My version of Li-ion modules for Toyota/Prius NiMH

Take your time... I'll be here.

I haven't spent enough time in both camps to fully understand the Toyota mindset...
...but if members on other Toyota forums behave like they do here on PriusChat, I think that's the answer.

Based on my experience with the Toyota's hybrid electronics, there isn't much difference between Honda and Toyota hacking from a technology perspective. Honestly, the Prius is easier from a hardware perspective, as the only electrical interface I have to interact with is a single 8p connector, whereas the Insight has quite a few more other signals to spoof/appease.

It's funny because most people think that moving to a serial-based architecture makes things harder, but IMO it makes everything easier... with just three wires on the CAN bus, you can basically do anything. This reminds me of a video Andy did a few years back:

To clarify: Someone from Team Jack thought they were insulting me by branding me with that moniker. I thought it was funny, hence the reason it's now in my signature. Honestly it correctly describes my affinity, so not sure why they thought it would bother me. Bless their heart.

I'm the first to admit I've logged very few hours in the Toyota ecosystem. However, so far my career odometer is around 70k hours, during which time I've become exceptionally good at assimilating, abstracting, and applying knowledge. In other words, my inexperience with Toyotas specifically doesn't really matter much, because with enough experience it doesn't really matter what logo is slapped onto the device you're tinkering in.

For example, even though I'm new to Toyota, I'm probably the only person outside Toyota who has deciphered their gen3+ hybrid serial data bus. tbh it wasn't even that hard, although it did require focus for numerous hours. I'm retired, so I get ~80 hours a week to spend on whatever engineering projects I'm most interested in. Not having deadlines breathing down my neck is both a blessing and a curse, particularly since I'm almost always working on several projects simultaneously. Time is certainly my most limiting factor.

LiBSU only became my primary interest in the last couple weeks. Right now I'm ramping up to 50 hours a week in Toyota Land.

I don't have a concrete answer to your question. Ask me again next year and maybe I'll have more Toyota domain knowledge.

 

Are you intending to place an inline power MOSFET in series with the pack positive?
If so, any out-of-bounds excursion will disconnect the traction battery from the hybrid system, which is a jarring customer experience. We're talking MIL and car dead on the side of the road. That design could meet ISO26262 ASIL-C safety requirements, but it won't be pleasant.

Active balancing is overkill for automotive traction batteries.
Packs in poor health won't benefit much from active balancing.
Packs in good health will stay balanced with C/100 passive balancing (e.g. 50 mA for a 5000 mAh pack).
Active balancing will work, but it's not necessary.

Yes, LiBSU generates every battery parameter, and has absolute control over battery charging/discharging. Specifically, LiBSU can prevent pack current whenever it wants (e.g. cell full, cell empty, too cold, too hot, etc).

Moreover, LiBSU has numerous levers to pull to gradually coax the hybrid system into charging/discharging/idling the pack. This allows LiBSU to force/limit/prevent assist/regen without using the nuclear option (i.e. opening a high power MOSFET). Of course, if the SHTF, LiBSU can open the OEM contactors and assert the MIL, too.

 

What will it block, and how?

 

I always appreciate people sharing information, so thanks.

The 'happy path' range I've seen spans from 12.9 volts to 17.9 volts, which is close to your observations. However, note that this is only the happy path, when all cells are at the same voltage. In the video I posted in #6, I describe some of the pitfalls you can run into if all four (or five) cells in a given BSU voltage measurement aren't the same.

I see four problems with a 5S/tap setup:

1: You should characterize whichever cell you're using before you commit to 5S. In most NMC lithium cells, the resting cell voltage plummets as you drop below 5% SoC. For example, the 5AhG3 cell I use in my Honda Insight lithium battery product has the following resting Voc->SoC LUT:
Vcell -> SoC
3.22 -> 5%
3.17 -> 4%
3.10 -> 3%
3.05 -> 2%
2.95 -> 1%
2.70 -> 0%

2: Cell resistance increases dramatically below 10%. The Prius computer will fault out when it sees high resistance (e.g. voltage sag during heavy assist).

////////////////

3: In normal operation, you absolutely don't want to operate near a cell's maximum safe operating voltages (e.g. 2.75 or 4.25 volts).
In most lithium cells, cycle lifetime decreases dramatically below 10% (and also above 85%). For example:
-a cell continuously cycled from 10% to 80% might last QTY5000 cycles, whereas;
-a cell continuously cycled from 00% to 80% might last QTY250 cycles.

Cells typically like to stay between 10% & 85% SoC. You'll need to look at your cell datasheet to find cycle lifetime information. So then for my 5AhG3 NMC lithium cell, my product keeps cell voltage between 3.45 volts (10% SoC) & 4.05 volts (85% SoC). This is to maximize cell lifetime, and also go provide headroom to prevent the cell from over/under charging.

For maximum lifetime, you ideally want to keep your cells between:
-13.8 & 16.2 volts for a 4S/tap BSU blade setup (right in the middle of the BSU's 12.9 to 17.9 volt range), or;
-17.2 & 20.2 volts for a 5S/tap BSU blade setup (outside the BSU's range; the BSU will overdischarge a 5S NMC setup)

As you can see, without a replacement BSU designed for lithium, neither 4S nor 5S is suitable for use with the Prius' NiMH BSU.

////////////////

4: A 5S/tap configuration will leave the cells chronically undercharged. What happens when a customer turns their car off with the pack at 0% SoC (e.g. extended airport stay)? There's basically zero energy available in the pack, so the pack voltage will plummet rapidly over the next few days. I strongly encourage you to measure your cell's energy curve (in mAh) versus voltage... if you do the math right, you'll see that a 5S setup will only use a small portion of the cell's nominal mAh rating.

What is your 'soft' protection mechanism?
Please don't say active balancing (which doesn't work for reasons I described in my video in post#6).

Where does your BMS source energy if the pack is overdischarged? Please don't say "from other cells", because that energy isn't available even during the happy path case (i.e. all cells balanced, yet overdischarged).

Where does your BMS sink energy if the pack is overcharged? Please don't say "into the balance resistors", because they can't dump the 75000 mA the electric motor will dump into the pack if the OEM electronics don't know anything is wrong with the pack.

My point is that your 'soft' limits don't work/exist in this application.
Regardling 'hard' limits, a properly design high current MOSFET switch will add safety to your proposed system, but it's an inelegant solution. Are you sure that MOSFET can handle opening when Vds@t+ is 300 VDC? That' MOSFET pair (you need to block both directions) is going to need a chonker heat sink.

No issues here.

 

That didn't really answer the how question ...

... but that did, thanks,

About the link you sent earlier, do you mean the 3/4-kilobyte URL back in post post #14? I don't think PriusChat's auto-linkification worked right on that one.

A lot of the time, when a URL is that long, you can manage to shorten it. A lot of those parameters at the end are for things like remembering your browsing session and your searching history and what ads to show you and so on. It's often possible to omit a lot of that stuff and post a much less unwieldy link that just identifies the page to look at. Also, you can make it a link manually (with [url=...]...[/url]) so as not to depend on PriusChat's software getting it right.

 

How does your BMS tell the car to stop charging in this example?

How does your BMS tell the car to discharge in this example?

As @ChapmanF points out, the link you posted previously wasn't formatted correctly, but this time around I got it working with:
https://www.aliexpress.us/item/3256805967270449.html
How will this device interface with the OEM BSU electronics (e.g. to limit assist and/or regen as needed, without opening the HVDC bus)?
More comments below.

Forced airflow doesn't make a passive balancing circuit 'active'. Active balancing requires joule shifting energy from cell A to B, typically through cell-to-stack transformers paired to each cell in series.

I think what you're saying is this BMS has a bidirectional switch, composed of two parallel N channel MOSFET sets, which with both sets placed in series (Drain|Source||Source|Drain). This is essentially a switch that can open the HVDC bus in the event the BMS determines a cell is out of range.

Each FET 'switch' would need to withstand the entire HVDC bus voltage. For example, if cell#12 causes BMS#1 to open, BMS#2 won't also open. Therefore, the entire HVDC bus voltage will exist across BMS#1's switch.

 

I think Vencedor was using active/passive in regard to heatsinks, and you're using active/passive in regard to balancing circuits.

 

That's my primary concern with your proposed off the shelf BMS solution.
Note that my LiBSU controller will cost less than said BMS solution... once it's available (no ETA).
I'm not trying to cram my product down your throat, but I do want to encourage you to consider it because it will make what you're doing safe.

I suspect there's actually only one switching element (i.e. two MOSFETs whose sources and gates are tied together, as I described previously). I suspect that switch is opened/closed by firmware running on the processor inside the BMS. This clarification doesn't change the BMS behavior you've described, I'm just highlighting the functional difference in how the switch actually works.

Note that the Toyota hybrid system won't work if the OEM hybrid computer isn't able to sink/source current into/from the battery when requested. For example, if the battery is full and your BMS circuit prevents regen into the battery, then the gas engine won't be able to sink energy into the pack, which is necessary for the eCVT functionality to work. It's counterintuitive, but the Toyota drivetrain sinks current into the pack when the gas engine drives the wheels from a standstill.

Overall, if your BMS system ever opens the HVDC switch, you should expect the dreaded hybrid system malfunction alarm and all its related side effects.

I can think of several over-discharge events:
-a healthy pack that's been left to sit for a while, and/or;
-an unbalanced/weak pack with a failing cell (which requires the pack to not charge/discharge for safety).

You aren't going to break the OEM BSU by opening the pack+ or pack-. However, opening the pack in the middle can only occur as specific points... otherwise, you can damage the multiplexed, isolated ADC inside the BSU.

Respectfully, I don't think you yet understand all the safety elements required for an ASIL-C product like this. I strongly encourage you to read up on proper battery safety design. I'm a stickler for safety, but IMO it's absolutely required to understand the design elements that lead to safety before you start selling lithium batteries to customers. I absolutely don't want to discourage your progression on this project, but I suspect you're about to start an exciting roller coaster ride:

Where will your BMS discharge the 75000 mA current source, which is somewhere around 18 kW? That's around a dozen space heaters worth of energy your BMS needs to absorb... unless it can tell the OEM BSU to stop. This feedback disable mechanism isn't impossible to achieve with the OEM BSU... in fact, I've described how to implement it several times in my various videos regarding this topic over the past year.

I agree your BMS can conceptually absolutely prevent charging in the 'hard' disable method. I say conceptually, as I'm still not convinced the FET switch opening would survive such an event. Even if your FET is somehow rated for 600 VDC (necessary to survive a sudden open at high current)... it would also need to not explode as the semiconductors absorb a massive amount of internal energy as the Vds goes from ~0 volts to ~250 volts. For example, if the FET opens in 5 ms, the area under the curve would be around 100 joules. You can certainly build a discrete MOSFET solution that can handle that much energy released that quickly... but it's not going to be inside the BMS you linked previously.

I wasn't correcting the statement 'active cooling'... but instead the 'active balancing', a different topic entirely. Typically a fan is 'forced air cooling', but 'active cooling' isn't wrong in your context.

I suppose I'll worry about it for you, then.
If you don't already know the sound a MOSFET makes when it opens into destruction, I suspect you'll learn the sound soon enough. Note that a MOSFET that doesn't literally explode during an overvoltage opening event is likely to fail short due to avalanche breakdown. When this happens, the gate can no longer deplete the channel carriers, which will cause the die to conduct heavily. In other words, your 'switch' will probably fail short, hence your safety mechanism to prevent charge/discharge will fail.

If your manufacturer says yes, you should probably ask for their datasheet.

...

Reminder: As I've said several times over the past year, I ultimately want these aftermarket battery products to succeed. My goal here is to provide the information necessary to design safe products. I'm not attacking you personally... only your proposed product's lacking safety elements. I am often blunt in my replies, but please understand that I have the best intensions for your (hopefully) safe product. I am willing to keep discussing the technical merits/deficiencies as long as you are.

 

Hi @Vencedor,

Really nice thread — great to see a constructive discussion here at my favorite corner of PriusChat!

We actually experimented with NCM lithium chemistry about five years ago. Unfortunately, several versions didn’t pass our End-of-Life (EOL) testing — and one of the test cars even caught fire as a result.

To simulate an EOL condition, just drop one of the block voltages about 1.0 V lower than the rest. No matter the Prius generation, the system will enter a “high-voltage state” while trying to rebalance the blocks. This is built into all Toyota hybrids running NiMH.

In this state, you’ll see the pack voltage climb up to around 260 V for an extended time. I’ll attach a Dr. Prius App screenshot as proof — hopefully it’ll save you a few years of frustration. Based on the math,

260 V / 56 cells = 4.64 V per cell,
which is way above the safe operating range for NCM lithium — that’s why that project failed.

For LiFePO₄, the math works out to

260 V / 70 cells = 3.71 V per cell,
which does function but severely reduces long-term battery life. That’s one of the key reasons we’ve now switched 100% to sodium-ion chemistry, which operates comfortably between 1.5 V and 4.0 V per cell. At the end of the day, safety is the most important thing!



Once your battery passes the EOL test, I recommend running these additional tests to verify long-term durability (I cover them in detail in one of my V3 videos):

1. Postman Test – Simulate a stop-and-go delivery route: accelerate to 45 mph, then stop to 0 mph repeatedly for 7 hours a day over 2 months.
   
   
2. Cold Shower Test – On a cold morning (~50 °F battery temp), roll downhill before the car warms up — repeat throughout winter.
   
   
3. Diarrhea Test – Discharge to 40 % SOC, then full-throttle for 10 seconds to drain the battery even further, repeat that every day for over 3 months straight.

We actually had to hire several beta testers to test all these scenarios, not an easy job!

As far as BMS, to construct a BMS that can handle 200v of lithium battery isn't an easy job, I believe Muddler can help you on that, just keep in mind that Prius battery is operating in a semi-open environment that is subject to dust, condensation and debris, also you can't just shut off the battery all of a sudden as the drive could be in the middle of highway merging, your best bet will be some type of warning or reduce power so that driver can get a chance to pull over safely.

Hope this information helps you and the Prius community, we've also test the LTO (Lithium Titanium) and the result can be found in one of my earlier post.

Best,
Jack

 

Thanks for taking the time to provide this background information, Jack.

LiBSU can prevent the EOL scenario you described from ever occurring. One method is to spoof a much larger blade voltage delta (e.g. 5 volts) if any lithium cell(s) enter(s) an unsafe operating region. Therefore, instead of entering your EOL condition, LiBSU can disable the entire hybrid system when a cell fails.

I agree that allowing NMC lithium cells to reach 4.64 V/cell is a recipe for failure. Fortunately, LiBSU will provide per-cell supervisory control, so we can prevent excess charge/discharge from over/under charging the cells.

...

Thanks again for taking the time to provide your empirical data from your previous prototyping. I'm certain that LiBSU's per-cell SCADA will drastically reduce thermal events to industry acceptable numbers. We can't eliminate all risk, but we can do our best to design products that don't put their operators in harms way.

 

With 5S/blade, you're looking at a per-cell operating range between:
~2.5 volts (under 0% SoC) on the low end, and;
~3.5 volts (15% SoC) on the high end.
(Note: I used the discharge curve for my 5AhG3 cells. Your lithium cells might be slightly different. This is just an example)

Therefore:
-you're only using 15% of the lithium capacity (e.g. 150 Wh in a 1 kWh pack), and;
-you aren't using 85% of the lithium capacity, and;
-you are operating below 10% 2/3 of the time (not ideal for cell lifetime), and;
-you are operating below 0% when the pack is empty (not ideal for cell lifetime), and;
-your cells are going to stay chronically undercharged, and;
-the cell resistance is going to be much higher below 10% SoC.

Similar to above, sodium's voltage range is very wide, which means the OEM BSU doesn't use much of the total cell energy. In my testing of NexPower's V3 sodium cell, they're only using about 1/3 of the total energy. It was something like 13 Wh usable per 33 Wh cell (I'd have to look at my notes for the exact value).

It's very likely that with a replacement BSU, we could stretch sodium's voltage curve, thus increasing the usable capacity. Based on what I've seen, it seems like the Prius' main hybrid computer blindly 'trusts' the blade voltage data sent by the BSU. In other words, if we tell the hybrid system the blade is at 12 volts – but it's actually at 14 volts – the car doesn't appear to mind (within reason, and I've barely tested this, so take it with a grain of salt). In short, I suspect a replacement BSU would allow us to use most of sodium's energy.

I agree with the science that sodium is undamaged by below freezing temps.
However, in my testing there's almost zero usable energy below freezing.
In my testing at -20 degC, usable pack energy dropped to less than 1% of the available energy at room temp.
Sodium still wants in subzero simply because there's no lithium plating issues.

However, I also observed incredible high self heating regardless of temperature. Both NexPower cylindrical cells I tested (beta and production) behaved more or less the same, and the results are abysmal... these cells overheated in my test gauntlet after just QTY3 cycles... I've never failed a lithium cell for overtemp. Typically this test runs for several months, but with sodium it was less than 20 minutes before the cells overheated.

This isn't correct. For example, CATL has several production sodium cells.
NexPower isn't the only player in the sodium automotive space.

Sodium is newer tech. I think it's ultimately best suited for stationary storage, as the energy density is abysmal compared to lithium. Price is the major reason to use sodium. Once solid state lithium cells exist, they will dominate the automotive market.

We'll see more sodium cells flood the market until solid state lithium cells exist. Of course, those are always five years out. Science is hard. But assuming it happens eventually, that's what'll be in cars.