Came home to an interesting errorr message that had shutdown the inverter.

I only had one inverter turn off via the LiFePO4 battery's BMS for low voltage and I recharged the battery from another fully charged battery using the Riden 6018 with a different PowerJack inverter.....i did not turn the PowerJack inverter back on until the battery was over 80 percent SOC, so really never saw the reverse polarity etc as i do not utilize the grid connection to any PowerJack inverter at all...

my concern is always to protect the expensive LiFePO4 battery 1st....

better???? , i can't say i had the exact failure commentors described in the gs6k inverters in any PowerJack inverter,,,,,knock on wood

and I have never blown all the mosfets up or any mosfets for that matter (again knock on wood)

I do not totally understand all these things either,,,, but hopefully you will get to the point of producing an inverter ready to ship....via all the new suppliers etc....

I am not an electrical engineer, or a licensed electrician >>>.just learning as I go....trying to avoid major catastrophes....

On 3/2/2023 at 9:45 AM, Sid Genetry Solar said:

I am suspecting this may have to do with the position of the AC input phase upon immediate restart--in other words, that the breaker trips if you happen to "slam" the transformer directly onto the peak of the AC wave...and by deduction it should theoretically be fine if the transformer was connected at the zero crossing.  If that is the case, then theoretically it would be OK to directly start the transformer from a zero crossing to full amplitude.

Notice that the transformer reaches basically a dead short (current flies through the roof, and voltage plummets back to zero) in less than 2mS when presented with a high dv/dt (i.e. slammed into a 55vDC supply!)  Conversely, an AC half-wave is 8.33mS--so you can easily see the issue here: if the transformer saturates due to whatever reason, you might as well say "bye bye FETs."  Especially on the GS12, where the transformer's 48v dead-short primary calculates out to MORE than 19,200A (and that's not a typo, either.  Yes, nineteen-thousand two-hundred amps.  FETs are only good for 8,000A if you try to use the 10uS "surge" rating!)

Very interesting information.

Bearing in mind my limited knowledge in this arena, and i'm theorizing based on some basic principals...

Seems like others suggest switching at the peak?  https://www.te.com/usa-en/products/relays-contactors-switches/relays/intersection/zero-crossover-switching-transformers.html?tab=pgp-story

This would kind of make sense to me because the transformer would already be at full potential, so no surge current necessary to restore it. Unfortunately, i don't think this could work for power-loss recovery.

On the other hand, it seems to me something similar to "parallel" mode could be useful for outage-recovery. The fets could track with the grid's wave, a large current increase felt by the fets (indicating the input voltage dropped or went out of phase) could then disengage the ATS. The fets would only be taking the load the original input was handling - the transformer never lost it's field?

If my thinking is sound (probably not :D), maybe zero export grid tie (rather the mechanism behind it) would kill 2 birds with one stone.

On 3/2/2023 at 9:45 AM, Sid Genetry Solar said:

Especially in your case with a 24v inverter, it's possible that the "ramp up" is taking longer than expected due to the batteries' ESR (as a result of the higher amperage than with a 48v inverter).  I know it's going to be basically impossible to capture, but if you can get a shot of the "DVLT" 'scope screen on a recloser event, that would be highly valuable for visualizing exactly how long the inverter's actually taking to regain output voltage on a power loss event.

I believe you're saying the sudden load causes a voltage drop on the batts which "counters" the inverter's ramping algorithm?
I have been investigating this possibility. One hangup i have is the fact that low load is more likely to cause issues.

That's a very, very interesting document.  They do rather succinctly go over why I don't want to "slam" the inverter transformer with the FETs after AC power loss...because huge surge loads like that are on the high voltage side!  400A on the high voltage side (of a 500W E-core transformer...not a 6,000W toroid!) multiplied by the transformer ratio on the low voltage side...yeah, that might not end very well.

Several years back when first trying to sort out the ATS transfer functions, I did some diagnostics along these lines--and came to the same conclusion as the above document but with a twist.  Engaging the relays at the peak of the AC wave was the best spot--BECAUSE by the time the mechanical relay contacts closed, we were at the zero crossing point.  In my testing along those lines, switching the transformer at zero crossing had a significantly lower "inrush spike" than high-peak switching.  Does beg the question, "where is the AC wave when a 6kw toroidal transformer connected to mains causes a 15A breaker to trip?"

Of course, the newer firmwares on the GS inverters do a seamless transfer from battery -> AC mains, so there is no inrush there anyway--as power never is disconnected.

Great thinking....but unfortunately there's one critical caveat...

If there's a power outage situation, the inverter must disconnect itself from the power input BEFORE attempting to drive power to the output.  Otherwise the inverter will end up trying to backfeed the entire local power grid--which will result in an overload shutdown and/or an "Output Shorted" shutdown.  (Not to mention completely failing U.L. requirements!)  And the disconnect device (AC mains relay) is a mechanical device, meaning that it will take time to physically disengage before the FETs can be driven.

This is where the catch-22 comes in: we don't shut off the AC mains relay unless AC power is lost--but once AC power is lost, we now already have a glitch that we don't want on the output!  Which is why the best solution is going to be to try to reduce the "glitch" delay.

(Worth noting that the methodology you've described is the basic operational premise between the [future!] seamless AC mains -> battery transition code.)

If you can capture a shot of the "DVLT" 'scope channel on a recloser event (or just yanking AC input power to the inverter!), that'll give me a good look at what's actually going on--and help determine how much it can be tightened up.

General rules of thumb is zero voltage turn on for resistive and capacitive loads and peak voltage turn on for inductive (e.g. transformer) loads.  Transformers can have additional issues due to remanence mentioned in the TE article.  Sudden turn off can cause remanence.  Generally, soft start helps minimize turn on issues and soft stop stop helps minimize turn off issues for inverter transformers.  However, this is not always practical if the inverter has to provide UPS functions.  In addition, inverter drive imbalance can cause remanence to build up in transformers that result in seemingly "mysterious" or "random" failures.  A reliable way to avoid this is to measure DC current and adjust inverter drive to ensure balance.

1 hour ago, JIT said:

In addition, inverter drive imbalance can cause remanence to build up in transformers that result in seemingly "mysterious" or "random" failures.

Can you explain what you mean by "inverter drive imbalance"?

23 hours ago, Sid Genetry Solar said:

Can you explain what you mean by "inverter drive imbalance"?

The imbalance refers to drive characteristics that cause net magnetizing flux difference between the positive and negative halves of a mains cycle.  Imbalance will cause remanence to build up.  One cause of imbalance is asymmetry in the PWM drive and associated circuit elements (e.g. driver or FET propagation delay differences).  External factors (e.g. load changes) can also trigger control responses that cause imbalance.  One solution is to monitor flux and make appropriate drive control adjustments when the flux get too high but it's important to make adjustments that don't degrade the output power quality too much.

5 hours ago, JIT said:

The imbalance refers to drive characteristics that cause net magnetizing flux difference between the positive and negative halves of a mains cycle. 

So I can definitely see where this could very easily happen if the transformer primary was not being firmly driven push-pull style.  (Most LF inverters, including the GS inverters utilize push-pull drive.)  But with a push-pull style SPWM drive with bipolar modulation...the transformer doesn't have the chance to build up "net magnetizing flux difference". 

It's actually quite fascinating to put the FETs into a "single-ended" mode and let the transformer dance it's own wave around the SPWM signal from the FETs--in this type of scenario, yes, I would expect things to easily get out of hand and cause random failures.

One other observation I will make from my (unscientific) testing...I noticed that sometimes when turning the GS inverters off, I'd hear a little "snap" from inside.

I traced it down to being the result of the inverter ending the sine wave "anywhere."  Changing the code to always end the sine wave (except for serious faults) at the zero crossing solved the issue--now the GS inverters don't randomly "snap" due to the collapsing transformer magnetic field when turned off.

It's possible with push-pull or full bridge PWM drive topology.  Also, as mentioned, other external factors can cause it.  Unless you are monitoring the flux you will never know sure if it's happening or not.