Feature Request

Posted by: @inphase

↑

I realized that it would just be a software issue because the inverter has a built-in transfer switch. If the MOSFETs fail or the transformer burns out, the brain board is likely to still be functional.

 

The Request:

Can you add a setting to the menu to call for generator start and transfer if INVERTER power fails?

So I understand what you're asking for/getting at.

Technically, however, the inverter does not have an internal transfer switch (DPDT)...what it has would be considered more of a bridge switch (DPST).  The internal "mains" relay simply bridges the AC input and AC output terminals when turned on--it doesn't "disconnect" anything as required for a transfer.

Even with that, though...in theory, input could be bridged to output in the event of a system failure.

Unfortunately, there's several "gotchas" that make it quite inadvisable to do so.  (Worth noting, if the inverter had an internal DPDT transfer switch that actually disconnected the transformer from the AC Output, this would be much more feasible...although still slightly problematic.)  Let me explain:

• If the FETs burn out, there are several failure scenarios:
  • Bank shoot-through, dead-shorting the DC bus, blowing the DC input breaker.  Problem: no power available to switch the relay.
  • Partial failure, but blows a FET driver IC, shorting the FET gate drive power supply out.  Problem: this flatlines the entire board power supply as well.  (There's actually a FET gate voltage shutdown error--"Drive Volt Low"--which I implemented in hopes of being able to stop chain-reaction FET failures...but because this takes down the CPU power supply as well, it's effectively useless.)
    • Part of the challenge here is that the gate drivers are each rated to 4.0A.  The main power chip, however, can only supply 0.8A continuously--i.e. it's very easy for a FET failure to short out the main supply.  (In normal operation, the spikey gate driver power requirements are sourced from several large caps on the control board, with an average gate power consumption of under 1W--well within the capabilities of the main power chip.)  In other words, fusing or limiting power to the FET supply (in hopes of the CPU being able to continue running) is a difficult proposition.
• It is worth noting that we are not aware of a single transformer failure on Genetry inverters--possibly because unlike most other inverter brands, we don't like running the transformer to well in excess of 150C (that's 302F!)  Max desired temp limit for the transformer surface on a GS inverter is roughly 82C (180F).
  • Most likely hypothetical transformer failure would be an insulation breakdown resulting in shorted windings.

The problem with all of the above scenarios, is that in each case, you would want to disconnect power from the affected/compromised items to prevent further damage or problems.  Unfortunately, bridging AC Input to the AC Output practically provides unlimited power to both of the above items--which doesn't quite fit the bill for a solution.  (FETs get "unlimited" power from the transformer if the transformer is bridged to the AC Input.  In normal operation, this voltage is well below the battery voltage, so there is no issue.  But if the battery breaker is tripped, the FET body diodes--if functional!--form a perfect inadvertent bridge rectifier...and there's gonna be a LOT of amps behind it!)

Another feature that I've thought of (but keep shying away from implementation), would be a "Bridge on Shutdown" feature.  See, the GS inverters' bridge relay is a latching relay--it does not require any power to maintain state.  That means that it's technically possible to have the inverter purposely bridge the input and output when powered off, allowing it to be a somewhat passive "power path" without requiring any power.  Again, however, there's several caveats--which is why I haven't implemented it:

• With the processors off, there is no active current limit monitoring, transformer heat monitoring, or fan control.  This won't be a problem in some scenarios (i.e. 240v in -> 240v out, with a common neutral--transformer basically doing nothing, but along for the ride).  However in a 120v in -> 240v out scenario, or a 240v single-phase -> 240v split-phase application, the transformer will be doing notable work--and as such needs at least temp-monitored to prevent literally melting the transformer down with overload.
• As lightly noted above, the FET body diodes form a perfect inadvertent bridge rectifier.  This means that in such a case, there will be voltage with basically unlimited current on the DC battery terminals (well, at least until the FETs go up in smoke).  Mathematically, this is lower than most battery voltages will/should go, so it's generally not an issue for the battery per se--although from a safety perspective, it's not ideal ("The inverter is off, why do I have 42v on the battery terminals?!")

I hope this helps explain?

P.S. we haven't gotten the final forum dump yet.  But your suggestions for improvements are on my list.  Gonna rotate that output box cover in the design--and make sure it meets the U.L. cubic inch requirements.  Oh, and a quick search for "mechanical lug" turned up exactly the elusive terminals I'd been trying to find when designing the inverters!

@sid-genetry-solar I see. I was confused by some terminology Sean said on a phone call then. He mentioned the "transfer time" to grid input, and I assumed there was a full transfer through a DPDT contactor. Oh well. I can do it externally.

So terminology can be challenging...technically, it is a "transfer" of power via two effective switches.  One side's the FETs ("disconnecting" the battery) and the other's the relay ("connecting" the AC input).

But if we completely disregard the DC side of things and look from a purely AC standpoint, the GS inverters don't have a "transfer" switch.