Hi Mike, that's never left on as I have remote switching via a big Relay AC relay, the dc coil is controlled via a mechanical on/off switch and small PSU, however I did wonder if the pressure controlled Water pump had come on and possibly shorted / failed, but everything was fine and switching all loads off still tripped another breaker with Mains AC, after the 3rd trip the short disappeared.

But even allowing for all of that, nothing could pull over 63A from the AC line for 6 minutes except a partial short (equipment burning up) or some organic matter across the AC lines somewhere.

Everything has been running for 24 Hours again with max power around 6kW, if everything aligned to run at once, worst case would be around 10kW to 12kW, not including startup surges, but 17kW for 6 minutes?? external monitoring confirmed that, followed by the Inverter C63 AC Breaker tripping.

I now believe from looking at the logs, whatever was drawing the extra 10kW suddenly shorted and the Breaker, already under stress, tripped instantly BEFORE the Inverter DC tripped.

When I reset the breaker the first time under Inverter AC, Inverter DC input current was heading north of 800A, and this time the Inverter tripped with both power board DC input trip LEDS lit and the AC Breaker also tripped again.  

One thing for sure, the DC protection works and the available power and load sharing on this Inverter design is impressive.

  analog8484 said  

  KeepIS said  

High current draw then Short on Inverter AC line.

Looking back at the LOG, an Unknown fault on the AC wiring had caused the DC input to run over 17Kw, over 30kW peak, into the Inverter for at least 6 minutes.

.

Good thing nothing was damaged and you were able to verify the protection mechanisms worked as designed.  So, this mean the inverter can run 17kW for at least 6 minutes?

I ran the initial single stage Inverter at over 8kW for over half an hour without any real issues, so yes, I new the Dual Inverter could run at that power level and for a lot longer than 6 minutes if built correctly, but I was unsure about the BMS's and Low value DC fuses, I use these fuses on each of the four paralleled LiFePO4 banks, these are my fail-safe should one of the BMS's go south or some other life changing event happens to the battery banks paralleled 53V DC supply BUS.


.
Edited 2025-03-16 16:10 by KeepIS

Hi KeepIS, any underground conduits involved?
I have a 100m of 6mm2 in a buried conduit descending a hillside.
After several decades of rare random combined breaker/safety switch trips it coincided with rain. Resetting did not repeat trip but delivered no power.
Pulling the cable told a story; three rat skeletons, each stripping a 200 to 300mm section before the inevitable. Finally water run down the conduit and the arcing burnt one leg open before tripping.

Your rats may have also taking shelter in the storm? The rain filled their jug to boil for a cup of tea?
The issue remains somewhere lying in wait to repeat at an inconvenient time. If there was arcing the carbon deposits may have increased the leakage to be measured. Did the sheds earthing remain intact?
It’s nearly Easter, enjoy as if hunting Easter eggs.

By the way, thank you for sharing your build steps and its evolution. I now have the Wiseguy pcb’s and bits ready to build. Uncanny, turns out with binoculars I can see his roof from mine.

I arrived at the same conclusion a few days ago, 25 year old underground feeds to the shed, a recent cyclone with prolonged rain, and process of elimination now points to that as the only logical conclusion.

Rats and Vermin cannot get access to the underground conduit, shed AC earthing is via a large independent ground connection in the conduit. So far it's playing well, but still needs to be checked under much higher power loads, only one earth leakage trip so far.

Great to see another build, there were many twist along the way to this final Dual build, many turns and dead ends in my threads, quite a few rabbit holes and many Senior moments, it's great to have forum members to point out the bleeding obvious when you're up to your armpits in Cabling, FETS and case construction.

I've finished rebuilding the Solar Charger Wall and it was worth the effort.

There are Seven * 60A solar chargers and a 1.5kW AC battery charger, the same physical size as the solar chargers, just slightly wider, likely never needed but got one for 75$ so what the heck.

The load sharing and switching between shaded panels is even better than before the rebuild. Each Charger runs a Max of 40A output for the amount of solar connected to each charger, this gives plenty of headroom in a 60A charger, so 280A @ 54V.

I then decided to try and measure the overall Inverter efficiency over Long term running under all load conditions encountered in the off grid system.

I rewrote my old kWh Code in my Solar Monitor, it's running on an ARM H7@480mHz with a 9" touch panel.

I have three commercial AC kWh meters, one in the Wall Fuse Box in the AC output path from the Auto Transfer switch, one on each Toriod output (1/2 AC power) with a combined AC kWh meter mounted on the Inverter.

So unless these are all crap and nothing can be believed, they should be able to give me some meaningful comparison long term results.

The idea is to measure accumulated DC kWh input into the Inverter and compare it to accumulated AC kWh out of the inverter over an extended time frame.

This should account for low power overnight running and high power daily running under all types of reactive and restive loads and combination of loads.

I note that bad loads (like Diode in one side of the mains feed) show a higher DC input compared to AC output load, over time these short transient loads should fade into the accumulated kWh background (I hope).

I programmed the Solar Monitor to allow me to enter a power loss in watts for the Inverter, from then on the Solar Monitor calculates the AC output in kWh from DC input kWh, accounting for Idle power + Conversion power loss and displays it in real time for total running time kWh, the display just ticks over every minute and all I do is look at the AC kWh meters and see if they match the DC calculated input kWh, changing the power loss value until they match and track within 0.03kWh will indicate the conversion loss as a % of total running kWh by this method, how useful and meaningful it turns out to be remains to be seen.

Right now it's tracking the two AC kWh meters to within 0.01 kWh after 30 hours.

The Test loss value entered is 50 Watts Idle + 70 watts conversion loss, a total of 120 Watts in kWh over each accumulated kWh.

The total conversion loss including Idle power is 14% after 30 hours running time, not really long enough but it's in the ballpark which is encouraging, it's getting less as the heaver daytime loads start mounting up in kWh, most of the 30 hours run time has been below 1kW which reduces efficiency due to % of idle power to output power.  

This should account for all loss through the entire path from battery to AC out, if nothing else, once set up, it might be a good indicator of changes in Inverter running efficiency.
.
Edited 2025-04-11 20:17 by KeepIS

What, no pictures?

So this 14% loss is measured from the output of the MPPT compared to the 240v output?

If so then that is really good, particularly when you consider that the 50w is about a 5% loss right there. Figure 1.5% for battery and 7.5% for inverter, not including standby power?

Quick question.  I notice at low loads your inverter PWM is running at ~71%.  How high does it get with a 6kw or larger load?
Edited 2025-04-11 20:47 by oreo

Just got back to the PC today, picture of the Solar wall to come, I'm still waiting for Solar Controller number 7 to arrive      

The commercial AC kWh Meters claim .05% to 1% and obviously take power factor, AC distortion etc correction into account.

The problem is Inverter DC input cannot compensate for the effect these have on DC input current. I've been busy getting the kWh granularity accurate at the millisecond level, I wanted to be absolutely sure that very fast repetitive transient high DC current variations are accurately counted.

I'm testing taking one sample per second, converted to kWs (Kw per second). I will then change that to accumulated samples every 10ms and 100ms for each minute.

It takes time to correctly verify between various sampling methods under complex loads to see if it really matters compared to one second sample times.

If you spend time looking at DC input current through a quality isolated Hall Current sensor, you will see some crazy HIGH current pulse patterns with various machinery and some Household electrical appliances.

I sometimes look in wonder at how this Inverter can handle these loads without flicking overhead lights or making a noise from the Toroids, I guess a 100kG Inverter with 6 big Toroids makes a small difference as well

The commercial AC kWh meters are pretty close in the results between them, and I have a commercial high quality AC analyzer which I will use once I have confirmed the full accuracy of the DC kWh reading, again I'm hoping that after a few weeks of accumulated kWh, I should be able to get useful values that can be used to verify if over time, the efficiency of the Inverter is changing and give me a meaningful efficiency %.

Currently my Solar Monitor with modified kWh code is indicating the following:

Inverter total loss of 12.31%
DC to AC conversion loss 4.06%

The above are updated every kWh.

Solar Current sensor mounted right after the combined Solar Reg output Bus.
Solar output voltage measured at the combined Solar Reg output Bus.
 
Inverter Current Sensor mounted on DC feed to the Inverter, close to the Inverter.
Inverter DC Input voltage is measured right at the Power Boards.

The Monitor has measured most of the following for years, but I've just rewritten the code for more accuracy for playing with this thought experiment.
   
The Monitor measures and displays kWh for:

Total Solar output from all Solar regulators
Total Solar to the Inverter
Total Solar to the Battery banks (charge)
Total Battery to Inverter (discharge)
.
Edited 2025-04-14 03:44 by KeepIS

This is running really well and tracking AC kWh meters correctly.

As many have found, the Solar regulators are a big source of Noise on the DC Bus, and last night with no solar and Load power under 520 Watts, the DC wattage on my monitor, minus Idle power and Efficiency loss was tracking AC Wattage perfectly.

The Monitor current sensors match my Clamp meters, I have few, and the DC voltages are accurate to within a 100mv.
 
These readings are very stable on the Monitor, even with all Solar chargers running, but with the Chargers off, the last digit is stable and I can compare the AC and DC Power readings to within a watt or two.    

I have programmed the monitor to allow me to enter two values, 1: Idle power and 2: Expected Conversion loss for DC to AC generation.

The monitor displays % of total Inverter loss and Conversion loss %.

Obviously these two entered values can be changed and Conversion loss is adjusted to change the % of total loss to match the DC and AC accumulated kWh over time.

As expected, efficiency drops as AC power consumption gets closer to Inverter  Idle Power.

I'm starting to see the Conversion loss value I entered stabilizing as more hours are accumulated, and it appears that this might give me a good indication of overall efficiency.

My only thought was, the Idle power comprises Inverter electronic loss, which may not change much, and Toriod magnetizing loss, could the Idle Toroid magnetizing loss change with load?
.
Edited 2025-04-14 15:17 by KeepIS

  KeepIS said  

My only thought was, the Idle power comprises Inverter electronic loss, which may not change much, and Toriod magnetizing loss, could the Idle Toroid magnetizing loss change with load?
.

The toroidal magnetising current should remain constant and not change with load.

For efficiency I prefer to use the incremental efficiency value as it removes the idling power from the result. ie idling no load = 15W, add a 1KW load = 1025W input power, incremental input power = 1025 - 15 or 1010W, load power = 1KW. Conversion efficiency ignoring idling losses = pload/pin  or 1000/1010 = 99%.

Otherwise it would have looked like 1000/1025 = 97.5% which is not the true conversion efficiency.
Edited 2025-04-14 19:44 by wiseguy

Thanks, good to know the Toroid should not change.

I am removing idling power and leaving only Conversion loss as a % of kWh conversion. Trying pload/pin with accumulated kWh skews the efficiency to the 99% range.

The way I'm doing it shows overall 93% at the moment (-idle power).

As the Monitor can accurately calculate watts pload/pin (not accumulated into kWh)
I will add pload/pin Conversion efficiency for running watts (-idle) and see what comes out tomorrow.

Hope all is going well with the projects, you must be busy and over it by now.

I did a "quick" efficiency test @ 3.2kW, the result was 97% allowing for the AC digital meter jumping around slightly.    

When I removed the "Idle loss" entry from my Solar Monitor kWh settings, the Accumulated 43 hour kWh result was 95.2%.

Considering this is with a lot of inductive loads from machinery, lots of AC plug-packs, other SMPS devices, Inverter Air conditioners, Microwave and Fridge freezers.

All make DC power increase above the AC power indicated on AC multi-function meters and of course the power factor varies with these loads which I assume the AC meters are compensating for in the AC power readings.      

But I feel that as I fine turn the Monitor code and reconfirm the accuracy of the AC kWh meters, this could work quite well.  

Inverter Stopped at 2:45am

Came out this morning and found we were running on street AC, the Inverter was still on and the LCD read. "* Restart Inverter *"  with  "VCap to Batt V Delta!" as the error.

The Monitor History showed it tripped at 2:45am and the Graph showed the Hot water had started 5 minutes prior to the trip, the DC input voltage was still at 52.8V @ 3kW load on the Graph, no glitches on the graph just normal running conditions.

I'm going to check the Connections between the "Contactor" Input and the Power boards just in case, but I doubt there is a problem, it's been running at 5kW for the past hour and everything is fine.

I had been thinking about adding a small delay to Vcap Delta logic, just in case there was some momentary noise pulse, or possibly the Moon cycle has something to do with it?

To that end, I've added a 50ms delay to make absolutely sure the measured Delta difference is real, I just have to test on the Inverter before Uploading it to Download page.                        

I'll post when it's uploaded
.
Edited 2025-04-15 11:13 by KeepIS

Version 8.1 uploaded.  Version now appears in the Sig below.

EDIT: For anyone wondering about that 50ms delay before halting SPWM and requesting an Inverter restart.

The Code now reads the Delta voltage 5 times. Once each complete code run cycle, which is every 10ms. If the voltage difference is still there on each run for a total of 5 passes (50ms) it will trip, otherwise it was a "once in a blue Moon event" or an act of Murphy, and can safely be ignored.  

Worth remembering, this Inverter had never tripped a "Delta Error" before, even under daily multiple 25kw surges and even a Mains short circuit and it's running 24/7.

.
Edited 2025-04-17 08:56 by KeepIS

  oreo said  

Quick question.  I notice at low loads your inverter PWM is running at ~71%.  How high does it get with a 6kw or larger load?

Just has a look at 5.2kW and it's at 79%  - mainly depends on your turns ratio, which likely effects efficiency at some power levels.

BTW I had a chance to do Wiseguys efficiency test today @ 3.2kW and it was between 97% and 98%.

I feel I should recheck the AC power Meter again at that power level just to be sure, when I get a chance.

  KeepIS said  

  oreo said  

Quick question.  I notice at low loads your inverter PWM is running at ~71%.  How high does it get with a 6kw or larger load?

Just has a look at 5.2kW and it's at 79%  - mainly depends on your turns ratio, which likely effects efficiency at some power levels.

BTW I had a chance to do Wiseguys efficiency test today @ 3.2kW and it was between 97% and 98%.

Thanks!  On some level, the increased duty cycle is a reflection of the inverter efficiency right?  An 8% increase in DC to support 5.2kw seems rather good to me.  I plan on using 2 ~3kw transformers in my build and it will be interesting to see how the duty cycle changes with 1 vs 2 transformers.

I don't think the Duty Cycle changes, it's still the same Power level and winding ratio, only difference is that each Power Stage/FETS/Toroid are doing half the work of a single stage inverter.

If  you wind a Toriod with the typical commercial "save money" turns ratio, you will often see the duty cycle really start climbing as the the Inverter tries to maintain the nominated AC output voltage under even smallish loads, and likely higher Idle power.

The other tell tail sign is a Toroid that needs at least 48V to get 240Vac output to regulate, the Idle Duty-Cycle can be as high as 92%, along with ever increasing Idle power.

I feel (could be wrong) that the real limit in Toroidal power ratings is the size of the wiring and heat, using a bigger core to fit heaver wiring of course means less number of turns, more metal, better airflow, and is therefore easier to keep cool, however keeping a smaller core temperature under control would be a problem, even if the windings could take the same power level.

My inverter build is creeping forward again and i think i need to get a nano programmed. Question on hardware requirements for loading the hex file onto the nano:

I seem to have old computers: (win xp sp2, now with Ubuntu v16 on dual boot, or my newest computer is a 2009 imac i was given running osx 10.11 (el-capitane?), and  was wondering if the software needed to upload the hex files needs anything in particular? Or if you could recommend one of the above OS's over the others?

Just trying to avoid giving myself a headache whilst trying to make things work on my old hardware.  

Thx

Hey.  You need 2 things to load a file and program the lock bits in a Nano.  One is the program AVRDUDESS and the other is a dongle which plugs into the 6 pin connector on top of the Nano board. KeepIS has written some excellent instructions on how to do this, however doing a quick search on AVRDUDESS operating system compatibility reports "Windows Vista is the minimum supported version, though AVRDUDESS may still work on XP. " and "Only supported up to macOS Mojave (10.14). " Of course YMMV.  One possible issue is that the drivers may not work with these older software versions.  I had trouble loading the drivers on my laptop (Win 11) where it reported that everything was successful, however AVRDUDESS would not connect.  I reverted to using my desktop (also Win 11) which worked perfectly.  

You can load the software, using AVRDUDESS and a USB cable, but in order to program the lock bits, you need a dongle.  KeepIS suggests a couple in his documentation.  I used one of these.  It comes with a cable which is pre-configured to plug into either 6 or 10 pin headers.  You can load the software and program the lock bits using the dongle.
Edited 2025-04-18 11:14 by oreo

Hi, the Full Download has the details for programming the HEX file, the free AvrDudess program and how to set it up and the info on the Very small low cost Programmer board.

I've only used this on Windows, I don't recall if there is a version for Ubuntu.

Once you have read the "Nano Hex and Menu.pdf" if there is anything unclear, I'm happy to try and answer any questions, if I don't have the answer, I'm sure someone on the forum will
.

Thanks oreo,
I'm reading through keepis's documentation as we speak :) . Avrdudess doesn't seem to run under Ubuntu anyway, but I will try the other two os's.

'Only supported up to mac os 10.14' suggest my older  mac os might actually work  

And yes, I will need to find an appropriate dongle programmer for sale somewhere too...

The ones I showed are available everywhere, they are also the fastest at programming, but speed is not a big deal if you are only uploading a program once, or for any updates. But you need to make an adapter cable or get an adapter form somewhere, the Board that oreo linked to is one I have used and it comes with the correct header for the Nano, that header also fits the Faster (and lower cost) programmer that I posted and use.

.
Edited 2025-04-18 11:41 by KeepIS