PIP inverter repairs and hardware modifications

[weber]

MODERATOR NOTES
This topic is about repairs and modifications to the internal electronic hardware of any of the 48 V 5 kVA PIP, Axpert or equivalent inverter models. It was started by Walde, with this post and the title PIP Inverter Error "error 09" bus soft start failed. We later moved all the internal hardware related stuff from the PIP-4048MS and PIP-5048MS inverters topic to this one, and renamed it to be more generally about the internal electronics, but not the firmware, of all these related inverters.

There are other topics for aspects other than internal electronics, for the
PIP-4048MS and PIP-5048MS inverters (also called the Axpert MKS 5K), and for the
PIP-5048MK inverter (also called the Axpert King 5K), and for the
PIP-5048GE, PIP-5048GK and PIP-5048MG inverters (also called the Axpert VM II 5000-48, Axpert VM III 5000-48 and Axpert MKS II 5K), and for the
Voltronic InfiniSolar inverters.


Index

Hardware

• Rinaldo's post with the 2015 service manual
  Coulomb's disassembly and reassembly instructions for removing and reinstalling the main board.
  Weber's post about upgrading the capacitors and MOSFETs for greater high battery voltage tolerance (updated March 2017).
  •  Also deciding which MOSFETs to replace.
    February 2017 model has upgraded parts.
  Start of discussion about rewinding the large toroidal inductor.
  Fan replacement and direction
  • Weber begins discussion
    T1 Terry's post (see also Weber's post immediately after).
    Coulomb's post on the Arctic Cooling fans and their connections.
    2015 model Fan "Rectification", i.e. changing direction from blowing down to blowing up.
    Original fans put back, due to the fan locked warning issue.
    Comparison of temperatures with downward and upward fan direction.
    Video of fan replacement without taking the inverter off the wall (7 minute).
  July 2015 model photos from Rinaldoparaipan and Coulomb, and Fan "Rectification". More photos from Rinaldo.
  Repair
  • Coulomb's post of the PIP main board repair, with the power flow explanation. Also the Power Flow Topology and block diagram from a service manual. Repair continues. Repair completed with Replacement of the 230 V inverter IGBTs, this time with integral diodes.
    Testing inverter gate drivers when the battery-side MOSFETs are removed.
    Another repair and a few testing hints.
    Replacing the burden resistor fixes error 07 (fault code 07).
    Cactus's notes on resistors to check when replacing MOSFETs
  Partial schematic traces.
  • Coulomb's partial schematic trace of the main power supply circuit. Also the utility power supply (where fitted), and the bus soft start circuit.
    Partial schematic trace of the inverter IGBT gate drive circuit.
    Small trace of the fan locked rotor detect circuit.
    Weber and Coulomb's partial schematic for the battery-side MOSFET drivers and 400-V-side IGBT gate drivers for the main DC-DC converter
    Coulomb's partial schematic for the reverse protection circuit (where fitted).
    Partial schematic trace near the SCC comms connector, including sniff circuit used during dynamic current control patch development.
    Partial schematic trace of the Comms Board.
  Relay configuration one and relay configuration three.
  Photos of the SCC board and its pair of relays, and what the relays do.
  The red and black wires are NOT SCC output.
  Decoding SMD marking codes

Serial Communications boards

• Coulomb's post with the RJ-45 and D9 (RS232 serial port) pinouts 2013 model and immediately after
  Scott's post on the 2014 version.
  Weber confirms 2015 pinout, measures voltage sag on +12 V
  Scott's comparison of 4 (now 6) RS232 boards.
  Replacing your communications board with an Arduino without the microcontroller plugged in. Note safety precautions in the next post.
  McMajen able to re-flash again by replacing opto-isolators.

END MODERATOR NOTES
====================================


[Post by weber]

Any chance of a photo showing where the voltage sensing section is? Hopefully they use the same circuit lay out for their 24v units as well as these are far more practical for use in motorhome house battery set ups.

I note that I am now having second thoughts about the need to scale the voltage like this. But here's the schematic for the battery voltage sensing in the inverter/AC-charger section. It is a differential amplifier with a gain of around 1/40.



[Edit: See a photo of the strings of 1 megohm resistors on the main board here.]

Here are the two 100k resistors outlined in yellow (thanks to Coulomb yesterday). Coulomb carefully soldered a 1M 1% resistor on top of each of these, to give us the 10% change we wanted.



Here's the schematic for the battery voltage sensing in the MPPT section.



We hypothesise that the three differential amplifiers with different gains are there to give maximum A-to-D accuracy for the three different battery voltages this device can work with: 12, 24, 48 V. but what is mystifying is that the 3 gains are around 1/4, 1/7 and 1/13.

Interesting the 20w waste heat increase when converting a higher solar input voltage, something the MPPT supporters have been denying being the case, but simple logic said it must as there is so much more heat sink and/or forced cooling required in an MPPT controller compared to a PWM controller.

That extra 20 watts is only 0.6% of the 3200 watts going into the battery at the time. As Kurt said, a PWM charge controller can waste far more power by not collecting it in the first place. It is dissipated as heat in the panels (but any temperature rise it might cause to the panels is probably insignificant since 85% of what falls on them is turned to heat anyway).

[Moderator note: See this post in the PIP-4048MS and PIP-5048MS inverters topic, for why this modification was not pursued.]

[weber]

What's driving me nuts, and I think is likely to drive the customer nuts, is the fan noise. With only 300 watts being supplied it's unpleasant to be in the same room. When the water pump comes on and the load jumps up to a whopping 600 watts the fan instantly roars to even higher speed. The air temperature coming out is indistinguishable from ambient. It doesn't sound much worse when I turn on the 2400 watt kettle.

Can anyone tell me why inverter designers choose to make the fan(s) blow out of the bottom of the case, thereby opposing natural convection and failing to force air to flow fast close to the surface of the heatsink?

I have used Vantec Stealth fans to replace the noisy ones in Latronics inverters in the past, but their air flow (at 27 CFM) is too low compared to the fans in the PIP-4048MS. These are ADDA AD0812XB-A7FGP (80x80x25 mm, 12 V, 6.6 W) which I can't find the airflow specs for, but extrapolating from similar part numbers looks to be about 70 CFM. [Admitting defeat on the non-SI units here]

There's some discussion in another forum here http://www.photovoltaikforum.com/pv-inselanlagen-f57/pip4048-mpp-solar-improvements-t104258.html

[Edit: Deleted text "Latronics inverters are the same." They are not. They sensibly blow upward onto the heatsink.]

[offgridQLD]

So by the sound of it the fan has two speeds fast and very fast. What is the trigger for the 2nd speed temp or load? Sounds more like it's triggered by load.

The inverter is sitting inside a room in your house so that would make it sound worse.I can say with 100% confidence if my selectronic inverter was mounted inside my house even in the garage I would go nuts in one day. Buzzzz..grrrr...hmmmm of the big transformer and although the fan doesn't come on to often it's got a real sound to it like a industrial dust extractor. Lucky It's in it's own cinder block power room separate from the house by about 4 meters and at the end of the house from all the living areas.

If the air flowing from the pip is only ambient temp then it sure sounds like not much heat is being wicked away.Perhaps the fans are shifting to much air unnecessarily?

Kurt

[Johny]

I have noticed that any high airflow 80mm fan is noisy. My controller uses two 5.7 W 80x80x25 fans and I hear them easily when stationary.

[offgridQLD]

A question regarding the current shunt.

For some reason I have it in my mind that shunts should go on the negative cable.(don't ask my why though)

So that's how I have my main 500A shunt wired on my battery and the two other 200A shunts (one for each MPPT controller).

Is there any rule for this or really doesn't it matter?

Kurt

[weber]

I was mistaken. Latronics inverters are not the same. They sensibly blow upward and onto the heatsink.

Yes the PIP fans seem to be responding instantly to load, not temperature. And they seem to be running unnecessarily fast at very low inverter/charger power levels.

For some reason I have it in my mind that shunts should go on the negative cable.(don't ask my why though)
...
Is there any rule for this or really doesn't it matter?

Yes. The rule is, you put it wherever the manufacturer (of the thing that's monitoring the shunt) says.

They usually do go in the negative lead, because the device that is monitoring it is typically also measuring the battery voltage, treating the negative terminal as the common reference for both measurements.

In our case, our IMU needs to also measure the PV array voltage, not the battery voltage. It needs to do this so it can protect the PIP's MPPT from array-open-circuit-voltages greater than 145 V, which may occur at dawn on frosty winter mornings with the 3S x 72-cell PV arrangement I'm using. It will do this by leaving off the Minitactor that isolates the array, until the array warms up and its open circuit voltage drops below 145 V.

The MPPT in the PIP has it's positive in common with battery positive and switches the negative. That's typical for non-isolated buck converters.

So to measure both PV array voltage and battery current, the IMU needs to use the positive as the common reference, and therefore the shunt needs to be in the battery positive. We've modified the IMU to measure negative voltage.

[Edit: Corrected the PIP's max array input voltage (open circuit) from "150 V" to "145 V".]

[weber]

I have noticed that any high airflow 80mm fan is noisy. My controller uses two 5.7 W 80x80x25 fans and I hear them easily when stationary.

Yeah. A single 120 mm fan can be a lot quieter than two 80 mm fans pushing the same amount of air. But some fans are optimised for low noise and some are not, at the same diameter and flow rate. The ADDA fans in the PIP clearly are not.

But they are 4-wire (speed controlled by a PWM signal).

I ordered two of these Noctua NF-R8 PWM fans from PC Case Gear in Melbourne,
http://www.noctua.at/main.php?show=productview&products_id=45&lng=en&set=1,
perhaps foolishly, because they can only do 31 CFM, about half the flow rate of the existing ADDA fans. But I plan to make them blow inwards (upwards) in the hope that this will compensate for the lower flow rate. Clearly all this will violate any warranty I might have still had.

I had some more thoughts on possibly why the fans are currently blowing out (downward). When blowing upward, the hot air leaving the top of the heatsinks on the main board will blow onto the electros on the MPPT. So they may be correcting one design-flaw with another. Unfortunately I only thought of this after ordering the new fans. Perhaps I can add a shield for the electros, but that would only protect them from short term peaks in hot air.

But surely, as the fans are now, they will be sucking most of their air, not from the top of the heatsinks downward, but sideways from the openings around the battery busbars at the bottom, right beside the fans. Blowing is directional, sucking is not. I suppose I could block these holes with some foam.

Edit1: See also this post regarding the changes required to the fan plugs: http://forums.aeva.asn.au/viewtopic.php?title=pip4048ms-inverter&p=54866&t=4332#p54866

Edit2: I eventually had to restore the original fans because the Noctuas occasionally triggered a "fan locked" warning (code 01) even though they were working perfectly. See this post http://forums.aeva.asn.au/viewtopic.php?p=62186&t=4332#p62186

[offgridQLD]

I had the same thought about them not wanting to blow already heated air onto the MPPT charge controller up top.

Edit : Misread your description...hang on no I didn't. Are you sure the fans wont take air from inside the case (the fan is in a baffle/shroud) how can it not take air from the opposite side two what it blows....need to do some googling on that one.

Yes most of the air will come from the air holes on the sides up top and not passing much air over the heat sink on top of the unit. But as mentioned first perhaps that's still better than soaking the mppt charge controllers PCB in hot air.

Kurt

[weber]

Are you sure the fans wont take air from inside the case (the fan is in a baffle/shroud) how can it not take air from the opposite side two what it blows....need to do some googling on that one.

I didn't say the fans wouldn't take air from inside the case, i.e. the opposite side from the one they blow to.

When I wrote

"... they will be sucking most of their air, not from the top of the heatsinks downward, but sideways from the openings around the battery busbars at the bottom, right beside the fans. Blowing is directional, sucking is not."

I was only referring to the two heatsinks inside the case, not the MPPT heatsink whose fins project upward outside the top of the case. Its fins will not be cooled by the fans no matter what direction they blow.

Yes, the folded plastic sheet that makes a duct inside the case, enclosing the two heatsinks on three sides, is the only thing that makes any air pass near the heatsinks, given that they're on the suction side of the fans. But there are three holes in the plate the fans are mounted in, that allow the right-hand fan in particular to suck some outside air, and therefore to fail to suck that quantity of air past any heatsink. These holes are inside the duct/shroud/baffle.

Here's a photo, looking up from under the inverter, with purple arrows pointing to the holes I'm talking about. They would not be a problem if the fans blew upward, since blowing is directional.

[T1 Terry]

We replaced the fans in our PIP with maglev fans from Jaycar, much quieter. I seem to remember in one of the branches of the menu there was a fan speed setting, one option was constant slow run and a variable speed up/down reliant on the temp of the heat sink, that was the one we selected and it's been that way ever since. The owner is not in the area for a few weeks due to Christmas commitments but I can follow up on the menu setting when he returns if you like.
We do not have the MPPT controller model though, so there could be a major difference in the programme menu between the models

T1 Terry

[weber]

We replaced the fans in our PIP with maglev fans from Jaycar, much quieter. I seem to remember in one of the branches of the menu there was a fan speed setting, one option was constant slow run and a variable speed up/down reliant on the temp of the heat sink, that was the one we selected and it's been that way ever since.

Thanks Terry. Good to know someone else is getting away with replacing the fans with a quieter lower-air-flow model.

Did you leave them blowing down out of the case?

I assume you mean these:
http://www.jaycar.com.au/productView.asp?ID=YX2580
But if so, I'm surprised they do variable speed since they are only 3-wire fans. No PWM input.

There's definitely no fan setting mentioned in my PIP-4048MS manual.

My fans speed does not seem to respond to temperature at all. Only to current or power.

[offgridQLD]

Yes I can see now thanks for the pic (my pip4048 wasn't in front of me befor) Yes that baffle plate should be completely sealing off the two chambers.

Terry,
      I don't think there is a fan control option on the pip4048 we have. Having the fan thermostatically controlled would be the best option and variable speed that hopefully would keep the fans off at night, slow speed during the day and the odd fast speed during heavy loads and hot weather.

I don't have the software in front of me but isn't the temp reported in the pc software display just to keep tabs on any modifications.

edit: I have three of the same blue Jaycar fans listed above (though a larger diameter version) in my power room exhausting air from different areas of the industrial cabinet that my power electronics are housed. They are quite , smooth quality sounding but (not silent)like some of the pc fans designed for HTPC's yes mine are only 3 wire.

Kurt

[weber]

I don't have the software in front of me but isn't the temp reported in the pc software display just to keep tabs on any modifications.

No temperature is displayed on either the PIP's LCD or by the WatchPower PC software, however the protocol manual kindly supplied by Eric shows that heatsink temperature is returned as the 12th item in response to a QPIGS<crc-16><cr> command.

The present fan speed control algorithm is just ridiculous. When the water pump starts, the fans roar to maximum jet-engine noise just for the half a second of starting surge and then drop back again. It could scare the doodoo out of you if you weren't expecting it.

[T1 Terry]

They appear to alter the drive voltage but I guess that is still via a PWM circuit. I installed them to blow out the top, I just assumed that was the way it was originally designed but to be honest, it was only an assumption, I didn't really check which way the old ones went, they just died and we replaced them.
I also fitted a second set on the top of the case with a 50 deg thermostat on the top of the heat sink, but that was more because the air was getting trapped inside the cupboard where it was housed so the idea was to push the air out through a roof vent. This unit lives inside a cupboard in a Hino motorhome so noise would be a serious problem. The original roof top rattler Dometic air con threw in the towel and was replaced with an inverter split system unit that is so quite you need to touch it to determine it's actually running. Under full load the fans can now be heard, but it certainly isn't a disturbing noise by any means.

T1 Terry

[weber]

I installed them to blow out the top, I just assumed that was the way it was originally designed ...

As you would. That's great to know, Terry.

[offgridQLD]

That is mad. It's like they are worried it's going to go into instantaneous thermal runaway the second a heavy load applied.

So some way to intercept the fans power feed and control it based on temperature. There is some real smart looking case fan controller units getting around for pc's that don't cost much but what they can do and how they work. I don't know as I haven't had any experience with them but could something like this or similar be of any use
http://www.ebay.com/itm/NEW-Enlight-Combo-Spy-Smart-4X-Fan-Controller-with-Temperature-Sensor-LCD-Panel-/271707604678

pdf instructions
http://thirty-day.thirtyday.com/combospy/combospy.pdf

something like that could look smart flush mount in a box lid that's already there with some space or new small box.

Kurt

[T1 Terry]

When we first fitted it up the fan would start up like that, like it initially powered the fan and then looked to see how fast it should be running, sort of like when you state up a ceiling fan, full power and then set it back to the desired speed. The option with the fan constantly running slow seemed to eliminate the need for a sudden speed ramp up, but there wasn't an option of a fan stopped without that high speed start up. Maybe they had issues in the early days with fan motors burning out and determined it was the result of an under voltage start up.
You could always ask the factory, they are very helpful people, well I've always found them that way.

T1 Terry

[coulomb]

From a page that Weber linked to back on page 7:

"One has to cut and resolder the proprietary plug - color by color - though."

This was in relation to an Arctic Cooling fan, which happens to use the black red yellow blue wiring scheme, as on the Adda fans on the original PIP. But it seems that there are two, not quite consistent wire colour schemes for fans:



It so happens that the Noctua fans Weber bought use the other wire colour scheme: black yellow green blue. Note that yellow changes meaning! Besides that, the linked-to post above doesn't mention that "color by color" actually results in swapping the middle two pins!

It also happens that to gain access to the fans, you have to remove a clear plastic cover (duct), which requires removing a PCB, and to get access to the screws or the supplied anti-vibration mounts, you have to move the bottom plate of the inverter, which in our case required removing six thick AC wires. Because we didn't imagine that the fan wires would need to be swapped, this had to be done twice

However, the final result was miraculous. We had to look at the fan blades to know whether they were actually spinning or not. At normal 200-300 W loads, the fans are silent. With a 2.4 kW load, the fans run a little faster, but still barely audible.

But was the heatsink staying cool? Fortunately, the PIP inverter has the QPIGS command (Device general status parameters inquiry), and one of the many result returned is the inverter heatsink temperature. Even after boiling a kettle, the temperature increased from 32C to only 35C. (Ambient was 24C.) (The CRC for this command is B7 A9, which on TeraTerm can be typed with right-alt-7 right-alt-) (assuming that right-alt has been enabled as the meta-key*).

Hopefully, this information will help someone else not have to pull the inverter apart twice.

[ * Edit: This means editing TERATERM.INI in the Tera Term program folder, changing MetaKey=??? to MetaKey=right, and Meta8Bit=??? to Meta8Bit=raw . Restart Tera Term to take effect. Tera Term is a free terminal emulation program, allowing you to talk to an RS232 serial port. ]

[offgridQLD]

That's a great outcome. Thanks for sharing the details.

Our pip4048's are slowly becoming special adition versions.

Kurt.

[weber]

Replacement caps & fets? Do you guys have a service manual or something? So far there is a dead PIPs & Rpi here, casualties of the learning curve. The Pi is a throwaway but we'd be interested in PIPs resurrection.

A fellow who lives near both Kurt and I, had an earlier version of the PIP-4048 for about a year. He had the model where the MPPT was not built in, but was a separate product. He contacted Kurt (and Kurt invited me) to show and tell about the PIP blowups he'd had. The first two times, it was fixed under warranty. He had AGM (lead-acid) cells. The MPPT was not properly controlling the voltage near end-of-charge and was making excursions to 63 V, which is the rating of the PIP's DC bus caps. These caps also had only a short rated life at their maximum ratings -- 2000 hours which is about 3 months. And that's about how long they lasted. Towards the end of this life the caps' impedance would have risen to the point where they could no longer control the inductive voltage spikes from the PWM switching and some MOSFET(s) would blow. The MOSFETs only had a 75 V rating. Hardly any headroom to speak of. It's common to use double the maximum DC bus voltage.

So I researched suitable replacements with higher voltage ratings. It wasn't easy, particularly when limited to 18 mm diameter caps by the PCB layout. But here's what I settled on:

Caps: United Chemi-con EKZN800ELL 182MM40S, 1800uF 80V radial
Digikey part 565-4129-ND

MOSFETs: International Rectifier IRFB4310ZPBF, N-ch 100V 120 A TO-220AB
Digikey part: IRFB4310ZPBF-ND

The caps (4 of) have much lower capacitance than the original Jamicon WLR332M1JL44RT9 caps, which would be bad if it were not for the fact that they have nearly the same 100 kHz impedance.

Capacitance, voltage, ripple, impedance, endurance (at 100 kHz and 105°C)
Original          3300 uF, 63 V, 4.57 A, 15 mR,   2,000 h
Replacement 1800 uF, 80 V, 3.86 A, 17 mR, 10,000 h

Hopefully the 5 times longer endurance more than makes up for the 15% lower ripple current rating.

I note that the "Z" in the replacement MOSFET part number is important. There is a part without the "Z" that is not up to the job. The IRFB4310Z has almost identical specs to the original IRFB3307 except for being 100V instead of 75V. I replaced all 16 MOSFETs. Replaced the silicone thermal washers too, on the 8 that had them, and used a thin smear of heatsink grease on the ones that didn't need insulation from the heatsink. The originals were put on dry, which is bad.

Despite replacing the caps and MOSFETs I have not yet succeeded in fully repairing the PIP. I found some resistors open circuit in the MOSFET driver circuitry as well, and replaced them, but the blowup must have propagated further back up the driver chain.

As Kurt says, I gave his PIP a pre-emptive upgrade. For complicated reasons, I did not end up doing this to the PIP in Monolith #1, which has been running as an off-grid solar power system for 2 months now (PV array 3.5 kW 72 V nominal, Battery LiFePO4 180 Ah 48 V nominal).

Kris, it would be great to know how yours blew up, so we might avoid this, if it isn't too embarrassing.

[Edit: Added Digikey links. Aligned cap specs. Changed "caps (8 of)" to "caps (4 of)".]

[PlanB]

Very enlightening Dave. Will be really interesting to see how long the unmodified monolith pips holds up then. Cheap electros must be the Achilles heel of electronics. I've resurrected more dead LCD monitors these last 10 years than I care to remember. Don't even need trouble shooting skills, just replace all the 'pregnant' looking swollen caps. Boy I sure hope Nissan are using quality units in the Leaf!?
I don't really know much about the demise of the PIPs it just refused to work one day, not one I was personally involved with. I'll see if I can find out more. The Pi expired when I plugged in a HDMI monitor while both were on. HDMI is supposedly hot swapable but the Pi B model didn't take to it too kindly.

[gmacd33]

Thanks Weber, that's awesome! Sorry I thought it would be saved in a folder already, but thanks for the Googling work!

I'm working on a concept to use 3 of these for a 3-phase system, without solar, just with batteries - to do "peak lopping" to reduce capacity charges. So your comms control of the charging could come in useful, as it won't be reset by the MPPT all the time!

[solamahn]

I have been installing mppsolar pip inverter/charge controllers in Papua New Guinea for about 2 years. About 75 sites now. From 812HS to multiple 4048MS. Price attracted me to this brand. I have had my fair share of problems though. Fans. The soldered terminals would corrode. Depending on which terminals corrode, you get either a non serious 01 alarm but fan runs, serious fan does not run or failure if the 12v fan power supply fails. I treat the soldered fan terminals with lanolin grease and mount them upside down. I notice that late model units have the terminals coated. I would like to know the model number of a non adda replacement. DC cable terminals on scc were not tight enough. If using a lot of panels, these terminals would get too hot and fail. I always tighten all the DC connections inside the inverter before installation but again notice that on late model units all the screws are right from the factory. They are listening to me. I have had MOSFET failures. I read with interest a member saying he replaced his 4048 mosfets and caps with higher rated units. I will try this. I suspect the lvd at 42 might have something to do with this. I set 29 to 47 but have asked the manufacturer to have this value load dependent. I use Pb now but would like to try lifepo4. I am thinking of using 15 CALB 400ah instead of 16. Any thoughts on this. A 3248ms with 12 x 280w panels and 8 x 12v250ah AGM is a good combination. 2 3248's with 24 panels and 16 batteries is another popular choice.

[offgridQLD]

Welcome to the forum Solamahn

Sounds like you have some valuable experience with the PIP range of inverters to share.

I guess corrosion would go hand in hand with how humid the air is or perhaps how close to the sea the location is. Thanks for tip. If you look back through this thread you can see details about replacing the fans with much quieter (most likely better quality) PC case fans. Although improving the noise like you say if the terminal or power supply fail the upgraded fans wouldn't help.

The first thing I noticed when I Pulled the cover off the PIP4048 was the DC cables from the charge controller. Particularly how they where attach to the tall standoff pins to clear the main PCB. As you say tighten them all up befor use.

I'm not 100% sure how much better the new MOSFET used to replace my stock fetts are. From memory they were only slightly better specs. I think just the temp rating? (Perhaps Weber who replaced them for me would have the specs at hand and could clarify that)

The caps that were replaced had a much longer life from memory something like 10,000hrs at 105C VS 3000hrs at 105c and higher voltage rating to 85 volts VS the marginal 63 volts of the stock caps.

With the Lifepo4 cells 16 cells is the number most people use to replace a 48v Nominal PB bank. I run 16x400Ah cells this has the voltage around 53v throughout the entire night running the house and for the most part your working in the 53V - 56v range while charging them over the day. Much different to lead acid that could be working anywhere from (48v - 60V). I have my LVD on my inverter set at 48V. Lifepo4 batterys have a very stiff voltage under load. So the typical inverter loads running a house in the 100A range on a 400A 48v bank will result in very little voltage sag. Unlike lead acid.

Edit: I just went back and found the cap and Mosfet specs in the thread. So the fets had a much higher voltage rating 100v and the caps where 80v vs 63v and a much longer life at max rating as mentioned above.

Caps: United Chemi-con EKZN800ELL 182MM40S, 1800uF 80V radial
Digikey part 565-4129-ND

MOSFETs: International Rectifier IRFB4310ZPBF, N-ch 100V 120 A TO-220AB
Digikey part: IRFB4310ZPBF-ND

Kurt

[solamahn]

Thankyou. Pb poses a problem with program 29. I would like to set it to 48v but then I could get lvd in the morning if lower soc and high load and then have to wait till 52v for the inverter to automatically turn on again. If the setting in 29 also was adjusted for load like in program 05, that would be better. 48v for load less than 20%, 46v load between 20% and 50% and 44v for load greater than 50%. Like you say, less of a problem with lifepo4. I set 26 to 57 and 27 to 54.8 with good results, although some sites could be better with different settings. Using 15 lifepo4 would get around the prog 29 48v maximum. With 15 lifepo4, could set 29 to 46.5, 26 to 53.5 and 27 to 49.5 and fine tune from there. Do you see any problem with that setup. I had another 4048 blow up yesterday. Mosfets Q14, 25, 26, 40, 23, 18, 13 and 19 all cooked. Some had their legs blown off. Not sure what caused it. I suspect low battery voltage. LVD was at manufacturer setting of 42, I think. I dont have any problems with 4048's using mains backup 12 set to 48, 13 to 52 and 11 to 10A or automatic genset start. Am hoping that setting 29 helps reduce failures also. Would be good to stop all failures with either modifications or programming.