nicknorman

Possible yes, but not normally unless very high usage. I’d say more like 10 years between repacking as an average with moderate use. I think you over-analysed my thinking a bit! If you have 2 pack blacking as we do, the boat comes out every 5 or 6 years. But my point was really that repacking is a trivial expense unless you also have to get the boat out of the water to do it.

Yes. Be aware that the packing comes in different sizes and you can get graphite-loaded or normal. Check the position of the pusher/stud length protruding and make a judgement call as to whether to repack. Since it is much easier to do this out of the water, and cheap, I’d suggest doing it anyway if it’s more than about 1/2 worn.

No I don’t think it was rubbish, the point about the castellated nut (or its absence) was cogent. I think the point is to try to avoid the nut being able to rotate a bit, which could cause fretting especially in an application where the direction of rotation / load reverses. Whether the tangs are bent circumferentially or radially is not significant since the bending is only to stop the pin from sliding out and there is no significant force trying to make that happen - in fact the offset mass of the pin is only going to pull it in when rotated fast. But locking things subject to heavy load only by means of a split pin isn’t great design, the primary means should be friction and bolt stretching, which admittedly is tricky with a fairly short “bolt” and dissimilar metals (with different thermal expansion coefficients). But fortunately the tapered shaft adds plenty of friction!

Quite possibly, hard to see from the photo. But my point wasn’t about the micro detail of that particular pin, more about the principle of the orientation of the pin. I didn’t make that clear, but it was the premise of the question. As you say, of course the pin should be the right size (snug fit) for the hole, and spilt pins shouldn’t be reused. All that said, it has obviously had some use and hasn’t come adrift so it is probably adequate for the purpose.

Speaking as a licensed aircraft engineer, I would say that there is nothing wrong with the way the split pin is inserted. An alternative way is to have the folded head of the pin rotated 90degrees and then one leg of the pin is bent back over the end of the shaft, the other has to be cut right down and is poked back towards the prop. But the bending is only to stop the pin sliding out. With the latter method the head does tend to sit in more neatly into the nut slot, but in terms of security it makes no odds. As mentioned it’s a safety locking thing, not the primary means of avoiding the nut undoing - the latter being achieved by tightening the nut!

Yes the 30mV was the difference between the highest 3P group and the lowest 3P group. As you know the curve of voltage against SoC is very steep around 3.55 to 3.6v so this represents very little charge. The lost capacity is really either of the things you mentioned and would manifest itself at the bottom end when the 3 cells would get to minimum voltage whilst the 4th still had a bit of usable. So only relevant if you cycle between 100% and zero%. I estimated it on how much charge had to be taken out of the high group by the balancing resistor. It’s my own design of balancing, devised before I realised that cells don’t go much out of balance and there is no need for high currents. But also it doesn’t require the cells to be held at high voltage whilst the balancing takes place. When the first cell hits 3.6v it looks at the differences between the lowest cell group and the other 3 cell groups and calculates an estimated time to turn on each balancing resistor even though charging has gone to float. It’s not perfect but closes the gaps at each iteration.

Just dragging up this old thread to mention that we are back down to the boat today and I thought I would check the top balance on my cells since I don’t think the balancing has kicked in for a couple of years. I took them to 100%, no balancing kicked in. But there was a bit of a voltage split (30mV or so max) and so I increased the charge voltage from 14.25 to 14.35. This pushed one cell up to 3.60v at which point the balancing kicked in. I have external balancing resistors that sink an amp or two. Not long afterwards the cells were balanced again and I would estimate that the out of balance was no more than 1 Ah. Which out of 600 Ah reflects a potential (and temporary) lost capacity of 0.17%. Not a big deal and in fact totally inconsequential. Bottom line is that 30mV right at the top (>3.55v) represents a very small imbalance in terms of Ah, and correspondingly only a tiny and not real-world detectable and reversible loss of capacity.

Alternative viewpoints are always respected! Obviously it depends on the application but the PICs have a much better selection of peripherals for low-power controller applications and are cheaper. I would agree that if you want to do graphics then the PIC is probably not the thing, but for small control applications I think it is. The Teensy 4.1 is very powerful in terms of processing power but has a price and power consumption to match and is more in the RPi SBC league than a microcontroller. I also vaguely remember assembly language but PICs these days are programmed by most people in C. I seriously dislike the arduino "sketch" thing and the abomination of C++ that is the default compiler, and the default IDE is atrocious (although other options are available). JLCPCB has been pretty good recently, my last order was sent on 21/5/25 and arrived 28/5/25. A recent oopsie with RS was to order a 3d accelerometer chip. Only after it arrived did I notice the dimensions on the foot print - 10 pads even though the overall dimensions were 2mm by 2mm. The pads were 0.2mm wide with 0.2mm gap between. Try doing that with black tape! But no probs for JLCPCB and a touch of solder paste, a steady hand and the hot air rework gun and it was done and functional.

Can I just mention JLCPCB in Hong Kong for bespoke printed circuit boards? There is no need for breadboard or rats nest things when you can get a top quality pcb with unlimited plated through vias, silk screen and solder mask top and bottom, for peanuts - literally £3.50 for 5 boards double sided or about £7.50 for five 4 layer boards, including shipping and import taxes. This is the latest board I sent away from, a 4 layer board for a PIC microcontroller thingy with a bluetooth module. And personally I would always use a PIC and not an Arduino, because the latter are quite limited.

Ours is a Clesse. Been OK for 10 years of leisure use. I see BES sell them, so they are probably of adequate quality.

Consider yourself mocked. Regarding JonathanA's point, it is a fair one but set against that, many people's inverters including ours are left on 24/7 when quite likely there is no-one around to notice the smoke coming out of it, let alone operate an isolator switch.

You can have 2 shunts in series. Each shunt drops a max of 50mV at it’s rated current, so with 2 you are only losing 0.1v and that only at the full rated current of the shunt. However I would argue against putting a 100A shunt in a circuit that can pass 200A or more. It would likely overheat. And bear in mind power dissipation is I^2R, ie double the current = 4 times the heat generated.

I just used the bus bars supplied with the cells. Although they are solid copper (with a huml), not several strips of thinner copper. But the supplier didn’t provide enough strips for my configuration so I made some more from copper sheet which was quite thin so I used 3 or4 layers. I did wonder whether several layers was a good idea in terms of electrical conductivity, but if this is the new gold standard then I guess it’s ok.

I’m not clear what you mean by flexible busbars. My interconnects (bus bars if you like) are copper strips with a hump - the hump being there to allow a small amount of movement. Since the problem has never recurred in that or any other cell (and we are about 3 years further on now) I don’t think it was an installation issue, more like a manufacturing issue.

I’m not sure what the point of putting stuff between the cells is. I certainly didn’t. I have a wooden battery box and my only aim was to ensure the cells couldn’t move, so there was a bit of packing out the box with thin ply, but just to take up a bit of clearance.

No. Lots of people get arrested on suspicion etc. But they were released, and so far have not been charged with any crime never mind found guilty of it. Only if the issues relates to their work.

Well BSS does not require a battery isolator for the connection to an inverter. I think it used to, which is perhaps where the confusion arises. A fuse is required. I wouldn’t bother with an isolator, it’s just another thing to drop voltage and get hot under heavy load. If you want to work on the inverter, remove it etc, you can undo the megafuse.

It depends on whether you believe in innocent until proven guilty and due process, or trial by social media. As I understand it no-one has been charged with any offence yet, never mind found guilty in court. There are suggestions of a hate campaign from disgruntled ex band members. Personally I don’t think people should be “cancelled” purely by the power of social media and without the backing of fact.

They charge faster than ordinary LA, less prone to sulphation and have a better cycle life. But starting from a low bar, "faster, less and better" may not be saying much. I think Li will be far better in all these parameters. And they are only marginally cheaper than Li, probably more expensive if you grow your own Li.

The damage mechanisms for low and high temperature are quite different. At low temperatures it’s about the slowness of the charging reaction resulting in “free” lithium getting plated on the electrode. But it is certainly not a binary thing. At low temperature Li can still be charged without Li plating, provided the current is kept low within the ability of the reaction to absorb the Li. A cheapo BMS with FET switching can’t unilaterally regulate the charge current, it has to be on or off. But a in a better designed system the BMS a would instruct the charge source to charge slower at low temperatures. I think around 10% of normal max charge rate is ok down to about -10c which is way below what a narrowboat installation is likely to encounter. In my case max charge rate is already only about 0.33C and the normal rate is about 0.15C, so I allow that down to zero and below that charge voltage is reduced to give a trickle charge. I have never found the batteries below zero though, as the inside-ish - modern aft engine compartment in trad stern boat.

I haven't looked at it in detail, but I would have thought a cheap BMS could be used to monitor cell voltages whilst not passing the main current through its MOSFETs. So the over and under charge built in disconnect would not work, but perhaps there is a digital signal output that could be used to send the stop charging command when there is a cell high voltage? Edit. Then again maybe not. Well the JBD BMSs that I looked at dont seem to offer any sort of digital "disconnect" output. Shame.

I built my own battery because at the time the alternatives were things like Victron and Mastervolt, which were very expensive. Now with the likes of Fogstar there is probably less reason. But apart from cost, I didn’t want MOSFET switches in the circuit, I wanted a relay. And I wanted to have a BMS that did exactly what I wanted it to do. If you are going to buy cells and an off the shelf BMS then I can’t see much advantage other that more confidence in build quality (if you are competent!) and an ability to determine the shape. Hence my idea about inserting a soft copper washer…

yes there would need to be some compression. Enough but not too much! But my point is that for a given compression (bolt torque), conductivity would be improved by having a soft copper washer. But this is only conjecture, I have no evidence!

If you look at a junction between two nominally flat surfaces (the battery terminal and the ring terminal) at the microscopic level, you would see a lot of peaks and troughs and generally rough surface. The actual metal to metal contact area is thus only between the places where the peaks align, and thus is a fraction of the apparent contact area. Copper, especially recently annealed copper, is quite soft and so my contention (not proven!) is that inserting a relatively soft and squishy conductor between them will to an extent conform to the rough surfaces and thus increase the contact area. Of course the cable-end ring crimps are notionally made of copper, but I think it’s some sort of alloy and they are tin plated which I suspect increases the surface hardness.

Yes. For a year or two the cells were fine. Then suddenly one suffered this poor internal connection issue. So most of the current went into the other 2 cells and fully charged them to 3.65v. The third dodgy cell was also apparently at 3.65v but this comprised (say) 3.4v cell voltage plus 0.25v dropped in the resistance of the poor connection whilst under charge. The effect was an apparent (though not real) loss of capacity of that group of cells. I remedied it by turning the big nut on top of the cell - the nut that says "do not turn this nut" in the instructions - just by 5 degrees or so, there was a bit of a click and everything has been fine ever since. These are CALB cells with a nut on top, a bit like the Winstons. For some reason the nut was well connected to the cell electrode, but the central bit on which the cable lug sits, wasn't. Which was very odd! Turning the nut must have removed some corrosion thing I guess. So it was not really about the cells being well matched or not, it was about a manufacturing issue that showed itself after a couple of years - some sort of poor internal connection to the terminal. Maybe I was just unlucky, but otherwise the cells have been faultless. Well I have the "NO!" configuration. The text is correct in theory but in practice, with the low C charge and discharge rates on a narrowboat (not having electric propulsion) I don't think it is an issue. The cell resistance is a bit more than 1mohm anyway, especially when you add the interconnects. At rest there will be a degree of self-levelling and the cells never get significantly warm from charging or discharging (just a few degrees when charging) so I don't see it as an issue and it makes life a lot easier.