PJ is standing me up......

Definitely a chinese solder job.  Looks like 6 is bridged to 7 (makes no difference if all switches on).

Have you tested for bad/blown resistors?  Dip switches actually closed (making contact)?

9600 watts continuous from a 15kw pj is doing pretty good!

On the 30kw, I wonder if you have a case of dual mainboards fighting each other?  As there is no provision for balancing between them, it's possible.  Don't know how you would test for that except by disconnecting one mainboard, then see what happens.  Sid probably has a better diagnosis.

Sean has talked in videos about "sync issues" with twin mainboards, etc.--well, let's just say that none of that is electrically possible.  The 2 connectors on the control board are hardwired together, so both mainboards are getting identical signals.

What CAN happen is...with two mainboards, now the (already) overloaded LF Driver has twice as much work.  With it struggling to turn FETs on and off quickly enough, the FETs end up spending a lot more time in their resistive range (potentially also encountering cross-conduction or shoot-through)--causing a LOT more heat.  Secondly, they usually use a ridiculously long ribbon cable on the 2nd mainboard (well, both for that matter)...which adds considerable inductive ring and resistance to the already-too-weak signals.  It's just a rash of poor design.

I knew he would!😁

So then the question becomes;  Is having two mainboards actually an advantage?  Do you gain any power-handling capacity? 

Seems to me you would gain more by adding a second transformer (or increasing single transformer size)

1 minute ago, dochubert said:

I knew he would!😁

So then the question becomes;  Is having two mainboards actually an advantage?  Do you gain any power-handling capacity? 

Seems to me you would gain more by adding a second transformer (or increasing single transformer size)

The way PJ does it...no, twin mainboards is a liability, not an asset.  This was pretty easy to determine with the 15kw inverters with twin LONG mainboards...simply unplugging the 2nd mainboard would drastically reduce heat and no-load.  (I actually was able to reduce no-load current on a PJ inverter by increasing the dead time setting on the CPU...which is pretty telling.) 

With the smaller mainboards...I recently installed a GS control board setup in a "15kw" PJ inverter which had twin smaller mainboards, and it seemed to be driving alright.  Worth noting that the GS LF Driver is much stronger (4.0A) than the PJ driver (2.5A / 1.0A), and so can handle heavier FET loads.  This PJ inverter had only 4 FETs per mainboard, and they were the lower capacitance NCEP039N10 (as compared to the much harder to drive HY3810 FETs they used to use).  So in total, across the 2 mainboards, it basically had the same capacitance as a single "long" mainboard with HY3810 FETs.  (Yes, I did remove over a foot of ribbon cable from the connectors!)  Last I heard, that inverter was still doing dandy.

The main power limitation in a PJ inverter is the transformer.  Second behind that is FET driving...and then maybe PCB layout and wiring.  Last on the list would be the FETs; more often than not, a very conservative rating on the FETs ends up being far higher than the inverter's rating. 

For example, the NCEP039N10 FETs have a 100C rating of 108A.  (I like to use this as a "conservative worst case" number.)  108 * 6 FETs per quadrant = 648A * 48v nominal (worst case) = 31.1kw redline on the FETs.  So for a 30kw inverter, PJ would need more than a single mainboard--but this of course is assuming that all of the other problems have been solved, otherwise the FETs will smoke long before reaching their true capability.  (Inductive ringing / signal crosstalk / induced noise in the ribbon cable can each VERY easily be fatal to the FETs even with very small inverter loads.)

Update: 

I installed a spare output control board that I have, or whatever it is called. I put all dips on and no resister in the spare spot. And it starts a/c with single flicker of the lights. That is a heavy surge to start. 40 amps 240v when it is running. Not sure what surge is.

The transformer in this thing is huge. Weight is 150lbs just that box alone. It has 2 long main boards with 24 mosfets and 6 capacitors each.

I guess trick now is to get PJ to send me a good board, lol. 

I see if I can do something to ribbon cables, maybe some ferites to stop noise?

We've experimented with ferrites on certain strands of the ribbon cable--but while we've tried all sorts of cable tricks, that's only half of the problem.  The remainder of the problem is with the mainboard layout....

It doesn't help that the high FET's source / gate wires are intertwined in the worst possible way--for functionality anyway.  If the goal is crosstalk, they couldn't have done better with the cable pinout 😉.  Still, with a completely different cable (separate single strands of shielded wire on each FET pair), the crosstalk problems are still there.

You would need a good DSO (digital storage scope) to be able to see what's going on with the FET gate signals.

But glad it's running.

I appreciate all the help and tips. Hopefully this is not all in vain and there is something for others to gain. I will follow up after If any more develops with PJ's repair solution.

Thanks again, Dave

I wouldn't be surprised if there's a hairline crack, bad solder joint, or some other issue between the E135-LF-L transformer and the shunt resistors, on the bad board.  While I don't know the layout of a 30kw PJ inverter, I know the <9kw PJ designs hang the "output" board off the end of the control board with 2 screws--that's pretty much asking for hairline cracks, broken parts or other delights.

On 5/25/2021 at 8:52 AM, Sid Genetry Solar said:

The way PJ does it...no, twin mainboards is a liability, not an asset.  This was pretty easy to determine with the 15kw inverters with twin LONG mainboards...simply unplugging the 2nd mainboard would drastically reduce heat and no-load. 

I got one of the dual longboard inverters awhile back.  Initial test run had it heating up very quickly at no load.  I removed the second mainboard and it runs fairly decently now.  It is still a backup as my old-style clamshell 3 transformer model runs much cooler, quieter, and more reliably handles higher loads. (V3.6a control board)

On 5/25/2021 at 8:52 AM, Sid Genetry Solar said:

With the smaller mainboards...I recently installed a GS control board setup in a "15kw" PJ inverter which had twin smaller mainboards, and it seemed to be driving alright.

Do you consider the smaller mainboards to be a better or 'less prone to problems' alternative to the larger mainboards? (talking single board, not dual)  Better pcb layout?  I have always considered them less desirable due to smaller heatsinks and therefore lower heat disipation capacity, but am willing to revise my view.  Also, they allow less caps.  Original design called for 6.  How many is 'enough' and how many is optimum?

1 hour ago, dochubert said:

Do you consider the smaller mainboards to be a better or 'less prone to problems' alternative to the larger mainboards? (talking single board, not dual)  Better pcb layout? 

No.  Same design flaws in each.

1 hour ago, dochubert said:

I have always considered them less desirable due to smaller heatsinks and therefore lower heat disipation capacity, but am willing to revise my view.

Personally, I have a hard time seeing the smaller mainboards being able to handle 600A without melting.  Yeah, smaller heatsinks and all...

1 hour ago, dochubert said:

Also, they allow less caps.  Original design called for 6.  How many is 'enough' and how many is optimum?

Not entirely sure; there's probably mathematics to figure this out though.  Basically, the caps main purpose is to suppress inductive oscillations in the battery cables / wiring.  Without them, the SPWM switching of the FETs will turn the battery cables into inductors--very quickly blowing the FETs to smithereens.  However, this power filtration comes at a cost: electrolytic capacitors have a maximum ripple current (a factor of the ESR and heat dissipation ability)...which if their ripple current is exceeded, they will overheat and fail.  (Let's not mention that most electrolytic caps are rated for a lifetime of only 2,000 hours--3 months--at full rated voltage, temperature, and ripple current.)  Adding more capacitors divides the ripple current out...the challenge is knowing how much ripple current there is, and what the caps are rated for.

In short: the danger of too-few (or junk) caps is that they'll overheat and fail.  Or not be able to squelch damaging oscillations in the battery wires (in which case the FETs would fail), etc.

9 minutes ago, Sid Genetry Solar said:

(Let's not mention that most electrolytic caps are rated for a lifetime of only 2,000 hours--3 months--at full rated voltage, temperature, and ripple current.)  Adding more capacitors divides the ripple current out.

Now I'm concerned!  Was blissfully unaware that I have exceeded the lifetime of my inverter's caps by many times already.  Ran this 48v/15kw pj inverter 24/7 from apr4 last year to oct26, then as sunlight was available thru winter.  Went 24/7 again mar15 this year.  Expect to go till late oct again.  Do you have a recommended replacement cap for these that has a bit more lifetime? (10000mfd 80v) 

It seems I should be replacing them at least once a year, and 4 times a year by your 3 month figure. (Just what I needed.  Something else to worry about.)😟

Old capacitors aren't necessarily faulty capacitors, but it's probably a safe assumption with anything PJ uses.  Victron had a discussion about AC side induced DC ripple and recommended that if you see more than 200mV (I think) peak ripple on the inverter's DC input under full load you need to improve the wiring and/or battery capacity but, by extension, if both those are good enough for the load it's time to look at those capacitors.

Didn't mean to cause you panic; the manufacturer's rated lifetime is an absolute worst-case scenario: running the capacitor at the maximum rated temp (often 105C--closer to 220F), maximum voltage, AND maximum ripple current.  If the cap is at none of these max ratings, they can often live for years without issues.

My introduction to the damaging effects of ripple current were when I overheated and caused a 100uF 400v electrolytic cap to pretty well explode into a huge cloud of smoke.  I had been running it at 50vDC, but with a PWM "brake" on a motor.  Needed the cap to suppress the motor's splash to save the FET--but the 60Hz short-circuit PWM pulses meant immense ripple current.  Took roughly 20 minutes of tests before the cap blew up.

PJ DID have a rash of bad caps about a year ago--they switched suppliers, and got rebranded garbage.  Took 'em a few months to realize that the caps were junk, and they went back to the old supplier.  Apart from that, we aren't aware of any undue garbage caps in PJ inverters.

My house inverter is just over 3 years old, still sporting the original FETs and caps (that and the chassis are about all that's original right now, hahaha!), no issues.  If the caps aren't swelling/bulging, you're probably just fine.

I installed a spare output control board that I have, or whatever it is called. I put all dips on and no resister in the spare spot. And it starts a/c with single flicker of the lights. That is a heavy surge to start. 40 amps 240v when it is running. Not sure what surge is 

The  30kw PJ  inverter  has two  large  mainboards   and  output   at least  40 amps  from this forum .    That  is  near  10000 watts  from  30kw  PJ  .    Some   20kw PJ inverter has  2  mini mainboards  with   4  mosfets per board  and  2 capacitors  on each  mainboard  .    His  20kw PJ  has  6 mosfets per board  and  more than 2 capacitors on each  mini  mainboard  and  output  9600 watts .     It  seems like  the large mainboard  and  6 mosfets  per board  and  6 capacitors on mainboard  make  a difference .   Is  that a   ASL 10  transformer  ?