BYD Atto 3 12V starter battery failure

BYD Atto 3 12V battery issue

The Atto3 EV uses a standard 12V 36Ah starter battery, which requires a charge absorption voltage of 14.4 to 14.8V. However, the maximum charging voltage I have recorded is only 13.8V, while the nominal charge voltage is only 13.75V. This is likely to be another part of the reason behind the battery failures. Not reaching the required absorption voltage will reduce the battery lifespan due to sulfation.
However, I also believe there was a faulty batch of batteries that were likely to be left in a partial state of charge for a prolonged amount of time, leading to some initial sulfation.

Kia has announced they will use deep-cycle AGM batteries (which they are calling EV 12V batteries), which makes more sense as they generally last longer in a partial state of charge condition. EVs do not have starter motors and do not require high cranking currents, so a general deep-cycle battery makes more sense.

Replacement battery options

A GEL VRLA battery would be more suitable, considering the 13.8V charging voltage more closely aligns with GEL (ref attached image)

I can see no reason why the 12V battery in the Atto 3 could not be replaced with a more durable 12V AGM or GEL lead-acid battery. Has anyone done this or, more so, has anyone found a battery that fits the Atto 3 battery clamp. I’m looking to fabricate a new clamp to house a more appropriate battery.


This issue has been discussed at length elsewhere, e.g. on NZ Geekzone. A well-respected EVSE manufacturer in NZ has been working hard on sorting it out. BYD NZ has issued at least three OTAs in an attempt to address the issue. (Note: I have no idea what country you are writing from, nor what sort of EVSE you are using.)

I don’t own an Atto 3 but I have looked carefully into the recharging of lead-acid batteries on the Nissan Leaf, and also into the (quite muddy waters of!) sulphation of lead-acid cells.

Much much depends on the particulars of the lead-acid cell, and as you point out there’s quite a bit of difference between the manufacturer-recommended charging regimes for the various general classes of cells: gels, flooded, AGM.

There’s also quite a lot of temperature-dependence, as well as voltage-dependence and even current-dependence, on the rate of sulphation. Very roughly: at low states of charge, or in high-current draws at moderate states of charge, sulphation will occur.

Desulphation is quite controversial. Some vendors sell specialised desulphation equipment, and some advertise it as a feature on their lead-acid chargers. It’s hard to generalise safely across all types of lead-acid cells, but the cell has to be quite hot for the desulphatiion to occur at any great rate… all to say that I’m not optimistic that, on the particular lead-acid cell that was standard equipment when you purchased your Atto 3, that a charging voltage above 13.9V (but below 15V) would result in much desulphation.

A higher charging voltage (if sustained for more than a few minutes) would, I suspect, be quite hazardous if the Atto 3 charging circuitry is similar to that of the Leaf, i.e. if it doesn’t control the charging current. And, unless you figure out how to control the charging current on your Atto 3, it’ll charge a gel battery too rapidly to preserve its life-expectancy… possibly even cause it to overheat or bulge (from the hydrogen gas that’s emitted from an overly-rapid charge).

I’d guess the BYD engineers picked 13.9V as the maximum charging voltage to ensure that the lead-acid battery doesn’t ever get into the danger-zone of an overly-rapid charge. Depending on the cell and the ambient temperature, 13.9V is pretty close to fully-charged, i.e. if you want to charge it any higher, you’d have to go into what is marked in your plot as an “Acceptance Phase” (where the current is limited).

I’d further guess that some of the BYD 12V problems are with cars which are routinely left on their EVSE overnight, by owners who have used a timer-function on their EVSE to start the charge some hours after they insert the EVSE’s plug into their vehicle. Depending on how the EVSE is programmed, this may prevent the Atto 3 from going fully to sleep… and when it’s “awake” it’s the itty-bitty 12V battery powering all of its microcontrollers, quite possibly discharging the 12V battery more rapidly than the Atto 3’s normal top-up cycles (from the traction battery) will recharge it.

I’d also further guess that some of the premature fails of 12V batteries on the Atto 3 are occurring in hot climates. The charging regime for a 12V lead-acid battery really “should” be temperature-adjusted, and it might be that 13.9V is actually too high a charging voltage for an Atto 3 that’s parked up outside in the desert sun.

All to say that, if you are looking for an alternative to a BYD-recommended replacement lead-acid battery for your Atto 3, you might consider spending large on a lithium-ion 12V automotive battery. The best of breed of these have charge-controllers that are model-specific, i.e. they’ll modify the firmware in the built-in charge-controller of their units so that it’ll cope with the charging voltages and timed top-up cycles of the vehicle. See e.g. Ohmmu. Happy hunting!

1 Like

The only way to do a timed charge is to use the Atto’s scheduling software; if you try to use an EVSE’s timer facility, the car will go to sleep soon after you plug it in and the charge will never happen**. If you use the car’s scheduling, once you plug in the EVSE, the traction battery is connected and the 12V battery is on a trickle charge until the session has finished. So scheduled charging is actually a good way of looking after the 12V battery.

** Evnex, who you alluded to, have come up with a workaround, but it works by stopping the car from going to sleep, and hence it keeps the traction battery connected and trickle charging the 12V battery.

1 Like

Yes, that’s also my understanding: Evnex has found a way to “smart slowcharge” an Atto 3, i.e. to have the Atto 3 on an EVSE indefinitely. This is already important in NZ (and I presume also in your location) for anyone who has a rooftop PV array and wants to “smart solar” charge their EV. And it is also important for people in NZ (and I’m sure in other locations) who want to charge their EV from the (significantly less expensive) electricity that’s supplied to households and businesses who have loads (e.g. for an electric hot-water cylinder or an EV) which can be “shed” by the lines company whenever there’s any threat of an excess of demand over supply on their local line.

Having an Atto3’s charging session being controlled by its onboard timer isn’t helpful – and may be a feature-clash or at least be quite confusing – when the EVSE is “smart” enough to be controlled by a lines company and/or a current-sensing apparatus (on the electrical service-entrance to the property) which indicates that the property is exporting at least the 1.5kW that’s required to maintain an AC slowcharge in a country with 230VAC.

My longer-term hope for “smart” EV charging is that low-power DC chargers become readily available and quite affordable. A DC charge session can be reliably “paused”, using a standard protocol (in the CCS and CHAdeMO series anyway, I dunno about a Tesla). However the only reliable way to keep an AC charge session “alive” is to supply at least its 6A minimum. It’s my understanding that some EV manufacturers have non-standard protocols which will allow an AC charging session to resume after a power-outage of more than a few minutes, but I don’t know their details and I’d be a bit concerned about the risks of an EV “resuming” a session that was interrupted by someone unplugging its connection to the EVSE!

Interesting times!

The Atto 3 arrangement is disappointing, no question. OTOH, most vendors of solar-aware chargers recommend going with a 6A minimum setting anyway, so from that point of view it’s not too big a deal.

I think it’s fair to say the charging side of things is still in its infancy. That was one reason I was reluctant to spend more than I needed to at this time.

6A is the minimum allowed charging current for an AC slowcharge – this limit has nothing to do with Evnex or the Atto 3. It was decided in the previous century – when J1772 was developed in Japan and subsequently endorsed by the USA’s SAE. A few years later (in 2003), IEC issued 61851-1 which picked up the 6A limit.

Folks like me who have a wimpy PV array on their roof, and have a 230VAC inverter, are unhappy with this 6A lower limit. See e.g. Is it possible to charge EV's at a lower power? | Speak EV - Electric Car Forums. My personal workaround is to use a 3:2 stepdown autotransformer, so that I can charge “full-solar” for at least a few hours every day. Without the stepdown tranny, the 1.5kW minimum ante for a bog-standard “smart-solar” charging system would pretty much preclude a full-solar charge of my EV, except for an hour or three on bright summer days.

A 6A minimum limit is quite appropriate IMHO for low-voltage countries such as Japan (100VAC) and the USA (115VAC). The AC-DC converter onboard an EV behaves rather inefficiently, with a poor power factor, when charging at a small fraction of their rated capacity. My 2014 e-NV200 has a 3.6kW AC-DC converter, so 6A at 100VAC is 600W: which is 16% of its rated capacity, so it’s reasonably efficient. However it’d be equally efficient if provided with 3A at 200VAC… so I reckon the J1772/IEC61851-1 lower limit of 6A was overly restrictive back in the days when a 3.6kW AC charging rate was normal. However 6A (especially at a wimpy 115VAC) is too low to efficiently charge an EV that can accept 22kW of AC power … and indeed I have heard the Renault Zoe will refuse to commence a charging session if the EVSE offers less than 10A during the negotiation phase of the session.

And don’t be too hard on those BYD engineers! My 2014 Nissan e-NV200 has a 12V battery top-up system which, as with the Atto 3, doesn’t allow any “top-up” charge of the 12V system while the vehicle is plugged into a bog-standard EVSE. It’s a long-standing “quirk” (aka “usability defect”) of the Nissan’s BEVs – that you’ll return to a soft-bricked vehicle (which will require a 12V “jump-start”) if you naively go away for month with it parked-up and plugged into your EVSE. See e.g. Leave Plugged in or not when not driving for a month | My Nissan Leaf Forum. The folks at BYD were apparently aware of this issue when writing a warning in the user’s manual:

“When ignition status is OFF and VTOL is connected without output for a long time, the vehicle’s static power consumption increases. It is recommended that the discharging/charging connectors are removed when the device is not in use.” (’s-Manual-NZ-08_2022.pdf)

Yes, I understand that. I’m just saying that other solar-aware chargers, like the Zappi, recommend using the setting that drops back to 6A from the grid rather than stop-starting. And on the Atto 3 it’s effectively ~5A, because the car draws lower than the EVSE signals.

1 Like

Yes indeed, AFAIK very few BEVs have a (non-standard!) feature on their AC-charging system which would allow it to keep a session running if the EVSE supplies less than a (nominal) 6A. That’d be necessary for V2H; but that’s not yet happening outside of pilot studies. And it’d be very important for load-shedding on overloaded lines, but it seems only the likes of Evnex are making headway on this.

And yes, when an EVSE and an EV “agree” to run a 6A charging session, it’s the “responsibility” of the EVSE to deliver no more than this much current, and the “responsibility” of the EV to avoid accepting any more charging current than its battery can safely accept.

Electrical safety regulations (at least in NZ, and I think also in the USA and the UK) effectively require a safety-margin on the “nameplate” current draw of an appliance. If an appliance is on a “6A” setting, its manufacturer is in deep trouble if it ever draws more than this, so long as the supply voltage and power-factor and ambient temperature are within its operating spec.

I don’t have any experience with wallboxes, but I have measured the current flows through several different IC-CPDs over the years… and all had at least a 10% safety margin, but no more than 20%. All to say that it doesn’t surprise me that you have observed that a wallbox that’s on its “6A” setting is actually limiting the current to roughly 5A. Also, I’d be very surprised if an Atto 3 is ever doing any current-limiting on such a slow charge session, even when charging from 99% SoC to 100% SoC. It’ll DC-charge at a hefty 10kW when topping off to 100% SoC, according to a charging-curve I found online just now at BYD Atto 3 charging curve & performance :: Then again, if the battery cells are too cold to charge rapidly, I suppose it’s possible that the Atto 3 would sometimes limit its AC charging current below 5A??

That’s not how I read the standard. All the EVSE does is says my limit is N amps, and it’s then up to the EV not to exceed that limit.

But that irrelevant in this case – an EV isn’t an appliance. You’re confusing issues. Set an EVSE to 6A and plug in, say, a Tesla, and it will draw 6A, near enough. Plug an Atto 3 into the same EVSE, and it will draw ~5A. It’s a quirk of BYDs (and some other Chinese EVs).

The charger doesn’t limit the current (except in dire circumstances) – the EV does.

I have read only sections of the relevant standards documents.

However it seems we have gathered very different understandings of the function of a mode-2 EVSE! In my understanding (and also my observation) the EVSE controls the current using pulse-width modulation of the incoming AC power – essentially clipping its waveform so that it has no more than the negotiated ampacity. The EV’s onboard charger is what does the AC-DC conversion. It could do some PWM to further limit the power, but this would be a common occurrence only on a very high-power (>10kW) AC charging session, or on an EV which is trickle-charging its cells to get them in balance at 100% SoC… which is my best guess for why a Tesla 3 will limit itself to a 6A charging rate on an EVSE which is indicating (on its CP signal) that it can supply up to 12A. See

I find a simplified discussion of the CP signalling in J1772 in Wikipedia at SAE J1772 - Wikipedia. “The PWM duty cycle of the 1 kHz CP signal indicates the maximum allowed mains current.” A simplified schematic indicates to me that the EVSE drives the CP with a square wave that has a duty cycle which indicates the maximum current it is able to source. The EV doesn’t change this duty cycle; but it can adjust the voltage on the CP by switching in a shunt resistor – this is how it indicates that it is “ready to charge”. (EVs with lead-acid batteries may also use this line to request that the EVSE engage the ventilation fans in the garage.)

But… I can’t guess why a Tesla might draw 6A from an EVSE that’s charging an Atto 3 at about 5A… unless perchance this particular EVSE has (effectively) no safety-margin on its ampacity? Or perchance you’re measuring the ampacity in two different ways on these two vehicles? Especially when an AC is not a pure sine wave (and the PWM in an EVSE injects quite a bit of “harmonic distortion” into its power supply), it’s far from trivial to make an accurate measurement of ampacity. Certainly I’m unable to do so, but instead I rely on an uncalibrated “kill-a-watt” plug-in device which is wildly inaccurate in readout of power factor but which is within cooee of the net kWh I can estimate by more roundabout methods (such as the readouts available to me through the LeafSpy app that monitors the OBD2 signals which indicate voltage readings, amperage readings, and cumulative kWh values as estimated by the vehicle’s charging system)…

No, it’s the CP signal that gets modulated – the actual AC power is essentially connected straight through. The EV then reads the CP signal and limits its power draw accordingly.

Right: the CP signal – not the AC – indicates – not controls.

Have a look at the Signaling section of the Wikipedia page you mentioned: it has the CP doing “supply equipment current capacity provided to PEV” followed by “PEV commands energy flow”.

Hmm… thanks for persevering with this! I stand corrected!

I’m now (belatedly!) realising that an IC-CPD would be much larger – and it’d throw off a lot of heat – if it were chopping the AC power. All it really has to do in the way of “protection” against an overcurrent condition is to terminate the session if the EV’s charge-controller is drawing more current than it should. Most simply – the IC-CPD could have a fuse in the “essentially straight through” connection you describe. But if it’s a variable-amperage IC-CPD I think it’d likely have a current-sensor and a relay that’d trip if an overcurrent condition (or any other fault condition) is detected.

I found a picture online of the (small!) circuit board in a (rather old) 8A/10A IC-CPD.

I’m in a unique position to shed some light on the BYD (supplied by Leoch) NS40 battery. I work for an Australian battery manufacturer (you can guess) and we were recently asked to supply these batteries to the MyCar/Eagers bunch as a middle man. We conducted routine testing and benchmarked against our current import and local made equivalents. Most results were acceptable. However, the highlight was exceptionally poor endurance testing results - particularly 50% DOD, which I believe is a major contributing factor to premature failure (especially during the early software releases when I look at the voltage graphs people have posted online).

You have observed that the charging voltage does not exceed 13.8v and have posited that this may lead to accelerated sulfation which is correct. More correctly we could say that over time the natural tendency of the battery to sulfate is insufficiently reversed by it’s charging and usage regime. Not exceeding 13.8v for a flooded battery and being charged at low temperatures (relative to an engine bay) will mean that the battery would not be desulfating much. We actually do open voltage constant current charging in bath temperatures of around ~40C to desulfate.

Your assumption that some of these batteries had deteriorated between manufacturing and delivery is also true - one of the primary reasons for this request.

A deep cycle is definitely a better alternative, but your unlikely to find something in that small a footprint that isn’t a recreational/backup battery (please let me know if you find something). There do exists B20/NS40 size hybrid auxiliary AGM’s (japan/Yuasa and korean/hankook made) that would suit your purposes, but may require adapters for skinny post.

If you don’t mind me asking, what issues are you having with the fitment? Is it due to wanting a larger size? I was under the impression that a standard B20/NS40 size would fit no issues, but I have not looked at one in person yet.

@dimter89, thanks for the insight and feedback. It’s great to hear from someone who understands the issue and has tested the durability of these batteries. Sadly, it sounds like they are prone to failure and will no doubt continue to fail if they battery absorption charge voltage is not corrected with a BYD software update. :thinking:

Kings LFP 12V battery testing in the Atto 3

I am currently testing a cheap ($250) Kings 60Ah LFP battery since it was one of the few options I could find that fit within the Atto 3 battery mount. However, you cannot use the standard battery foot clamp, and I’ve yet to fabricate a permanent bracket until I’m happy with the performance (Just using a heavy-duty tie strap at the moment).
The good thing about this battery is the terminals are in the correct position (positive on the left), and the tabs just screw straight to the flat battery lugs without any modification. Just remove the old terminal lugs from the cables and use the bolts that come with the LFP battery.

It’s been working perfectly for a few weeks. The only main concern is whether the 13.8V charge voltage is enough to balance the LFP cells and whether it will reduce the lifespan of the battery over time. :thinking: The maximum charge voltage listed on the Kings battery is 14.6V.

1 Like

Hi any up dates on the Kings LFP battery test ? I also have an Atto3 just over 12 months old and looking at replacement options . Has anyone put in a shunt and recorded what loads are pulled from the battery ?

Ok, I’ve been using the Kings 60Ah lithium battery for several months now with no problems. I’ve performed a full 14.6V recharge using an external charger once about a week ago to top balance the cells.

I don’t have a shunt, only a battery voltage monitor, but the history log shows the current draw never seems to pull the battery voltage below 13V. That being said, LFP cell voltage doesn’t drop much under load compared to lead-acid. However, the Kings battery is 60Ah which about double the capacity of the original 30Ah lead-acid starter battery

That’s great I have just ordered one .I think the charge voltage 13.8 which gives each cell 3.45v cells spot on .I have 3.55 as my over voltage setting for my home batteries and start balance at 3.45v .An active balancer would be great if the case can be easily opened maybe a JK smart bms could be used and then the cells could be monitored My only concern was the 60amps and not knowing what BMS is in the Battery .As most LFP cells have a 1c rating the BMS should be al least 60amps and the specs say 60amps continuous …No starter motor so shouldn’t have any large loads but could always be that one time when everything is turned on and the BMS shuts off. Which is why I asked if anyone has installed a smart shunt to monitor the loads .