Status of Bench Testing of GS inverts.

Naturally, and the reputable ones state it clearly as peak efficiency.  If it's not published already the reputable ones will usually supply graphs showing the full range on request.

3 minutes ago, InPhase said:

I would imagine that there is a "sweet spot" in the efficiency of any inverter. Transformers are most efficient when loaded to near their capacity yet the FETs go down in efficiency with more loading. The point where the lines cross on a graph would be the optimal efficiency.

Manufacturers likely state the efficiency at the optimal loading point vs maximum loading.

Seems to me that most of the efficiency ratings I've seen were at 100% load.  Which of course, as I have pretty much dealt with Chinese inverters (apart from the Mean Well TS-3000), I suppose is largely meaningless.

Could you expound further on FETs "going down in efficiency" with increased load?  They have a fixed "on" resistance, meaning that the more load, the more heat generated...BUT I would contend that the lower loads are harder for the FETs, as the PWM pulses are even narrower--meaning that in the lower parts of the wave, the FETs may not get fully turned on (= resistive loss).  Higher load = bigger PWM pulses = more time spent in fully on state, instead of in transition.  Of course, this is assuming there are not other variables turning the tables at higher loads--as is the case so far with the 12kw GS inverter.  (I think we're 90% of the way there...a new mainboard and new FET boards together with the custom shielded cable might just be the ticket.)

Transformers most efficient at full load...I suppose I could see that.  There's a minimum magnetizing power, which largely is responsible for the inverter no-load current...BUT among all the different variables of a transformer, the windings DO have an electrical resistance.  This is where the PJ trannies fall so flat on their face--the winding resistances are so high that the winding wires themselves generate a considerable amount of heat. 
Still, without any graphs, curves, or in-depth knowledge into transformers, I would still expect loss to go up somewhat linearly as load increases--the limit being when it can no longer dissipate the heat generated by the losses.

All the generic Chinese inverters quote the same efficiency figures in my experience.  There's no research there, it's all just cut and paste from someone else's manual.  I've tested a few generic Chinese HF inverters, and as limited as my testing capabilities are, I can still throw a resistive load at them across a range of wattages up to full power.  The stated efficiency figure has little to do with real world performance.  It might as well be, and probably is, 100% made up.

Never more true than with the PJ documentation 😉

I completely believe what you wrote, too...

45 minutes ago, Sid Genetry Solar said:

Seems to me that most of the efficiency ratings I've seen were at 100% load.  Which of course, as I have pretty much dealt with Chinese inverters (apart from the Mean Well TS-3000), I suppose is largely meaningless.

Could you expound further on FETs "going down in efficiency" with increased load?  They have a fixed "on" resistance, meaning that the more load, the more heat generated...BUT I would contend that the lower loads are harder for the FETs, as the PWM pulses are even narrower--meaning that in the lower parts of the wave, the FETs may not get fully turned on (= resistive loss).  Higher load = bigger PWM pulses = more time spent in fully on state, instead of in transition.  Of course, this is assuming there are not other variables turning the tables at higher loads--as is the case so far with the 12kw GS inverter.  (I think we're 90% of the way there...a new mainboard and new FET boards together with the custom shielded cable might just be the ticket.)

Transformers most efficient at full load...I suppose I could see that.  There's a minimum magnetizing power, which largely is responsible for the inverter no-load current...BUT among all the different variables of a transformer, the windings DO have an electrical resistance.  This is where the PJ trannies fall so flat on their face--the winding resistances are so high that the winding wires themselves generate a considerable amount of heat. 
Still, without any graphs, curves, or in-depth knowledge into transformers, I would still expect loss to go up somewhat linearly as load increases--the limit being when it can no longer dissipate the heat generated by the losses.

Re: FET resistance vs. load. You are absolutely right about the longer pulse widths. At low load the FETs stay in the linear region longer but the current is minimal. Down to the magnetizing current. However, at high load, they burn watts like a power resistor. Somewhere in between there is an optimum balance where a maximal amount of current is flowing across a minimal voltage drop. Below and beyond which it is more wasteful, and has no obligation to match the transformer maximum efficiency point. Drive and signal problems not withstanding.🤣

Transformer efficiency is pretty intuitive. At zero load, it is still burning watts with magnetizing current. It is 0% efficient at no load. It can only get better from there.... To a point. At some point the resistive losses and steel losses shift the slope the other way. So, like the FETs, there is a point between no load and smoking hot where a maximal number of watts is dissipated in the secondary load with a minimal number of watts wasted in the core.

If I were a shady corporate inverter manufacturer, I would find the narrow region where both of those quantities are optimum and use that as my efficiency statement. And if I were a shady Chinese government sponsored manufacturer, I would completely make up a wattage for a substandard inverter and then claim it was 99.99% efficient at 210% load.

I fixed the last statement for you.😉

Uhm Sid. Help?

Sean said on the dasy vid that he connect the master to the slave and the load is out of the slave?  Uhm,   Now I'm am confused.  I ran two 25a brackers. . . do I need to make it one breaker a 50a?  geesh.  

No.  Load is on a shared terminal rail; master inverter's output goes to the panel, and the slave inverters' inputs go to the panel (separately breakered).  In Sean's case, it's a test bench, no breakers, just testing the setup.

Its justseem odd he testing it that way when its going to be hooked up different, last thing I need is to hook it up and find a issue because of it being setup different then what he doing.

Nope, same way...just without the breakers or a panel.

So do you think I'll get my babies this week, I just order a 8s bms and it came in like two days, I was thinking it would take a month, so now just installing the 8 3.2 presmatic cells, geesh 250ah, and they are reading at 3.38v, yeah boy, fixing to give my sky blue a work out tomorrow. . .

Sure hope so.  Need to tweak a few things so they work as intended, then they should be good to go.

Do the wifi has the update that you setup for me to enable it?

Also, Hope sean wrote down my new number I gave him, I've not got a email or a call.  I gave him my new number what I last talked to him and my old number is no longer active.

No.  That's one of five or six projects on the list to address in the code...after I get other more pressing things done, like making necessary upgrades to the inverter control board design, ordering sample runs of a new mainboard (and MOS boards), sample control boards...so we can order another batch of inverters, plus get the 12kw design nailed down...

Got all 397 balancers assembled and tested 3 ways over the past week...shipped to Sean today, now I need to write the manual for them...

...and tomorrow I'm back to work at the McMansion, so even less time in the day to get things done...

I can't do everything instantly.

Will buzz him to make sure.

Update on bench testing status of the parallel/"daisy" mode:

I have it working quite well on my bench...BUT it turns out my setup is a good bit different than most people will be using.  I have been running 120v parallel-inverter mode (single-phase), backfeeding an electrical outlet that had at least 30 feet of wire (and a breaker panel) before it reached the mains inverter.  Worked fine, I tested backfeeding up to 13 amps of power, with the master inverter throttling the slave down as necessary.  Figured it'd be no problem to run 240v functionality...

Sean has a setup for a much more "typical use case"...240v parallel mode, with the inverters right next to each other, with only a few feet of wire between them.  Seems that makes quite a big difference to the system--there are some regulation issues I need to address before these parallel inverters can be shipped out in good faith!

Basically, the slave inverter is able to backfeed a considerable amount of power (>2kw) through the master inverter before it has time to react.  Evidently, the "parallel" regulation functionality needs rethought.

The issue has nothing to do with the parallel function (or any hardware design), nor the "parallel" voltage being 240v instead of 120v.  One test I had Sean run, was to put his 3500W heating element in series between the inverters.  Worked a charm...though we did notice that the higher he turned the slave's regulation voltage, the more voltage drop across the heating element (unsurprisingly).  Evidently, the long lengths of wiring between the master and slave inverters on my bench tests...was serving a similar purpose.

The good news is that there are no hardware faults...it's all in software. 

Back to work 😉.  Thank you all for your patience.