MultiPlus 24/1200 DIY repair (EDIT: unknown charging cutoff issue since unboxing brand new unit)

[slay3rone]

Hi. I have a very specific question, I’m really hoping for some help here. I have an issue with my MultiPlus 24/1200. Let me explain.

First of all, it works perfectly fine with no issue whatsoever running in inverter mode. But when charging, this is how it behaves:

• When I flick it on for the first time on a cold-ish day (winter on its way out here), connected to the mains for charging at the full 25A it can deliver, it works totally fine as far as I can tell. After 2 and a half minutes (or was it 3min) the internal fan kicks on, and it just keeps on charging no problem.
• But if the MultiPlus was already operating a little bit (enough to warm up I suppose), then I enable charging, after the 2 and a half minutes delay, the moment the internal fan kicks on it immediately shuts down the AC IN followed by the internal fan, then proceeds to resynchronise with the mains, and starts another charge cycle (I have to stop it, or it will overheat otherwise after multiple cycles like that). I get no alarm whatsoever plugged in to VE.Bus, tried VEConfig and Victron Connect. I just cuts off, and does not tell me why at all.
• After the first cut off, nothing is too hot inside the MultiPlus. In fact, I tried having a fan blowing air at the internals of the MultiPlus during the first 2 and half minutes delay, everything remains quite cool inside, but the problem keeps happenning no matter what.
• Now, if I run the AC in the room at 16°C, and have a fan blowing air full blast on the MultiPlus internals to really cool it down, then I begin another charging cycle, it works flawlessly. So it clearly seems temperature related, something not sensing temperature quite right.
• Note that UPS mode is also completely useless. When the cut off happens, it takes quite a few ms for the inverter to take over, and it shuts down a computer connected to it… So that’s really bad.

I think I have a pretty good idea what might be the culprit here, and if it is indeed this component, it is totally my fault I’m sure I probably broke it… That’s quite dumb. And that’s where I need some help figuring out the component model to replace it. More on that below.
EDIT: I was wrong, I can confirm nothing was damaged by any of my mods. That issue was present since day one, on unboxing. I just didn’t realise it (see second post below).

So, my MultiPlus is not exactly stock anymore… and definitely not under warranty at all after what I did (through it’s brand new, I just bought it a couple months ago).

(PS: That switch is in fact a dual pole AC breaker for AC IN. When off, the IEC connector can be used solely to earth the MultiPlus and nothing else.)

So yes, this is the MultiPlus in question. Quite modded indeed… Perfectly functional except for what seems to be this one thermistor below, R73.
EDIT: Thermistor is tested good (see second post below)

The reason why I think it’s this one is. Well, it’s the only single part that I had to fight to remove when taking the original MultiPlus apart. It was glued with some rock solid thermal adhesive to something that absolutely had to come off, or I would not have been able to proceed with the teardown. That thermistor was glued solid, and would not move under any circumstances, I had to improvise and it really felt awful when it finally came off, I was scared I might have broken it. I then tested the MultiPlus and it was performing seemingly totally fine, so I kept going, thinking it was all fine. Maybe not after all… after further testing later down the line, revealing this issue.

So, considering the “R73” naming, it is clearly a thermistor, I read that the MultiPlus has all its internal thermistors in series, or something. I suppose if one has a fault, it could definitely cause global issues. I’m surprised the inverter mode works flawlessly, cold or warm, and even charging works fine with the MultiPlus is completely cold. But it is what it is.

So any advice on replacing this R73 thermistor? Considering it was glued to the plastic housing of the DC input terminals, I assume its purpose was to monitor any bad connexions that might cause the terminals to heat up.

Technically, I don’t use the terminals anymore. I kept them installed to act as physical supports, to keep the positive and negative posts at the original spacing, to avoid shorts if anything was to ever move. Monitoring terminals temperature here don’t make sense in my setup, so I could either replace the temp sensor by a fixed resistor value matching a 25°C temperature reported by the original thermistor, or I could use extended wires and monitor the temperature of the QS8 connector I’m using, further away.

Either way. I would really appreciate if anyone here could give me some guidance regarding the model number or the specifications of a replacement termistor, so I can give a shot at replacing it, hoping this fixes my issue here.

Thank you very much!

---

Now, more infos on the MultiPlus mods, for those interested. This is completely unrelated to the issue above, as what is documented below was done much later, after noticing the issue.

The two vents at the front behind the handle, and the two fans behind them are now, since recently, controlled by a small internal DC to DC converter, switched on by an assistant via the MultiPlus relay.

There is also a temperature switch I added on the MOSFET heatsinks, wired to the temp sensor input of the MultiPlus. This is what triggers the assistants.

It kicks the two fans on at around 45°C MOSFETs temp, and kicks off at around 37°C or so, independantly from the original MultiPlus fan. So under low loads, it is possible to have only the stock MultiPlus fan running automatically, and under high loads, all fans can be running to ensure better cooling. Also, if the built in MultiPlus fan stops running, but the MOSFETs are still a bit hot, as long as the MultiPlus is still powered up, the extra fans will keep running until the MOSFETs reach below 37°C.

I custom designed an anodised aluminium heatsink that thermally joins all MOSFETs together. It is electrically fully insulated with proper high quality thermal pads from the output MOSFETs, but electrically tied to the input battery rail. That works perfectly fine, and lowered the MOSFETs temperature quite a bit, plus it works very well with the extra two fans I added.

The temperature sensor is built into this heatsink, so it monitors the temperature of any of the MOSFETs, and controls the fans accordingly.

---

Second thing. The VE.Bus MK3-USB module is modular, and can be removed if needed for some other project. The RJ45 cable can be easily disconnected from it, and connected to the direct RJ45 output port next to the USB B port. That way direct connection to a Cerbo or Ekrano can be done, if using this MultiPlus in a stationary configuration some day.

---

Third thing. Yes, despite the Milwaukee Packout Compact box being plastic, compared to the original MultiPlus bottom metal frame, I retained a full PE plane under the entirety of the electronics. There is a full length copper clad board (isolated with full kapton tape cover) that is connected to the PE connection on the MultiPlus PCB, and has the earthing wires tied to it as well.

---

This is the complete power pack, with the custom Packout MultiPlus at the bottom, and a 1kWh Li-Ion battery at the top.

The battery has the following specs:

• 6S16P (96 cells) Samsung 33J NMC cells - total of 1kWh - 10A per-cell individual fuses (cells I had already at my disposal, or I would have used better cells… up to 2kWh can fit in the same space, in there, for example Reliance RS50 or EVE 50PL tabless cells for the best cells on the market at this time)
• 100Balance BMS 60A continuous / 90A peak
• 50A Blue Sea COTS toggle breaker (for the rear ports, and the lid accessories)
• Internal 80A MIDI fuse, for the always on QS8 port
• Modded Victron BMV-712 (see below), with DIY shunt
• Victron SmartSolar MPPT 100/20 with fused XT60 solar input (can be used with a DC source), and XT60 load output (and a toggle to switch it, with status LED)
• 2x (for now only 1x wired up) USB C bidirectional 140W charger removable modules, it can be used to either charge a device at up to 140W (28V 5A), or charge the battery itself from a 140W wall charger. Two USB C ports, one bidirectional, one unidirectional shared with a USB A port. Max 280W combined output.
• The rear has 3x XT60 and 3x XT90, to plug in accessories. Those don’t have anti-spark though, but it should be fine for most applications. I made some anti spark small extensions when really needed.

Here is the pack charge and discharge plotted curve. Charging stops at 4.1V on these cells instead of 4.2V, and discharging cuts off at 3.0V (3.1V conservative here), as per the datasheet, for high cycle life.

Here is a peek at the modded BMV-712 (that works perfectly fine).

Here is a peek at the rear connectors panel.

Here is a peek at the USB C charger module, will add a second one later, right next to it.

I later will build a Venus OS module as well, with a Raspberry Pi Zero 2W I assume, and wire everything to it. Cerbo itself is too big I think. I won’t be using it all the time, as it will consume power, and is not really needed for my application here.

I wish the MultiPlus could be connected to the same VE.Smart Network as the BMV-712 and the MPPT, by using the Bluetooth VE.Bus donble, to track battery voltage precisely and adjust its charging accordingly. It would have been the absolute perfect minimal setup for this power pack here. But this is unfortunately not possible at this time at least…

---

Right. Don’t ask me why I did this, it makes technically no sense, and was way way more overpriced compared to something like a brand name power station, for less power at the same volume. But I had a bunch of Samsung 33J cells laying around, and I was looking for something to do with them, and really love Victron stuff, so I went for this!

Those Samsung 33J turned out to be a bad choice for this project here, so maybe later I will upgrade. The bottom of the cell is aluminium, so it can’t be spot welded with standard equipment (it requires laser welding). So, this is the reason why I had to use spring loaded cell holders. The cells themselves don’t have the lowest internal resistance, plus the holders seemingly adds a bit of ressistance as well. So in the end, with the MultiPlus I get quite a high Vdc ripple. It it still in the acceptable range, as it works flawlessly while inverting, but it is far from ideal, and will need addressing later. Maybe upgrading to a proper spot welded Reliance RS50 pack, and high amperage BMS.

Overall, I really like how it all turned out so far! But sadly, I was not expecting the trouble with the MultiPlus, and the batteries as well to some extent. Now just hoping to fix the MultiPlus so I can have something that fully functions, and can be used as a very solid UPS for example.

---

Fun side thing.

Here is the MultiPlus running my bedroom AC off a handheld slightly modified JNS S34 powerbank (I added a XT60 direct IN/OUT port, connected to the shunt and internal BMS MOSFETs). It won’t last long with its 100Wh, but it sure powers the whole AC unit at full 610W DC (24.88A 24.48V here). Note it’s a recent model AC unit, with smooth ramp up of the compressor, so no inrush at all.

This powerbank is DIY, and I mounted 7S Molicel P45B cells (not soldered, clamped copper contacts).

---

Anyways, thank you all if you read this far, appreciate your time and interest for my projects! Hoping someone out here can help with my issue.

[slay3rone]

An update on that situation.

I went ahead and removed the PTC thermistor, it measured 64Ohm and looked to behave like a working PTC should when heating it up, with its resistance rising noticeably.

I proceeded to compare its value with the transformer built-in PTC, and that one measured 54Ohms. So very similar values.

I still went ahead and did the short circuit test as mentionned down here (not with the transformer PTC of course, but the other one I removed), then ran a charge cycle.

First time it ran fine, but that’s how it used to be already. Unfortunately after manually stopping charging, and restarting after a few minutes (letting things cool down a bit), it cut off the moment the internal fan started spinning. Just like before. So this particular thermistor was probably totally alright after all, and the issue I’m having has nothing to do with that thermistor. Back to square one.

To add some precisions. When I first had the MultiPlus, I made sure it was working fine before taking it apart.

Inverter mode worked fine, I tested the UPS mode as well shortly, and I noticed it cut off… That was literally on the day I unboxed the MultiPlus, weeks before taking anything apart. I unfortunately don’t have any photos or videos of that. For my own sanity, I checked a chat with a friend where I told him I had issues with the MultiPlus charging, on that day I received it. So I can be certain I was not remembering that wrong.

At the time I thought, well I have a pretty bad wiring, pushing 25A through rather thin and long wiring during those tests. Plus a bit high internal resistance of the battery itself (I was using brass bus bars at the time, then rebuilt later using twice as thick copper). That was showing upwards of 0.7Vdc ripple or more, so I was convinced the battery and wiring had to be the issue, and definitely not a brand new MultiPlus. So I thought I would deal with that later when I rebuild my battery, as I was convinced it must have been the culprit. Turns out it was the MultiPlus itself.

I definitely should have done more testing on day one, as this is a bit of a sneaky fault, and return the MultiPlus under warranty when it was still under warranty as a DOA unit, and not completely modified. Now it’s too late, I have to figure out how to fix it myself.

So, well, I have no idea what could possibly cause charging to cut off after the MultiPlus has slightly warmed up. I unfortunately can’t see any alarm.

This testing right now is done on a proper 8S LF280K pack (I got cells a few days ago), with a JK 200A BMS, Vdc ripple is now around 0.1 to 0.2V, so the issue definitely does not come from that either, as I suspected on day one.

[slay3rone]

I just ran another test with the AC at 16°C and a big Vornado fan blowing air right at the MultiPlus electronics, and it started charging just fine.

As long as it starts charging correctly and the internal fan kicks on, then I can remove any extra cooling and shut down the AC, it will keep on charging just fine (and not ever overheat, of course, there is enough cooling nothing heats up too much here).

So it definitely is something temperature related that triggers at a specific moment (few minutes after charging starts) and only when charging, not at all when inverting. Maybe another PTC that is indeed faulty somewhere else, or some other temperature sensor I don’t know about.

That’s weird, and a very annoying issue. Every time I feel like I finally found the cause, nope.

[slay3rone]

Update, I checked the values of all the PTCs I could find, not long after a charging cutoff (after it charged at 25A for 3min):

• Transformer PTC (J10 PTC) → around 50 Ohms
• Input terminals PTC (R73) → was around 65 Ohms
• Battery input MOSFETs heatsink PTC (PTC to U3) → around 80 Ohms
• One side of the transformer drive MOSFETs heatsink PTC (PTC to U5) → around 85 Ohms

Then there is a PTC in 2-pin TO-92 form factor (R176, MOD-PTC?), it’s on the other side of the same heatsink as “PTC to U5”. Not sure what that’s for but seems to be part of a separate circuit, this one measures around 1.56k after cutoff.

All the regular PTCs seem to be more or less in the same range, not a single one reads a very high abnormal value, so I would assume those are all fine. I have no point of comparison for the other one (R176), but I can at least say blowing air on it makes it lower in value in the 1.42k range, so at least it’s working as it should it seems.

Nothing so far then. I have to keep investigating somehow.

[slay3rone]

One more update. This time with some really weird testing I did with what I had at hand, without taking the whole MultiPlus apart. But I think that helped narrow things down / confirm some stuff.

So earlier, I noticed that if I blow cold air on the whole MultiPlus electronics to cool everything down to around 20-25°C, then I started charging again it would work. But I was not sure what exactly was causing that, it if was a temperature sensor on the heatsinks, or some other component somewhere else that I missed. Blowing air cools everything down evenly, and it’s difficult to narrow down where the issue comes from.

Well, I happen to have a pretty powerful 165W peltier cooling stack I built for some random project some time ago, so I thought why not use this to only cool down the MOSFETs heatsink inside the MultiPlus, as fast as possible, without circulating air anywhere else and avoid cooling extra stuff around as much as possible. So I did that.

After charging for a few min and shutting down the MultiPlus manually, the heatsink was around 45°C. All fans were shut down completely, everything unplugged for safety, and I let the peltier bring the heatsink only down to around 25°C (takes 10min or so). So at this point any other component in the MultiPlus that is not directly attached to the heatsink remains warm. Then after firing up the MultiPlus again, it charges flawlessly. I retried multiple times, and every single time it worked totally fine.

So whatever causes the issue is definitely a warm component on the heatsink itself, I assume. The MOSFETs are definitely not the culprit… so there are the 3 temperature sensors:

• PTC to U3
• PTC to U5
• R176 (MOD-PTC?)

Weirdly, I already checked the value of those before and it all seemed rather normal, but measuring without taking things apart was not easy, and some values might have been off a bit. I will have a closer look and try to understand this a bit better.

[slay3rone]

Well, I found out some interesting trick that kinda solves my problem. If I disconnect for a couple seconds and reconnect one of the PTCs of the series PTC chain (for example the transformer PTC which is easy to disconnect as it’s just a connector), as expected it will trigger an overtemperature alarm. The internal fan will proceed to immediately ramp up to 100%. So far nothing new.

Since the PTC has been quickly reconnected, the alarm will then go out, and the fan begins to slowly ramp down but does NOT stop. It just slows down to a slow speed and remains like that for what looks to be 17min (not sure if that’s fixed or depends on something).

The thing is, if I start charging, the fan then proceeds to ramp up immediately, following the charge current! Unlike the original behavior, where there was a good 2 and a half minute delay before the fan even started! Enough time for everything to overheat (though that does not seem normal at all).

With the fan running from the beginning, I can charge at the full 25A, stop charging, restart, it seems to behave and not cut off at all, and MOSFETs temperature remain controlled, nothing overheats.

Disconnecting and reconnecting the PTC fast enough does not shut the inverter down seemingly. Charging does “pause” current goes down to zero immediately, but unlike a complete cutoff that disconnected AC IN, breaking UPS mode, this keeps AC IN connected, and charging resumes very shortly after the PTC is reconnected.

So I’m not saying I will disconnect and reconnect the PTC to solve my issue here, but this was an interesting observation.

Now, I really don’t like those fan startup delays and the fact that fan speed seems only tied to current and not temperature. So now that I know my problem is completely solved with more reactive fan control, I will just design a small circuit with a microcontroller, attach extra analog temperature sensors to all 3 heatsinks, each of them giving separate temperature readings. I will rewire the internal MultiPlus fan and my two additional fans to this circuit, and have all fans ramp up according to the MultiPlus temperature (highest of the 3 reported temps from my sensors), and ramp down to a stop only after everything has cooled down. That should completely fix the issue I’m having with my MultiPlus. I will then see if I can make that visible somehow in Venus OS as well.

All that would not be necessary if there was a more agressive fan profile available in the assistants. Just like the quiet fan profile that already exists. Something that dictates the fan to start up immediately, and not after a delay. It’s unfortunate.