8k and 5k ups charging

[Mike T]

I have an older Powerjack 24v 5k ups and a brand new 12v 8k unit.  The 5k unit was something that I previously used on an intermittent project basis, using batteries with lead acid chemistry.  I just sold the 5k to a friend with a promise to help him get it working in a permanent installation to back up household freezers, using Lifepo4 batteries.  The 8k I just bought is going (permanently) into my 5th wheel where I also intend to use lifepo4 batteries.

Q1: What is the low voltage cutoff value for the powerjack inverters?  I understand that there is an alarm for low voltage, but I need to know what the cutoff value is because if it's too low and the system is unattended during an extended power outage, we may find ourselves in a situation where the batteries are drained so low the internal BMS's go open, and then we have a whole new can of worms to address.

Q2: I've observed charging voltage over the last 24 hours or so and I don't see the charger settling into the "float" voltage indicated in the manual.  So far, voltage seems to stay put in the high 28V range with the battery type rotary switch set to #2.  My understanding is that I need this to settle down to about mid 27v range to avoid long-term degradation of the lifepo4 batteries.

Thanks in advance!

[Sid Genetry Solar]

40 minutes ago, Mike T said:

Q1: What is the low voltage cutoff value for the powerjack inverters?  I understand that there is an alarm for low voltage, but I need to know what the cutoff value is because if it's too low and the system is unattended during an extended power outage, we may find ourselves in a situation where the batteries are drained so low the internal BMS's go open, and then we have a whole new can of worms to address.

Well, it's really hard to say.  Most PJ inverters have had 2 "voltage adjust" knobs on the front panel.  One's for battery voltage, and the other's for AC output voltage.

There's one critically crucial problem: the "battery voltage adjust" adjusts ALL the battery voltage setpoints!  Including under-voltage alarm and shutdown, over-voltage alarm and shutdown, as well as ALL charge setpoints.  All at once--and this is why it's impossible to say for certain "what the low voltage cutoff is."

There is no way to individually adjust any of the setpoints--apart from the 10-step knob.  (Quite comically, several of the higher charge voltage setpoints are above the overvoltage shutdown threshold!)

 

45 minutes ago, Mike T said:

Q2: I've observed charging voltage over the last 24 hours or so and I don't see the charger settling into the "float" voltage indicated in the manual.  So far, voltage seems to stay put in the high 28V range with the battery type rotary switch set to #2.  My understanding is that I need this to settle down to about mid 27v range to avoid long-term degradation of the lifepo4 batteries.

Let's just say...the less said about the PJ charge firmware, the better.

[Mike T]

Hi Sid -- thanks for that info -- the voltage trim may at least allow me to achieve a safe "float" voltage on the 8k which doesn't cause premature damage.  I have not tested charge voltage (or anything else) yet on the 8k -- I took it out of the box last night but that's as far as I got.   I suspect issues, however, based on the fact that the 5k is charging much higher than the manual suggests on all of the available settings.  Is there an internal pot or a simple modification I can do to add a trim pot for charge voltage on the 5k?  (I am not a "good" technician by any means, but I have a basic understanding of solid state electronics and am usually able to solder/desolder components without damaging anything...)

For low voltage shutdown, I've hatched an idea based on something I read today -- I'm going to simply send the center tap on the power switch through a $12 ebay low-voltage cutoff module on both machines.  I understand that the center tap on the power switch simply passes low-current battery voltage to either of the two switch poles (haven't tested this yet) so breaking that circuit via the relay in one of these cheap cutoff modules should be a fairly elegant way to solve the problem.  Digital setpoint adjustment on these modules should ensure accuracy.   If you see a problem with this approach, please advise!

 

[Sid Genetry Solar]

49 minutes ago, Mike T said:

I understand that the center tap on the power switch simply passes low-current battery voltage to either of the two switch poles (haven't tested this yet)

It's actually battery positive on the center tap of the switch.  Yes, +24 or +48v from the batteries.

Power switch literally provides or cuts power to the little forward converter that powers the logic on the control board.  It'll generally draw <50mA (though it will have a surge when turned on with charging up the power supply caps), so rather trivial to work with.

 

51 minutes ago, Mike T said:

Is there an internal pot or a simple modification I can do to add a trim pot for charge voltage on the 5k?

Depends what version internals it has--but the best you will be able to do is same as the 8k: add a pot to mess up the battery voltage feedback to the main MCU.

And just to remind you: that will adjust EVERY battery voltage setpoint.  Undervolt shutdown, overvolt shutdown, charge, etc.

[Mike T]

I'll get back to you when I can tell you what the control board is on the 5k -- maybe we can add a few components and dial down the voltage a bit.  As for it messing up overvoltage, I don't think I'll ever have an issue -- as for undervoltage, I THINK I have a solution...

Question on the split phase 8k: the input indicates "L, N, E" -- doesn't that suggest 110VAC input?  It seems that if it were 220VAC, labeling would indicate L1 and L2 instead of L and N. 

 

[Sid Genetry Solar]

1 hour ago, Mike T said:

Question on the split phase 8k: the input indicates "L, N, E" -- doesn't that suggest 110VAC input?  It seems that if it were 220VAC, labeling would indicate L1 and L2 instead of L and N. 

Yeah, but now you're trusting the labeling 🤪

[NotMario]

If L2 goes directly to case, the AC input will be 120.

I find 120v input strange without power-assist. Half passthru capacity...

[Mike T]

RE my brand new 8k split phase 220 LF UPS unit with "L,N,E" specified for AC input: 

I figured undervoltage on the AC input couldn't hurt anything so I initially hooked it up on 110vac, L,N,E, as indicated on the case.  I did this assuming it might not work and, if so, the next test would need to be on 220VAC input.  It so happens that the charge and UPS functions worked perfectly with a 110-V input -- Output was 220VAC between L1 and L2 outputs, 110VAC between L1/N and between L2/N, regardless of whether the unit was inverting from battery power or passing AC from the 110VAC input.  This tells me that the unit must incorporate a step up transformer -- which I did not expect.  Voltages cited here are approximate, of course, but within acceptable tolerance. 

This all works great for my RV installation, as the services at some RV camp sites will show 220V between L1 and L2, while a twin parallel 110V service is all that's required and is often all that's present.  With the inverter requiring only 110V, the charge function will work in all possible scenarios -- both 110v and 220v camp sites.  There is an issue regarding ups handling of parallel "L" circuits which may or may not sum to 220v, but I plan to use a DPDT 80A 110V relay I have sitting around, as a transfer switch of sorts.  This sketches out great on paper and should work fine in application.

I tested the inverter's charging function only briefly -- my bluetooth enabled LIFEPO4 battery never indicated anything past about 33A of charge (this is how I monitored charge rate).  Do you (anyone) know the charge amperage rating for the newest 8k Power Jack LF UPS inverters? -- or maybe I should ask what they are actually capable of (regardless of fake ratings)?

--greatly appreciating the input here!

[Sid Genetry Solar]

13 minutes ago, Mike T said:

This tells me that the unit must incorporate a step up transformer -- which I did not expect.

Autoformer principle of a center-tapped winding.  It's actually intrinsic to basic transformer design.  AC input Neutral would be directly wired to AC Output Neutral, while AC Input Line will go through the relay and (likely) to AC Output L1.  That energizes the entire transformer core...which causes output on the L2 winding.

 

15 minutes ago, Mike T said:

I tested the inverter's charging function only briefly -- my bluetooth enabled LIFEPO4 battery never indicated anything past about 33A of charge (this is how I monitored charge rate).  Do you (anyone) know the charge amperage rating for the newest 8k Power Jack LF UPS inverters? -- or maybe I should ask what they are actually capable of (regardless of fake ratings)?

Max charge amps....really up in the air.  Likely going to be limited by internal temps (in one way or another!)...whether that limit is above or below the software setpoint is anyone's guess!

What I'd be really curious to know is if you measured charge efficiency (i.e. watts in vs watts out).  I'm expecting it to be ~60% or less.  If it's ~50%, I'll get a good laugh.

[Mike T]

Since this thread has just been me airing my own challenges, I'll follow on with my new problem which is no doubt the fault of a major idiot: me. 

I installed the unit, threw the switch and BOOM!  Mosfets on the left side of the unit, with battery terminals facing me, are blown in spectacular fashion and this is the only visible damage -- I assume the driver board may be shot but there's no visible evidence to suggest it.  Mosfets on the opposite side appear fine but I haven't tested yet.

I did bench test the unit prior to installation so we have to assume my installation was amiss.   The installation, in total, includes grid power via 50A RV service (either 220 or twin parallel 110VAC circuits depending on the RV park), a factory installed backup generator and ATS,  a second ATS installed by me to switch between output of the first ATS and the new LFPSW unit  (I'm aware that this PJ8k won't replace a 50A RV service but it should be sufficient to our needs). 

The second ATS unit was installed to address the fact that this LFPSW 220 sp unit, despite it's internal ATS function, only accepts one leg of 110VAC for input, whereas my RV's 50A service either requires 220VAC OR twin parallel legs of 110VAC.   The external ATS was required to manage switching for both legs.  Despite ATS duty being handled externally, 110VAC was still required at the LFPSW unit's inputs so that it can charge the battery and intelligently switch between charge and invert modes.  (hope this made sense?)

So, with that understood, I suspect the cause of the spectacular failure, was that the neutral connected to the LFPSW's input N terminal has already been bonded to ground somewhere up stream (generator?  ATS???).  When the outputs of the LFPSW unit route through the second transfer switch to the breaker box, ground is bonded back to neutral and you end up with a situation where input neutral, output neutral and ground, as connected to the LFPSW unit, are all at the same electrical potential.  I don't know this for a fact, but I suspect this unit isn't designed to have the input and output neutrals at the same potential...

Am I correct -- is this my problem?

Also - is it likely that anything else was damaged inside the unit besides mosfets and driver board?  I'm in a hurry to get this thing sorted and if I'm going to need more parts, I need to get them ordered immediately!

[Sid Genetry Solar]

1 hour ago, Mike T said:

So, with that understood, I suspect the cause of the spectacular failure, was that the neutral connected to the LFPSW's input N terminal has already been bonded to ground somewhere up stream (generator?  ATS???).  When the outputs of the LFPSW unit route through the second transfer switch to the breaker box, ground is bonded back to neutral and you end up with a situation where input neutral, output neutral and ground, as connected to the LFPSW unit, are all at the same electrical potential.  I don't know this for a fact, but I suspect this unit isn't designed to have the input and output neutrals at the same potential...

There is no difference between "output neutral" and "input neutral" at all on a PJ inverter.

The "mains relay" literally only disconnects L1.  It doesn't disconnect anything else.  You can probably continuity check "input neutral" straight to "output neutral."  Unless some funny miswiring occurred inside and they're crossed--which wouldn't surprise me in the slightest.

This makes a "backfeed" highly possible, where if there's a power potential across L2 - N (for 240v input), the inverter cannot disconnect it.  However, if this PJ inverter has 120vAC single-phase input, that would rule that out.

 

Oh, and grounding should mean absolutely nothing either.

 

1 hour ago, Mike T said:

Also - is it likely that anything else was damaged inside the unit besides mosfets and driver board?  I'm in a hurry to get this thing sorted and if I'm going to need more parts, I need to get them ordered immediately!

That's always the hope, and generally the case.  Unfortunately, the PJ design is not 100% isolated at the driver board--and sometimes damage backfeeds up the line.  I've got a control board (from a customer) that has a hole in the main MCU...from driver board damage that fed back up the line.

[dickson]

I did bench test the unit prior to installation so we have to assume my installation was amiss. 

MY  rev 11.3  control board only has 2 LED  on when running  and  the LF driver  LED   and is off .    Do your bench test  show  2 LED light up  on rev 11.3 control board ?    IF only one LED or if LED is blinking   then the rev 11.3 control board  is bad  .      The rev 11.3  control board  is 100 to 150 dollars  from China .   

[Mike T]

13 hours ago, dickson said:

"MY  rev 11.3  control board only has 2 LED  on when running  and  the LF driver  LED   and is off .    Do your bench test  show  2 LED light up  on rev 11.3 control board ?    IF only one LED or if LED is blinking   then the rev 11.3 control board  is bad  .      The rev 11.3  control board  is 100 to 150 dollars  from China .   "
 

 

well, my  bench test was just a quick case of connecting everything up, input 110VAC and battery power, to be sure the unit worked out of the box.  It did, as far as I was able to ascertain, but failed instantly when hooking it up in a permanent installation.

Not sure which control board I have but I'll check shortly and reply back.  In general, however, can I do a bench test with 12VDC applied and the FET's out of the circuit to test the control board, observing LED's operation as proof of operation? 

Edited December 10, 2022 by Mike T
misquote

[dickson]

Not sure which control board I have but I'll check shortly and reply back

Powerjack only  sell the  rev 11.3 control board .        LED  on the  rev 11.3  control board  is  proof of operation .     The LF driver  LED is off   and 2  LED   is on .    The control board is bad if  LF driver  LED is ON     and if control board  LED is blinking fast   or only one LED is  ON  .     IT  will be difficult to repair  if your control board is NOT  rev 11.3   but    getting  a new complete  main board  of rev 11.3  will  work with the rev 11.3 control board  ribbon cable  .          

[dochubert]

Best advice on powerjack inverters;

Don't use the charging function!

Don't connect anything to the 'input' connector!

Despite it's limitations (lots of them!) the powerjack inverters will do a good job of running your stuff, assuming you stay within those limitations.  Just don't expect it to charge your batteries too.  Use an external charging source.  You'll have way less headaches.

 

[Paul]

Not to mention that the H-bridge charge function is incredibly inefficient. You can tell that just by the amount of heat the thing will blow out when in charge mode. I'd be suprised if the efficiency is over 60%.

I use a Powerswitch PSW2460 (full range specs at https://www.moriraddrizzatori.it/en/products/battery-charger/high-frequency/power-switch.html) when I want to charge my 24v bank of SLA GEL batteries from grid power. Quoted efficiency is > 85%.

[Sid Genetry Solar]

11 minutes ago, Paul said:

Not to mention that the H-bridge charge function is incredibly inefficient. You can tell that just by the amount of heat the thing will blow out when in charge mode. I'd be suprised if the efficiency is over 60%.

I would be surprised if a PJ inverter's charge efficiency is over 50%.

Because I've measured GS inverter charge at 58-60%...utilizing a stock PJ mainboard, though.  With either a GS-design mainboard or GS "LegacyFET" boards, the charge efficiency increases to ~72-75%.  I'm hoping to figure out how to get it to at least 80% at some point...

[Mike T]

For now I'll be using an outboard charger -- made that decision in the last week or so.

The GS FET boards are in and I'm installing now.

1.  I've got thermal paste but none that's specifically electrically conductive.  I suspect it's relatively dialectic in nature (the white stuff) but it's certainly not going to prohibit metal to metal contact in spots, not to mention that screw contact on the opposite side.  I feel like that's plenty of electrical conductivity for use with a dialectric thermal grease -- but please correct me if I'm wrong!

2.  More importantly, these GS FET boards have only 4 FET's per board with the outer two positions on the boards unpopulated.  That's 4 FET's on the GS boards vs 6 on the failed PJ FET boards.  I understand that you're using different FET's which perform differently (and theoretically better) than the originals; please confirm that these 4-FET boards are a satisfactory replacement for the 6-FET boards used in the latest 8k units.

Thanks!

[Sid Genetry Solar]

32 minutes ago, Mike T said:

1.  I've got thermal paste but none that's specifically electrically conductive.  I suspect it's relatively dialectic in nature (the white stuff) but it's certainly not going to prohibit metal to metal contact in spots, not to mention that screw contact on the opposite side.  I feel like that's plenty of electrical conductivity for use with a dialectric thermal grease -- but please correct me if I'm wrong!

Wouldn't make a difference, I don't think.

Thing is, the FETs generate so little heat in a proper GS design, that heatsink paste is kinda an unnecessary cost/complexity.

 

33 minutes ago, Mike T said:

2.  More importantly, these GS FET boards have only 4 FET's per board with the outer two positions on the boards unpopulated.  That's 4 FET's on the GS boards vs 6 on the failed PJ FET boards.  I understand that you're using different FET's which perform differently (and theoretically better) than the originals; please confirm that these 4-FET boards are a satisfactory replacement for the 6-FET boards used in the latest 8k units.

Umm...?????!?

I'm not aware of a single GS FET board yet released/sent to the field that should have just 4 FETs/board.  Can you provide a photo of the boards you have?  ...PJ's the only one that ever has done 4 FETs/board.

[Mike T]

I didn't wait for reply to above -- temporarily installed the FET boards without thermal grease and installed the new driver board.  I inspected everything else and couldn't visually spot any problems.  I connected the inverter unit to a car battery with a set of 6 ga jumper cables, as these were easily accessible to my work bench.   One LED on the main board illuminates and one led on the driver board illuminates.  This seems to be different than Dickson indicated in his post above.  It is, by the way, the version 11.3 board.

With a large shop vac as test load, the unit went into low-voltage protect mode, probably due to small power cables and battery.  With a small box fan as a test load, however, the inverter works just fine.  

Now that I see your reply, I'm a bit alarmed.  Photos are attached -- please advise! 
 

[Mike T]

1 hour ago, Sid Genetry Solar said:

Umm...?????!?

I'm not aware of a single GS FET board yet released/sent to the field that should have just 4 FETs/board.  Can you provide a photo of the boards you have?  ...PJ's the only one that ever has done 4 FETs/board.

Photos show 4 FET's -- I didn't notice this when ordering...

https://www.genetrysolar.com/parts/mosfet-board-for-48-volt-10k15k-power-jack-inverter-684zr-7d748

[Sid Genetry Solar]

22 minutes ago, Mike T said:

Photos show 4 FET's -- I didn't notice this when ordering...

https://www.genetrysolar.com/parts/mosfet-board-for-48-volt-10k15k-power-jack-inverter-684zr-7d748

Yeah, those are Power Jack spare parts from the GS website.

Proper (new) GS FET boards look considerably different.

[Sid Genetry Solar]

If you're working with a PJ inverter...heatsink compound won't hurt.  Things run so hot anyway.  I gooped up the FETs in my "9kw PJ" with thermal compound before putting it into service (tranny overheats at 3kw of course).

[dickson]

One LED on the main board illuminates and one led on the driver board illuminates.  This seems to be different than Dickson indicated in his post above.  It is, by the way, the version 11.3 board.

Your  mainboard  has the 4  mosboard from Sean  and  look like  the  4 mosboard  LED  is on green and  all   FETs  are good .      Some  LF driver  like  your  the LED  is on green and is working  good .      MY 

[dickson]

My  new  LF driver  LED is OFF  and is  ok  ,        IF the LED  is blinking then the  control board is bad .      Your  control board   may have 2  LED  on STEADY and should not be Blinking .         Look  like your inverter is working  good .        Your  inverter will work with  4 FETs  per mosboard  just as well as 6 FETs per mosboard .        I    do not  use  paste  on the FETs   .           A   12v dc  Delta  fan will  keep the FETs  cool .