Coulomb

Actually, it has to be that. So, the connection to the battery had better not be "little". A friend of mine is using 50 mm^2 cable, though that's largely because of the high (18 kA) short circuit current of the lithium battery, and an Australian Standard requirement to withstand a short circuit for 1 second. (The battery cables are actually two such 50 mm^2 high temperature rating double insulated wires in parallel). I'm copying his design, except I'm using one inverter (he has two paralleled, single phase). So we have plenty of 50 V bus sharing. They need to be good for over 80 A continuous (4000 W divided by 50 V, possibly 46 V for lead acid, plus losses) in any case. Edit: oops! I could be guilty of foot-in-mouth disease here. By little black and red, you were probably referring to the "current sharing" cables. I was talking about the inverter to battery cables. My apologies if I seemed pompous above.

No problem, Gnome. This is actually a case where we have source code from the manufacturer. I hear a collective gasp. Well, I imagined one. How was this obtained? One of the users on the AEVA forum asked them by email, and they immediately provided it. Who'd think of that? Though now that I think about it, that was very early on, first or second page of the loong PIP/Axpert topic, when we thought that MPP Solar was the manufacturer. Perhaps they were more helpful than Voltronic would have been, had the user asked thrm instead. In any case, the sample code has the three cases where the CRC character is incremented. If this was a deduced thing, or even a reverse engineered thing, then we might fear that there could be more cases. It's good to hear that the CRCs hold up, i.e. that the slow serial line is reliable, even with a high power switch mode converter in the same metal box.

I've not noticed this, but haven't been looking closely. QPIGS has a very long response, and the response values change a lot. So it's difficult to spot if one data element has changed a little. Perhaps set up a small script or something to spit out regular shorter commands that have a fixed response, e.g. QID or QVFW. If they never seem to vary, then perhaps try larger responses with fixed data (e.g. QPIRI, ratings enquiry). If the CRC when wrong always has one character that is one more than you expect, then perhaps you;re not handling that situation where the CRC would end up with a carriage return or linefeed or open parenthesis; in these three cases, the CRC character is incremented (so CR (0x0D) to ^N (0x0E), LF (0x0A) to ^K (0x0B), and '(' (0x28) to ')' (0x29)). If you get errors, even increment by one, for other received CRCs, then either the CRC is being sent incorrectly or received incorrectly. To figure out which of these, I guess you need to try different computers, USB to serial adapters (if used), different and shorter cables (if relevant), and so on.

I've added a second edit to my post at the top of the page; it's been pointed out to me that on the AC side, there could be say 3300 VA worth of utility (AC in) charging current as well as bypassing to loads.

Indeed, there doesn't seem to be anything suggesting what current (and hence power) the Axperts can carry in line mode. In the absence of such information, I'd assume that the line mode output is the same as the inverter (battery) mode output, i.e. 5 kVA for a 5 kVA unit. The relays are rated at 40 A (at least, that's what is written on them), and the AC input circuit breaker is a 40 A unit. That would suggest 230 x 40 = 9200 VA, but I think you'd be mad to put 40 A through one unit, or even 8000/240 = 33 A. All that current has to flow through printed circuit tracks, line filter chokes, the circuit breaker, and several relay contacts. [ Edit: the bypass on overload could also be for if the inverter gets too hot, i.e. overloaded in the temperature sense as opposed to the current sense. ] [ Edit2: You could conceivably be charging the battery with 3000 W from the AC input as well as 5000 VA from the inverter, so that's getting up near 40 A on the AC input side. So that likely explains the 40 A circuit breaker value. But I don't think it's intended to supply 40 A continuously. Indeed, with 3000 W into the battery (call it 3300/220 = 15 A max), you could only have 40-15 = 25 A out without exceeding the rating of the relays on the AC input side. ] [ Edit 3: Then again, the up to 200 A of battery current also runs through PCB tracks, albeit somewhat reinforced, so 40 A isn't necessarily ludicrous. ]

The two possible issues I'm aware of are EEPROM wear and contact wear on the line relays. Most of us would have fridges and freezers as loads, and their motors have significant inductance. That can be hard on relay contacts. However, the inverter seems to sync with the line, so switching should be moderately soft. So overall, I see the risks as low. Why not use it to its full potential?

I'd be very surprised if the patched firmware fixed the problem of parameter 01 changing every night. Have you checked that the two inverters separately (with the other turned off) are set the same? Just grasping at straws here. The other thought is monitoring software; it may be changing modes and reconfiguring it may be needed.

I also think it may be normal. I don't use lead acid batteries, so the SOC reading is even more useless to me, so I ignore it. I know that certain events can cause the "battery low" level to change, and the battery low level is used in the SOC calculations, so some step changes in reported SOC are to be expected. If you have or get a BMV, then use it for state of charge and ignore anything from the inverter either with the battery bar graph on the LCD, or any SOC readings on Watchpower or other software monitors that are based on the inverter's concept of SOC.

Ok, so then you won't get any benefit from updating to official firmware, unless you need the patched firmware. > What is causing a problem is that setting 01 which I set to SbU changes to Uti by itself every now and then. I've noticed that with paralleled machines, the master seems to override settings on the slave, in the sense that if you read a setting on the slave, the master may give its setting. I don't know for sure, but the slave might still retain a different setting, and sometimes what is now a slave may later become a master, so then you'd get the different setting. To make sure this can't happen, you could program the settings independentl to make sure both machines agree. Maybe I've got that wrong, I haven't played with paralleled machines all that much. A friend has a setup where he can power up one or the other machine by pulling a fuse before starting up. It does pre-charge automatically with contactors, so it's quick and no splat.

People, sorry for the confusion. I neglected to use the word "patched" in the original post. Now corrected.

Check the files section of this forum. There are some repair manuals with some good information in them.

Sorry, the index page was a bit ambiguous, especially if you are mainly interested in the lead acid version (as most of you here are). Hopefully clearer now. Mark, I replied about the same time as your question; sometimes the post seems to disappear. Perhaps refresh the page. the essence was: no, no dramatic changes, unless you have a machine that has and needs 72.60 firmware.

I've not heard of 2.40 SCC firmware. Usually 72.40 is paired with 04.00. Could you have misread the U2 display screen perhaps? I would not upgrade just to go from 72.40 to 72.60, unless you wanted a patched 72.60. However, if you really have 2.40 SCC firmware, that might be worth an upgrade. But if everything seems to be working well enough, then no. A USB to serial/RS232 converter is fine; as you say, very few of us have computers with real RS232 ports these days. Just be aware of the potential problem with the virtual COM port having a numeric part of 10 or more (COM10 and higher doesn't seem to be seen by the reflash software).

The changes from 72.40 to 72.60 seem to be essentially limited to supporting the new hardware with 80 A MPPT(s). So if you're happily running the lead acid patch, then there is no real need to update. Sorry, I should have mentioned that for this audience. If you are running a LiFePO₄ battery, then you may benefit from the higher range for back to utility voltage (parameter 12), and if you have higher than average parasitic loads, you may benefit from the divide by 24 criterion rather than the divide by 30 criterion, for deciding when to transition from bulk/absorb to float. There are also cosmetic changes like the "font" rendering, so if you're one of those types that has to have the latest and greatest...

New patched firmware is posted here: http://forums.aeva.asn.au/forum_posts.asp?TID=4332&PID=63036&title=pip4048ms-inverter#63036 Edit: my apologies! I forgot thr all-important word "patched". So there is nothing new from Voltronics here, but a new patched firmware based on 72.60 rather than 72.40.

A colleague has just confirmed that 72.60 doesn't work properly with 04.00, so maybe 04.10 won't work properly with 72.40. So my suggestion is that if you change one, change the other as well. With 72.60 and 04.00, when the total charge current is > 60 A, then the inverter acts as though the SCC isn't there (except you still see a U2 display screen). There is no solar panel icon, the PV voltage and power report as zero, PV current doesn't even appear, and no solar charging occurs. For me, in this condition (mismatched firmware and max charge current > 60 A), it also seems to reset every half to one minute or so. My colleague doesn't seem to get that. Maybe it's my temporary connections (longish 8 mm² battery cable, for example). But updating to 04.10 SCC firmware fixes the SCC being ignored when the max charge current setting is over 60 A.

It turns out that the name of the command isn't QFTY as suggested by the function name; it's QFTY5200. This is an unusual command in that no CRC is needed, but also must not be provided. I hope I'm not revealing what is supposed to be a secret here But someone posted something on the internet that was wrong, and it was me! So I had to correct that info.

Peter, I have an MPP Solar branded PIP-4048, and it has much the same label, except the model number on the label is "PIP 4048MS". As others have mentioned, it's probably one distributor saying to go to the other distributor that made and sold it to you. All the real Voltronic models are designed in Taiwan and manufactured in China. If the version number of the firmware (U1 setting) is 72 40, then I'd be very confident that it was the same thing. Developing the firmware is a large part of the cost of clones, so they usually copy it verbatim. That's why the later Voltronic firmwares have a command that spits out their name. Maybe I should not be saying that, maybe it's supposed to catch out the copiers. But they would have to be pretty slack not to have figured that out, and it's easy enough to patch. Most in Australia use MPP Solar or Giant Power brands. In South Aftica, the Axpert and Meccer brands seem popular, but I think that there are others. As far as I know, these are all made in the Voltronic factory in China, or made for them by someone else, and are all identical apart from cosmetic things like logos and colours, perhaps slightly different metalwork. Likely the electronics is 100% the same, or at least functionally identical. As noted elsewhere, the lack of manufacturer name on the label is curious, but all the labels seem to be like that, so it's not a red flag by itself. The only illegal clones (as opposed to officially sanctioned branded models) I know of are branded MUST, and all of these seem to have a stripe at the bottom of the front panel. But of course, these things can and will change quickly as the copiers attempt to dodge law suits and recognition by potential customers. http://www.mppsolar.com/v3/company-announcement-must-power-must-ups/ To be completely 100% sure that you can update your firmware, you could copy your existing software so that you can update it back again if necessary. Or merely compare the binary or hex files with the official firmware file. It's not very hard to do this, but you need some uncommon equipment, and it helps to be familiar with the TI C2800 chip family. I have all that, but I'm unable to make house calls to South Africa. If you're really paranoid, you could take out your processor card and post it to me for firmware checking. But that would leave you out of commission for a week or so, depending on shipping speed. I suppose since you have master and slave, you could send the processor from the slave and limp along on one inverter while it's away. Send me a PM if you are serious about this.

Maybe the gradual dawn is still your problem, and by resetting in the middle of the day, you start bulk charging with good sunlight and it gets past the termination condition and you get a full charge. But perhaps tomorrow or the next day you'll have the same problem at dawn.

You repaired board likely has the latest firmware, so upgrading firmware is probably not an option. A patch could be developed, but I don't have a 24 V system, so you'd have to wait for someone else to make one. This doesn't seem likely, unfortunately.

[ Edit: oops, didn't notice Chris' reply over the page. ] Every software thing needed for updating firmware is included in the update archive (.rar or .zip file). You don't need any other tool or software, other than the Windows operating system. Well, plus something for reading the .doc file, but most Windows installations would have that.

You certainly need lots of support from the firmware to do paralleling. I haven't looked at the Infini firmware much, and what I have doesn't seem to match what people are running, so it's impossible for me to be definitive. But there does seem to be different firmware with the 3 verses the 3+.

I think what may be happening is that the long term load of the microwave is not too much for the inverter to handle, but there is a short term overload. Many microwaves have a large 50 Hz transformer. (I believe the ones with an inverter do not.) Any large transformer will have a high inrush current to establish the magnetising current. The magnitude of this current will vary with the part of the AC cycle that it starts on; counterintively, the magnitude will be least if it starts at the maximum voltage point. So experiments with this sort of thing have a random component to them. So perhaps the surge from the microwave is enough to trigger bypass if bypass on overload is enabled, but not enough to trigger an overload fault if bypass on overload is disabled. But it's also a mystery as to why the bypass doesn't come off (with bypass on overload enabled) shortly after the overload is no longer detected. So I don't think that there is much wrong with your setup. Perhaps when it comes time to replace the microwave, consider a model that has an inverter inside, so you're not needlessly bypassing to grid so often.

It could be the premature floating bug. You say your solar power comes up in a rush when the sun clears the mountains, but that means you could have an extended dawn with little light for half an hour while the sunlight is indirect. During that time, there may be enough light to trigger the bulk charge, but not enough to get over the threshold where it goes from bulk/absorb to float. So it thinks that the battery charge current is tapering down at the end of the absorb phase, when in reality it is tapering up at the start of the bulk phase. So by the time your blast of direct sunlight comes through, it may often already be in the float phase. Some observations of the charge LEDs around that time should confirm. The bulk/absorb charge termination threshold does change with paralleled units. Did the change in behaviour from usually getting to bulk to always seeming to be in float coincide to the second Axpert being installed? Edit: if the extended dawn theory is correct, my lead acid patched firmware should fix it. Details are on the AEVA PIP-4048 thread.

I don't know the Infinis much, but I'd be surprised if DIY paralleling was possible. Keep on mind that imperfect synchronisation when paralleling likely means kablooey.