15000W LF, 48volt to 240 volt, what wattage?

3 hours ago, Paul said:

I'll try to get some close up pictures of the MCU and FETs to see what they were actually using at this time, as I'm also curious.

FETs are HY3410 according to the photos you posted.  Some time more recently, PJ switched to the HY3810 (much better rated), then to the NCEP039N10--which are much better yet from a gate-driving standpoint.  (HY3810 has a relatively high "reverse transfer capacitance", a.k.a. self-destruction ability...where the FET will turn itself on if a sufficiently fast voltage transition occurs on the drain lead.  The NCEP FETs have almost no "reverse transfer capacitance", as well as a significantly reduced gate capacitance.)

HY3410 FETs have a higher on-resistance than the HY3810 (6.2 milliohms vs 5.0 milliohms), which would cause them to run slightly hotter (if apples to apples comparison).

NCEP039N10 FETs by comparison are rated 3.65 milliohms (0.00365 ohms) on-resistance...significantly better.

I believe I looked up the MCU from where someone else posted photos of actually the same PJ control board version...and IIRC that MCU is extremely barebones compared to the Microchip PIC they use today.  More details here: (<a contenteditable="false" data-ipshover="" data-ipshover-target="/profile/6-dochubert/?do=hovercard" data-mentionid="6" href="/profile/6-dochubert/" rel="">@dochubert):

<iframe allowfullscreen="" data-embedauthorid="6" data-embedcontent="" data-embedid="embed7929448947" scrolling="no" style="height:275px;max-width:500px;" data-embed-src="/topic/24-up-owe-her/?do=embed&comment=479&embedComment=479&embedDo=findComment">

3 hours ago, Paul said:

The only negative point that I have found is the high idle power (~70w). 

Do they have a ferrite core on one lead of the transformer?  If so, and you dare to mess with a functioning inverter...it needs a full 2 turns of transformer lead wrapped around it for best operation.

It did not. I tried a ferrite of similar parameters to the one used in the newer upower unit, but this only reduced the idle current from 2.5 to 2.3A @25V DC. I tried various other chokes with no better results. The only one I haven't tried yet is the ones I recovered from old 48V DC rectifiers like I tested my new tranny build with. The reason being there simply isn't enough space for them in the old PJ 'torpedo' housing.

HY3410 FETs have a higher on-resistance than the HY3810 (6.2 milliohms vs 5.0 milliohms), which would cause them to run slightly hotter (if apples to apples comparison).

NCEP039N10 FETs by comparison are rated 3.65 milliohms (0.00365 ohms) on-resistance...significantly better.

RUH1H150R   is the  mosfet on the new  mosboard  I bought on ebay from a powerjack seller .    It  run cooler  than the   NCEP039N10 FETs that  get to  135 degree F  running the heat pump . Now I only need run 2  Delta fans .    I  guess  one mosboard  cost powerjack  less than  8 dollars  and  sell on ebay  for 20 dollars  each .  Powerjack  can make the inverter  last  a long time  like in the beginning  but I think  they make more money  on repair parts  by making the inverter  last  past  2 weeks .   I   also  bought  a new  design  LF  Driver  that  has no  LED  indicator light cost 30  dollars .   It is not  Sid design  LF driver  .  

<a href="/monthly_2021_11/998475804_Screenshot(752250)mosfet.png.a2056841b07ac4cee13bbb29a3bdb04c.png" class="ipsAttachLink ipsAttachLink_image"><img data-fileid="861" src="//forums.genetrysolar.com/applications/core/interface/js/spacer.png" data-src="/monthly_2021_11/1514938812_Screenshot(752250)mosfet.thumb.png.3ae4f463dce63da9724708b039fa59bf.png" data-ratio="56.1" width="1000" class="ipsImage ipsImage_thumbnailed" alt="Screenshot (752250) mosfet.png">

I would be interested in more info on a new design lf driver.  Can you post a link?

I would be interested in more info on a new design lf driver.  Can you post a link?

The picture of the  LF driver for sale on ebay  is now more than 30 dollars  so I thought it is a new design  without  indicator light  .   Picture of the new LF driver on a new rev11.1  control board and there is no  light on the LF board  with the inverter  loaded  and connected  to the house .  The transformer  run more quiet  with the new LF board  so I  try to see what is different  .   The picture  show there is no  different  in  the  2  LF board  except the  price  and the transformer  is more quiet .   

@dickson I don't know what you're showing...at least as far as I can see in all of the above photos, you have 2 identical LF Drivers (both with the LEDs), on a 10.3C control board--not a 11.1 control board.

If there is a substantial sound difference (and power usage) in the inverter between the boards, then one of them is damaged.

Here's the "Sid LF Driver" design I sent PJ some time back--perfectly balanced setup (voltage / amperage on all 4 channels), using the TLP350 chips they are using.

Seems it worked fine up to about 6kw before blowing FETs--pretty much what a GS driver on a PJ mainboard does.  (In light of these issues, they are not using the design.)  I don't know how the PJ inverter design can be just bad enough to make it actually work...

<fileStore.core_Attachment>/monthly_2021_11/image.png.f5ab1226c03dc9d5422c91d0eb0a3b7a.png

<fileStore.core_Attachment>/monthly_2021_11/image.png.32e1a99e5b80940ea5ee47d1319d7b73.png

So, no new driver design.  Except for Sid's, which doesn't work on large inverters.

Sid, what about those electrolytic caps?  Since you didn't include them in your design, what was their purpose? Or are those little smd's in place of them?

Electrolytics are already on the PJ control board, so they were somewhat redundant.  Any FET driver requires ceramic caps as close as possible to the chip for the high amperage spikes required for FET driving; as the PJ design uses isolated transformer taps for the high-side power source, no large electrolytics are technically necessary.

PJ uses a TL431 and transistor voltage regulator to reduce the floating supplies down to ~18v for the FET high-side gate drive; I used a 78L15 voltage regulator to do the same far simpler.

The problems with the "upgraded LF Driver" and a GS LF Driver w/ a PJ mainboard...have to do with the actual mainboard layout (traces haphazardly routed with no consideration to the amperage spikes necessary), and the far-too-thin and terribly pinned ribbon cable.  Fixing those issues (plus a few other design tweaks) have gotten a 12kw GS inverter viable...with the same GS LF Driver.  (Currently awaiting another set of transformer load tests.)

I notice looking closer at the PJ LF driver...there are NOT any ceramic caps on ANY of the four drivers (2 TLP350s and 2 totem pole drivers).  That will significantly reduce the inrush current for driving.  Yes, the 2 electrolytics are fairly close to the chips (but far beyond the manufacturer specs), BUT electrolytic caps have a (comparatively) very high ESR, and can't deliver a huge dump of current.

I'm beginning to wonder whether this might be the trick that makes a PJ inverter half function...

 I don't know what you're showing...at least as far as I can see in all of the above photos, you have 2 identical LF Drivers (both with the LEDs), on a 10.3C control board--not a 11.1 control board.

Sorry I  post the  picture of the  rev 10.3 control board  before I  replace it with the rev 11.1  which did not  make the inverter run  better  .  I  try to  show  that the  30 dollar LF driver I  brought from  ebay  look  the same as the one I replace  but the LED  on the new LF diver  never light on but still  work .  For some unknown reason  the new LF driver with the LED not working  make the transformer  more quiet .  The LF driver with the working LED  also work but the transformer  is more  noisy .  I  do not  know if powerjack  will  send all  new LF driver with the LED not working  from now on  .    I  try to  make my  powerjack run better by trying different parts  but  can not make much improvement .

3 hours ago, Sid Genetry Solar said:

I'm beginning to wonder whether this might be the trick that makes a PJ inverter half function...

My main reason for asking the question.  To get you thinking about it.  Figured it must nag at you.

Your driver design is an improvement for the reasons you stated(better stability, cleaner fet driving, etc)  Yet something about the differences between the two designs is obviously why a pj blows above 6kw with yours but not with a pj designed driver.  Putting aside the poor pj mainboard and ribbon cable design, if you could pinpoint the culprit, then you could 'fix' the design to work on a high wattage pj (theoretically).  This would be a divergent path from your GS design of course, just for pjs.

I would think there would have to be a huge market for a drop-in lf driver board for v7/8 and v9/10 pj controls.  I would also think you could easily sell at a lower price than the current ebay price of $19.99 each.  Better yet, extend that thought to a drop-in universal control board (using that 'fixed' driver board) for all older pj inverters.  It must require no significant mods to the existing pj mainboard, no wifi board, no other significant changes.  No bells, whistles, talking error codes, blinking green lights, no charging function,etc,  I kinda like power saver mode, but if it needs to go, I'd survive. Any selling price you could manage under $100 would surely sell many units.

I realize you spend most of your time on your GS inverters, as you should.  Maybe in your spare time (hah!)  It would be a separate money maker once done that wouldn't require updates/revisions.  I wonder how many powerjack inverters have been sold in the US since 2013 or so.  Thousands probably.

Thanks for listening

Your driver design is an improvement for the reasons you stated(better stability, cleaner fet driving, etc)  Yet something about the differences between the two designs is obviously why a pj blows above 6kw with yours but not with a pj designed driver.

The picture of  my  PJ 12v  8kw  rev 11.1 control board I  buy on ebay  last month .  It  has SID  LF driver  and  works  good  no problem with FETs  .   My  15kw  rev 11.1 control board  has  the powerjack v 9.0  LF driver with the LED  purposely  not turn on  .  The  old  rev 10.3 control board  on my 8kw and 15kw is  still  working  when replaced with rev 11.1  control board .  

<a href="/monthly_2021_11/DSCF5975.JPG.d2cfa05cccf0aba35228dac0c41afcd7.JPG" class="ipsAttachLink ipsAttachLink_image"><img data-fileid="875" src="//forums.genetrysolar.com/applications/core/interface/js/spacer.png" data-src="/monthly_2021_11/DSCF5975.thumb.JPG.4c08b1256c0dd0e1b5f6875e18fff9f9.JPG" data-ratio="75" width="1000" class="ipsImage ipsImage_thumbnailed" alt="DSCF5975.JPG">

<a href="/monthly_2021_11/DSCF5976.JPG.bb06f682d61479814945ada806f5f9cf.JPG" class="ipsAttachLink ipsAttachLink_image"><img data-fileid="876" src="//forums.genetrysolar.com/applications/core/interface/js/spacer.png" data-src="/monthly_2021_11/DSCF5976.thumb.JPG.c9eb894f44fd1e432ef02ba4f020afc3.JPG" data-ratio="75" width="1000" class="ipsImage ipsImage_thumbnailed" alt="DSCF5976.JPG">

Well, the culprit IS the poor mainboard and ribbon cable design.  Everything else is basically Band-Aids around an existing problem.

If the FETs are driven cleanly:

• they don't get hot.  I have seen here where customers are mentioning PJ FET temps exceeding 130F under moderate loads (with fans on them!)  The FETs in the GS 12kw inverter at full tilt...barely break room temp (with a fan on them, mind you.)  On a GS inverter, the FETs will get gently warm after approx. half an hour of no load (no fans.)
• they are MUCH more durable due to spending very little time in transitionary states.
• much higher efficiency, as they're doing exactly and only exactly what they are intended to do.

If the FETs are driven weakly:

• they get quite hot under moderate loads (or even no load, for that matter!)
• they are far more prone to catastrophic failure.  Any switch is most vulnerable when transiting between "off" and "on" (as well as vice-versa).  Weakly driving the FETs lengthens this transition time--considerably increasing their most vulnerable time.  This in itself makes the inverter vulnerable to spikes/surges/dirty loads.
• reduced efficiency due to heat losses AND potential "transition deadshorting" (due to the longer transition time), where the FETs on either side of the H-bridge are partially on at the same time.  (Case in point: I was able to reduce the no-load current of a certain PJ inverter simply by increasing the dead time of the CPU's H-Bridge module.)

With a PJ cable/mainboard, it appears that a clean and firm driving methodology causes issues with induced spikes in the cable (and induced noise across the H-bridge channels due to the mainboard layout) >6kw.  Since said issues seem to be caused by the faster rise/fall times on the FET gates, it should seem that increasing the FET gate resistors to slow said signals down would work.

Problem is, I've already tried that (on earlier GS 12kw tests).  Yes, it DOES slow said signals down, nicely rounding off what was previously a switching "ring".  But then at higher loads the FET reverse transfer capacitance ("Miller capacitance") rears its ugly head, and the FETs deadshort themselves across your battery.  Basically, increasing the gate resistors then allows the FET to pull it's own gate up into "transition threshold" if the drain slew rate caused by the opposing FET is fast enough.

The PJ driver seems to solve this problem by using small gate resistors (sorta bypassing what Miller capacitance issues are left)--and violating the manufacturer datasheet requirements for the driver chips...

...by NOT putting the required ceramic capacitor anywhere on the board (emphasis mine).  If the power source is a (comparatively) high-ESR electrolytic cap, this will reduce the slew rate of the FET "on" signal--simply because the driver can't pull enough power to reach it's rated max amperage. 

In short: by keeping the gate resistors small, the driver can still "hold down" the FETs in off state.  But by omitting the otherwise required ceramic cap, the drivers' "pull up" current is seriously limited / slowed down.  This reduces the rise time of the FETs, which seems to help mitigate induced noise in the thin ribbon cable & mainboard.

Worth noting that the PJ design powers the control board off the low-side FET source leads.  That doesn't help either...a GS control board will blow itself up without a separate ground connection (due to all the noise on the too-thin ribbon cable).

At a bankrupting loss, yes.  I checked eBay, and the $19.99 price for a PJ LF Driver includes "free shipping." 

The problem is, there is no such thing as "Free shipping."  The buyer either pays for it separately, or the item cost is increased to cover said "Free" shipping.

You'll be doing good to ship even a small package via UPS for $16 (coincidentally, we've been burned each time we've tried shipping via USPS since November of 2020--we are not using USPS anymore).  Add eBay (and PayPal) fees.  Packaging to ship.  Supply chain logistics.  Board manufacturing.  Parts procurement.  SMD assembly.  Living costs.  With $3?

Heck, the driver chips themselves are $1/ea @ 100pcs pricing.

The PJ control board isn't a bad setup by itself.  While all the added "junk" between the CPU and the LF Driver board is an annoyance to handle from a firmware perspective ("over load protect board", etc)...the rest of the design is actually fairly decent.

If I may say so myself (not having a v1.4 control board to examine closely)...the v1.4 board is actually a decent start.  Unfortunately for PJ, all of their "improvements" to the board design in the time since, have only hindered forward progress.  But that's because they don't know how to change the CPU firmware--instead, modifying the PCB to try to address a certain (firmware) issue......

The more I look into it, the more I'm convinced that the FETs blowing in charge are a result of bad firmware (with a side of poor FET design not helping either).  Power Save Mode's complete nonfunctioning is also directly a result of bad firmware.  Extra circuitry on the control board "just for Power Save" doesn't help at all--actually in my revised firmware, that entire circuit and corresponding CPU input is ignored.

Since we at GS are no longer supporting PJ upgrades (due to the extreme headache of all the "variants" and changes), I am planning to provide PJ with the rewritten CPU firmware at some point here.  It'll benefit them, and I'm sure they'll like that things actually work (Power Save, ATS, charge, etc.)

For that matter, I DID sorta start on a one-size-fits-all revised control board design for PJ that gets rid of all the junk...and would give them room to grow.  Utilizing as many of their original parts as possible (supply chain), but meeting their needs as best as possible.  The headache that killed progress on THAT was when my balanced driver board was a bust...

...we'll see, it might yet happen, now that I'm aware of what is likely the root issue causing FET blowups with a properly balanced LF Driver.  But it'd be a design you'd have to purchase from PJ; I don't foresee us selling it at Genetry Solar.  Got more than enough on our plate as it is.

Well, the culprit IS the poor mainboard and ribbon cable design.  Everything else is Band-Aids on an existing problem

The  large  old  15kw mainboard  problem  is the  circuit design of the  copper trace on the circuit board . I  test for short on the mainboard  by removing all the  capacitors  and all the mosboard  after the FETs  blow up and the solder melted .   I  find the circuit trace inside the  mainboard was shorted .   I  fix the short and the  picture of the  mainboard  use as  storage  for 6 more  capacitor  added to the 6 capacitor on the mainboard  inside the  inverter .   The  red cable is connected to the  positive  battery input  and the  black cable  connected  to the negative  battery input   in parallel  with the battery .   Be sure to precharge  the 12  capacitors before  connecting  battery  or  it will sound  like a shotgun  blast .  This  really  help  with  inductive load  by  keeping the  output  ac voltage  constant  at 116 vac .    Before  when  inductive load switch on the  output ac voltage  drop to  106 vac  temporary .  

Well, the "Sid LF Driver" does appear to be machine assembled.  I haven't heard anything further from PJ since they discovered the new design driver was at fault for blown FETs in their bigger inverters.

Half wondering if it'd work better (with PJ inverters) if I put a 4.7-ohm resistor between the driver IC and the ceramic cap.  Runs the risk of the chip messing up and blowing all the FETs out, but....

...another option would be to use diodes and separate resistors on the driver board to slow the rise time (but not the fall time).