Pre-Order Discussions

Wow that is a squeeze to make all that fit.

The  rev 11.3 has 4 capacitors and 24  RUI FETs from Powerjack .    This 15kw has 2  large main boards that I remove .  I added  6 capacitors on a separate  board  and replace the  working rev 11.1 control board with a  rev 11.3  control board . There is a lot of room in this  15kw PJ inverter .  There is no room in the GS 12kw inverter  to do any kind of repair .   The  3 year warranty  should make it not necessary to have spare parts  .    After 3 years I  will just buy another one  as the saving in utility cost will pay for another one .    I  can repair the PJ  maybe  forever  with  spare parts .   

Yeah that works nice.  I don't like to use the warranty if I can do it myself I will, so still hoping that they sell spares even if I never use them its a mental piece of mind to have them. 

If the inverter is in Error mode (of any kind), it's basically completely shut down, and will not restart automatically (except in several selectable cases of overheat, overload, and OVP/UVP).

A risk of making the inverter "easy to start into charge under UVP" is that if the batteries are very low (or even the wrong voltage!), the moment the mains AC relay switches on, it will dump uncontrolled power into the batteries due to the LF inverter hardware.  This could be rather problematic.

23 hours ago, Steve said:

The driver board is isolated so the logic board should be fine even after failure

Generally speaking, yes.  Biggest issue has been the power chip getting zapped with overvoltage (or negative!)...but nothing else has been damaged.  Gotta love the Chinese chips...each time I check the datasheet on the main power chip (XL7015), it's significantly changed.  And there's no markings or insignia of any sort to indicate which revision the chips are, and which limitations/improvements apply to them!

23 hours ago, Steve said:

Much of what's gone into the last several months has been to make these very hard to kill unless shorted 

The FET current limit should protect against a short circuit.  We have not formally tested this, but Sean accidentally tested it earlier this year (flipping on a breaker that went to a dryer...but which wasn't connected yet).  Inverter passed the test with a FET Amp Limit error.  Not scientific, but encouraging nonetheless!

What is basically impossible to protect against is an AC backfeed or power surge.  The issue with this is that backfeed power gets intrinsically and uncontrollably dumped across the low-side FETs--and it doesn't take too much to fry said FETs.

Yes, the GS inverters DO make sure the output is at zero BEFORE they start inverter mode (as a backfeed protection)--but what can't be effectively protected against is a backfeed under normal running operation (i.e. click a breaker on and accidentally connect the inverter output to the AC Mains while the inverter's on).  Any protection I might put here on the AC current sense lines would likely make the inverter very "trip happy" with heavy inductive loads (like an A/C compressor).  Which would be annoying to say the least.

23 hours ago, Sid Genetry Solar said:

We have not formally tested this, but Sean accidentally tested it earlier this year (flipping on a breaker that went to a dryer...but which wasn't connected yet)

Any chance that's what killed it previously?

Possible.  The previous issues were with boards that didn't have either the TVS protection diodes or the Miller clamp circuits on.  It was likely just a slow degredation over time as a result.

What's the idle down to

The clamps should effect it 

Biggest change on the new FET boards affecting the idle is that I changed the snubber circuit specification.  The previous spec (PJ spec, haha!) with 100R and 100nF got very hot.  Unsurprisingly, this was increasing idle; the revised FET boards utilize a different spec here, and the idle is noticeably reduced.

but the Miller clamp circuits enabled/disabled do not change idle current--at least to the 1mA resolution of my power supply.  Boy you can sure hear a difference in the tranny hum with/without them, though!

The biggest help of the Miller clamp circuits is under higher loads, when the Miller spike gets considerably worse--and that's where it'd kill the FETs under surge loads.  Or maybe they were dying due to avalanche breakdown.  Or maybe because body diode reverse recovery time vs FET switching time..........I dunno!

My understanding was that the spike was causing a dead short pulse

It may not have been long enough to register on the meter but would have been massive current 

It starts out very small at no load, and increases under load (and by the type of load).  As long as the FETs are in the linear region (i.e. low amplitude spike), the only noticeable result will be a very minuscule efficiency loss...and slightly increased temps.

It's when the spike starts to reach full turn-on that problems occur.  Again, a Miller spike is of such a minuscule duration that you will have a very hard time seeing this in efficiency losses.  For example, an SPWM carrier of 24KHz has a cycle period of ~42uS.  The Miller spikes seen on the 'scope generally are <50nS in duration--less than a measly 0.12% of the SPWM cycle period.

It will be very hard to see a efficiency difference when the problem is at 0.12% duty cycle--not to mention functionally insignificant at no load.

And as the Miller spike is directly related to the amount and type of loading, this explains why surge loads tend to push the FETs over the edge, by pushing them into "full-on" (=deadshort across the power rails).  The large filter caps on the GS12 mainboard are MORE than sufficient to destroy an entire bank of FETs--I know, I've done it!

November 9, 2022:

• Asked the factory for a progress update...last time we asked, they told us that they were "finishing up some other order", and would start the Genetry inverter production this week.
  • well, a lot could be meant by that...but the report now is that the Genetry inverter PCBs haven't even been produced yet--despite my providing updated PCB files literally one month ago.
  • I have a feeling we're getting strung along a bit..........an eerie repeat of the last preorder, for whatever strange reason.

2 hours ago, Sid Genetry Solar said:

 

• • I have a feeling we're getting strung along a bit..........an eerie repeat of the last preorder, for whatever strange reason.

The Chinese economy is somewhat chaotic right now. 

1 hour ago, RobertM said:

The Chinese economy is somewhat chaotic right now. 

Same could be said for the U.S., I'm sure..........

It's not a economic/trade issue that I'm aware of--because I could literally order these boards from JLCPCB and have them delivered at my doorstep in 2 weeks' time or less.

I hope we don't get the new Zero here.  Wow its like they shut everything down a few days ago. 

November 14: 2022:

• Both Sean and I have been pestering the factory with "what's going on?"  Got some responses:
  • production is now expected to finish on December 5th, after which we have the shipping time.  Seems a reasonable timeframe, considering they've only just started on the order...
  • Two delays they mentioned were:
    • the initial payment getting to their usable accounts.  PayPal is giving them grief (surprise, surprise!), and Stripe apparently took 20 days to release the payment.  Seems they could have told us about this earlier!
    • after payment was finally released on their side, they started ordering items.  (Still, that doesn't explain why PCB manufacture has only *just* started).
    • Apparently there's been some delay in getting the transformer cores, as the raw materials are imported from other countries.
  • We'll get these at some point here, if a lot later than expected.
  • EDIT: Forgot to mention that PCB production should be done this Saturday, with SMD assembly done on Tuesday.  I should therefore have photos of Genetry boards in a week....