to source or emitter, then they're usually fine. Of course, they all have gate to source/emitter resistors, so to be sure, you have to remove either the MOSFET or the gate resistors (and the MOSFETs have four in parallel). I've written up a little on doing these tests here: http://forums.aeva.asn.au/viewtopic.php?p=65588#p65588 I update frequently, so use the index on page one and search for the keyword repair. (Yes, the index is getting that large these days... ). The good news is, you may not need to replace all the MOSFETs for a second-round repair (but likely half of them). Edit : Oh, and definitely replace the capacitors as Chris mentioned, with long life low ESR types, preferably the ones recommended in the post.

Updating to 80 A MPPT could well be different to updating to the 5 kW inverter. I'd be very surprised to see 5 kW coming from an inverter with 72.40 firmware. They can of course all do 5 kVA. Edit : I'm not saying anyone is wrong, just predicting my level of surprise.

So you are saying that new inverter-charger hardware with old 72.40 firmware results in 5 kW output?

A few posts back I mentioned that I thought the battery-side MOSFETs would be increased from 4P to 5P. My evidence for this is the higher losses (more switching losses from the extra devices despite the lower on-resistance losses), and the fact that the reverse polarity connection is now by the fuse. My colleague Weber pointed out to me some time ago, and I completely forgot, that the extra power also has to go through the high frequency transformer. He found evidence that the PIP-5048 models weigh more than the PIP-4048 models. There was one document where the weight was the same, but the rated power was also given as 4 kW, so this was presumably a copy and paste error where one table was not updated from 4 kW to 5 kW figures. In fact, he thought that the MOSFETs might be able to handle the extra current, so the MOSFETs might still be 4P. I would be wary of any inverter that has been transformed from 4 kW to 5 kW merely by firmware change. The hardware might not be able to sustain the extra 25% of power. In the case of the high frequency transformer, I would expect the transformer to saturate with the extra power, which I think would cause instantaneous MOSFET failure (not merely overheating and an ordered shutdown due to over-temperature). But I'm happy to be proved wrong with evidence (photos of the MOSFETs and/or measurements of the high frequency transformer).

Exactly! My suspicion is that they changed the DC-DC boost stage MOSFETs from 4 paralleled devices to 5. [ Edit: they may well have had to change the high frequency transformer as well; see my later post on this. ] To fit these in, they would have extended the heatsink, and to make room for that, I believe that they removed the reverse polarity protection MOSFETs, which were four paralleled devices of the same size. The latter may have been replaced with a honking great diode antiparalleled with the battery after the fuse, so reverse connection will now result in a huge splat and a blown fuse. The catalogue mentioned fuse protection against reverse polarity, from memory. So the inverter IGBTs were always capable of 5 kVA (not the 5000 kVA that at least two posters mentioned ), but now if the load is unity power factor, it can drag 5000 W from the DC-DC. There is no suggestion of correcting the power factor of the load. I hope it's not merely a firmware update as @Chris Hobson has suggested (based on others' statements I'm sure), because the DC-DC section seems to be the weakest link in the power chain already, and working it 25% harder without hardware modification seems reckless. Maybe it's technically possible to apply 5 kW firmware to 4 kW hardware, at least for a short while, but I hope that's not officially allowed. But maybe I'm wrong and the DC-DC stage is perfectly capable of 5 kW operation. That would mean over 100 A continuous distributed over 4 TO-220 packages, or over 25 A each.

I had posted a replied to this, but the post seems to be have been lost in the recent rollback. Briefly, the Duo and Trio MPPT models seem to be variants of the Axpert 5 kVA model, with an additional one or MPPT Solar Charge Controllers. The Duo and Trio models seem to need firmware version 73.XX, but 73.XX can run single MPPT Axperts as well. It should be possible to combine them in the same system, provided that they are running the same 73.XX firmware. You may have to ask for the firmware update file from your supplier. Unfortunately, the charge bug is not fixed as of 73.00, the latest I know about, and there is no 73.XX patched firmware at present.

Actually, the inverter checks its output with the same sense resistors that it uses to figure out how much PWM to give the IGBTs. So that doesn't sound likely after all. It has to be some sort of control issue, which is handled by the firmware. I suppose if the bypass relay is stuck on, it would read utility voltage, which could be over 250 VAC some of the time. So as the service manual suggests, try running with no utility and no load. If no problem for a while, turn on loads; if no problem for another while, turn on the AC input. Perhaps a particular load might cause the inverter to become unstable; a few such loads have been identified. So if the problem occurs only with the loads on, try turning off all loads and gradually turn them back on to see which one might trigger the issue.

In a profile post by @jussclay: I've not seen an error 06. It happens when in battery mode the inverter output voltage is 20 V over the nominal voltage for 10 "ticks". A tick is 1/50th of a second, or 20 ms, so it's allowed to go over for 180 ms, but not 200 ms. The nominal voltage is usually 230 VAC, but can be set to 220 V or 240 V with the undocumented POPV command. This is an actual fault code, not merely a warning, meaning that the inverter is in Fault mode (as opposed to battery or line modes, there are a few others). In fault mode, the inverter doesn't do much, and in my experience usually shuts down completely (even the LCD light goes off and the processor cuts its own power) after a minute or so. So the only way to "reset it" is to start the inverter again. If the fault occurs again as soon as it goes to battery mode, then it has to be repaired. As to what might cause this, I can only think that the resistor divider chain has failed and it things the voltage is > 250 V when it's really 230 V. If there was a problem with the inverter, like faulty IGBTs, then you'd get no voltage at all. The voltage would be set by changing the PWM, so if it can make any sort of sine wave, it would make one that is 230 V RMS, it seems to me. You might be able to watch the output voltage with a multimeter, but if the output only lasts 200 ms, it might be a challenge. Perhaps use your multimeter's "peak" facility to see what the maximum voltage was. The good news is that if I'm right and it's just the resistor divider, such a repair should be relatively easy. The voltage divider resistors are usually easy to spot - often four or more of them in series, usually in a straight line with few to no other components nearby. The AC output tracks are easy to spot, since they are pretty thick and striped with solder for good conductivity. The large toroidal inductor is part of the AC output circuit, to give you somewhere to start. You still have to open the case to get the main board out. The bad news is that if you can't repair it yourself or have a friend do it, it probably doesn't matter how easy the repair is, your supplier will probably replace the main board and throw the old one away, even though the cost of the replacement parts is probably less than one Rand. With labour and workshop fees, that will likely come to nearly the cost of a new unit. Of course, I could easily be wrong about what could cause this fault, so it might cost a technician hours to try and locate the fault. So the repair logic makes sense, I've attached what one of the service manuals says about fault code 06.

Be aware that the patched firmware is for the Axpert 5 kVA (4 kW) models only. @jaber, perhaps you meant 5 kVA or 4 kW in your original post. If you meant the 3.2 kW (4 kVA) model, then the patched firmware *will not work*. It is full of assumptions about the hardware being capable of 5 kVA operation, so even if the firmware loader accepted the patched firmware (hopefully it would not), it would not work, and then you would not be able to revert to factory firmware. Firmware files seem to be harder and harder to get these days, but if you need them, your supplier should be able to get you a suitable firmware file. I occasionally collect firmware for other models, but I don't have anything for the 3.2 kW model. The Czech site sometimes has firmware, but nothing since 2015 and I don't see anything for the 3.2 kW model. Sorry.

There are some reasonable service manuals here. Not many schematics, but we've traced the more important parts on the AEVA site, as Chris noted.

Unfortunately, it usually means shorted IGBTs, either in the 230V inverter, or the high-side of the DC-DC converter. It"s a Bus Soft Start error. There is a small power supply that charges the DC bus capacitors to about 8x the battery voltage (some 400 V), which is switched on very soon after powering on. If there are no paralleled machines, it has about 15 seconds to reach that voltage, or a bus soft start is asserted. A friend and I had one recently (after a repair), and the first time it took about a minute before it came up. Probably the IGBTs were OK at that point. Next startup, the error came up quickly, possibly in just over 15 seconds. There was a dead short across the DC bus. It's possible to check this with a multimeter without removing the main board, but it's tricky on some models. Measure on a voltage range first in case there isn't a short. The 2x 470 uF capacitors can hold a wallop, so care is advised. If no voltage, it's safe to measure on ohms. A working inverter should measure in the megohms, slowly changing.

When you say the SOC shows 80%, is that according to the Axpert display (or equivalently from monitoring software), or from a BMV or similar external device? As others have mentioned, the Axpert SOC meter is voltage based, with only crude load compensation, so it's not very accurate. Plus or minus 20% for a lead acid battery (plus or minus 80% for LiFePO4). What does the specfic gravity suggest? Also, what maximum battery voltage is reached, and does charge termination correlate with cloud cover?

That's to be expected. When the battery voltage drops below the float voltage setting due to a load or any other reason, the charge controller will increase its output enough to maintain the float voltage, or if it can't maintain that voltage, it will supply all it can. That is weird. I don't have a dual MPPT firmware to read, so I can't be sure. I would have expected that the DSP in the inverter, which calls the shots for a single MPPT machine, would also call the shots for a dual or triple MPPT machine. I'm pretty sure it sets a current set point, not a voltage set point (in the case of one MPPT). So I'm surprised by that behaviour, but it probably means that the multiple MPPT chargers work differently. Sorry, it didn't dawn on me till now that you have a dual MPPT pair of machines, so you can't use the [ edit: existing 72.XXy] patched firmware to fix the charging bug, if it exists in your machines. [ Edit : sadly it does.] A colleague had a single MPPT machine that came with main firmware version 73.00, and it still exhibited the same charging bug. We were unable to extract the 73.00 firmware to examine or patch. However, we were able to downgrade it to 72.70b patched firmware without apparent issue. That suggests to me that 73.00 will still have the same bug when running on a multiple MPPT machine, [ Edit: I've since learned that 73.00 is a single-MPPT only firmware] but maybe the bug only affects one of the two or three MPPTs. Maybe one or two charge controllers are effectively independent, just packaged in the same box for convenience. That does sound crazy, I'll admit. So sorry, I don't have a solution for you; if the bug exists, and it seems likely that it does , you'll have to wait for Voltronic to fix it, and ask them or your supplier for an update. [ Edit: we have a patch for dual/tri MPPT models now; see Weber's message several posts below.]

The Axpert is perfectly capable of pushing the current into the battery when there is enough solar power available (e.g. when the Outback charge controller would do so), it's just a logic error that makes it think that charging is over and it's time to float the batteries, not charge them. I run 2S of 72-cell panels with no issues that I'm aware of. Sometimes the MPPT voltage will be down to around 63 V, but it still seems to charge just fine. In summer I get nominal or even over nominal power for short periods of time. My lithium iron phosphate (LiFePO₄) battery charges at a few volts lower than lead acid (my CV voltage is set at 55.2 V), but I don't think that would affect solar charging significantly. The charge bug issue is described in the "discussion post" about the latest patched firmware. The essence of the problem is that the Axpert judges the charge to be finished when the charge current drops below a certain value. That could be because the battery is full, or because there is a cloud. It can even happen with utility charging, with perfectly steady charge current, if you charge at a rate less than the threshold. For most cases, that threshold is a fifth of the maximum charge current setting (parameter 2) (the threshold is different for the LiFePO₄ patched firmware and/or with paralleled inverters). Prove it for yourself with a test utility charge: set parameter 2 (maximum charge current) to 120 A and parameter 11 (maximum utility charge current) to 20 A. The threshold will be 60 A, and the charge current will be 20 A, so after 10 minutes (after only 30 seconds with pre-72.70 official firmware) the charge will change from bulk charging at 20 A to float charging at a handful of amps (depending on the exact state of your battery). Details here. You might not notice the charge bug, especially with official firmware version 72.70, because when you watch it, you might not get 10 minutes of uninterrupted cloud. It's harder to miss with the earlier versions, because they only needed 30 seconds of cloud to terminate the charge, but you could still miss it, especially if your weather has been like it has been in Brisbane Australia lately (weeks of nothing but blue sky; my rainwater tanks are empty). If you remove *all* solar input, the SCC will turn off, but you can probably simulate 11 minutes of cloud by turning off the isolators / breakers for most (not all) of your solar strings. Watch the battery voltage on your monitoring software or on the LCD display, and watch for the change from bulk/absorb to float charge, on your monitoring software or watching the charge LED on the front panel go from solid (bulk/absorb) to flashing (float). It's really scandalous that this charge bug hasn't been fixed after all these years. Charging batteries is a major job for an inverter/charger, and a battery worth more (sometimes many times more) than the cost of the inverter may be ruined as a result of the bug, which is probably a single line of code to fix.

Pretty much, yes. There are some relays inside to do the bypass for you. No, the power only flows into the battery when in bypass mode, never from battery to load and AC in. [ Edit : only hybrid inverters can blend power from the battery and inverter with an AC source. The Axpert does however combine any solar charging with AC charging.] I'm not aware of a three stage switch ; that sounds like something added to your installation. You can't force the inverter into bypass mode with isolators, but I believe that you can do it with monitoring software that sends the appropriate commands (e.g. output source priority). The inverter will switch to bypass mode automatically and without fuss when the load exceeds 5 kVA or 4 kW for more than a few seconds, or if the battery voltage droops too low. When battery and load are suitable for 10 minutes, it will switch back to battery mode (if priority is set that way) automatically. The inverter output phase is synchronised with AC input to make the transfer smooth.