New BMS for bq76PL455A based batteries

[Pete9008]

Edit: Might design a board with just the transformer. This would be small enough to heatshrink inline with the cable for testing.

Turns out SOIC16 breakout boards can be bought off the shelf, so will just use one of those and solder the wires, at least for testing.

That's a fix not a bodge! A bodge would be wires soldered to the pins (possibly with hot melt glue or epoxy to reinforce it).

Don't get me wrong here, it's great to see things being done properly!

[catphish]

Good news! I was able to briefly test the transformer in the car today. It was running from laptop USB but with a good connection to chassis GND so hopefully a valid test.

I was able to charge at 10A without issue!

I'm slightly disappointed now that I didn't try this sooner. I will build some cables with inline transformers for a permanent installation.

[Pete9008]

Great news

Curious what the waveforms look like now. The low DC impedance of the transformer will mean that the idle state is very different so your likely to see lower signal levels. Not that that's a problem if the signal to noise is sooo much better.

Hindsight's a wonderful thing isn't it!

[catphish]

The low DC impedance of the transformer will mean that the idle state is very different so your likely to see lower signal levels.

Not the case, because I still have the AC coupling capacitors. You have just pointed out exactly why I *shouldn't* remove them though!

I should be able to make up the cables with inline transformers today, and will try to capture waveforms for comparison.

[catphish]

Transformer and breakout board.

[Pete9008]

Not the case, because I still have the AC coupling capacitors. You have just pointed out exactly why I *shouldn't* remove them though!

Good point!

Transformer and breakout board.

PXL_20220709_112619843.jpg

Looks ideal. Are you going to add something to support the cables (the interface between the soldered and stranded part of the cable is very susceptible to vibration)?

[catphish]

Waveforms as promised. Here's a capture of a multi-byte transmission followed by a single byte reception as seen at the differential receiver. The second capture is a close up of a received byte. The transforrmer is definitely making the pull resistors a lot more impactful, almost to the point where at the end of the byte the differential voltage is pretty small.

[catphish]

Looks ideal. Are you going to add something to support the cables (the interface between the soldered and stranded part of the cable is very susceptible to vibration)?

Hot glue and heatshrink

[Pete9008]

Waveforms as promised. Here's a capture of a multi-byte transmission followed by a single byte reception as seen at the differential receiver. The second capture is a close up of a received byte. The transforrmer is definitely making the pull resistors a lot more impactful, almost to the point where at the end of the byte the differential voltage is pretty small.
PXL_20220709_121428752.jpg
PXL_20220709_121646050.jpg

Curious what it's like at the far end, with the smaller capacitors it could be worse.

Hot glue and heatshrink

That should work. If you don't already use it I'd recommend the adhesive lined stuff for use on cars. Keeps the moisture out and also does a better job of strain relief. The 4:1 shrink ratio stuff also makes life a lot easier.

[catphish]

Curious what it's like at the far end, with the smaller capacitors it could be worse.

It actually seems that a lot of this was my scope. Switched to 10x scope probe and the waveform looks somewhat different (notice 2V rather than 1V scale), and with twice as much differential voltage. 2V isn't nearly the desirable 10V but good enough for communication to work reliably it seems.

[catphish]

That should work. If you don't already use it I'd recommend the adhesive lined stuff for use on cars. Keeps the moisture out and also does a better job of strain relief. The 4:1 shrink ratio stuff also makes life a lot easier.

Yep I've been using this stuff, really makes a difference. The differential signal on the other (battery) side of the transformer looks excellent.

[Pete9008]

Well spotted. Much better!

Can't see a good reason why it would be better on the battery side though??

I try to never use 1x on a scope probe. Generally the only time I ever switch mine to 1x is if I'm using them backwards to feed a signal in from a sig gen (not recommended practise but often convenient!). Too much capacitance, you get the scope input capacitance, plus the lead capacitance, plus the probe capacitance, it can be as much as 100pF on a cheap probe - it can do a lot of damage on a high frequency high impedance signal!

Edit - just worked it out ~800R impedance on a 2MHz signal!

[catphish]

Well spotted. Much better!

Can't see a good reason why it would be better on the battery side though??

I try to never use 1x on a scope probe. Generally the only time I ever switch mine to 1x is if I'm using them backwards to feed a signal in from a sig gen (not recommended practise but often convenient!). Too much capacitance, you get the scope input capacitance, plus the lead capacitance, plus the probe capacitance, it can be as much as 100pF on a cheap probe - it can do a lot of damage on a high frequency high impedance signal!

Edit - just worked it out ~800R impedance on a 2MHz signal!

My probes are certainly cheap, and when I was testing previously, lots of configurations didn't work at all when probes were connected.

Just made up the inline boards. They seem pretty secure. I'll coat the back in hot glue then heatstrink. Hopefully the cables still reach!

[Pete9008]

Looks good, the loop of cable should help improve the robustness a lot.

TBH even good quality 10x probes aren't great and all of them will load the circuit to some degree. For any precision stuff at high frequencies you really need FET input probes, trouble is they are expensive, have limited voltage range, are easy to kill and either need a compatible scope to power them or a separate power supply. Very good performance though!

[Pete9008]

Just thought of a potential issue. While the Ethernet transformer is isolating it doesn't have a specified working voltage (a hipot test is specified but not quite the same thing). Ethernet networks are usually all at the same potential (mains earth) so it's not an issue, the isolation is there to break earth loops not to allow the equipment to float at HV!

In this application it's not providing safety isolation (the dual redundant caps are doing that so definitely keep them!) but it is likely to have a significant voltage drop across it for extended periods or time. It will probably last like this quite happily long term but then again it might not

If you do end up re-spinning a board it might still be worth thinking about putting an isolated interface on it.

[Volkmar]

Hi,

I am completly new in this forum and I am afraid, I am a coding fool

I have purchased for my homemade electric vehicle a complete Volvo XC90 hybrid battery and I have a SimpBMS for Tesla modules at home. Do you think there is a possibility to use the Simp BMS with the Volvo Batteries?

Just to clarify: I am really a complete novice in the area of programming and electronics. I'm good at steel construction and can also design a chassis, but I just can't get my head around the electronics. So please excuse my question!

It would be wonderful if you could help me!

Thank you in advance and best regards from Austria

Volkmar

[catphish]

I have purchased for my homemade electric vehicle a complete Volvo XC90 hybrid battery and I have a SimpBMS for Tesla modules at home. Do you think there is a possibility to use the Simp BMS with the Volvo Batteries?

Sorry I didn't reply sooner. Can you confirm exactly what batteries you have? Do you have a photo?

[catphish]

If you do end up re-spinning a board it might still be worth thinking about putting an isolated interface on it.

I built a new batch of these boards for Silent Classics this week. I updated the board to have 2 x battery interfaces and an Ethernet transformer integrated. I haven't tested this yet, but the performance should be the same as my external transformer setup.

[Volkmar]

Sorry I didn't reply sooner. Can you confirm exactly what batteries you have? Do you have a photo?

Now I have to say sorry for my late reply!

I have bought the complete battery pack of a Volvo XC 90 Plugin Hybrid from 2017.

I will send you detailed picture tomorrow, when I am back in my Workshop.

Have a nice weekend

[catphish]

I have bought the complete battery pack of a Volvo XC 90 Plugin Hybrid from 2017.

I've been wanting to know for a long time where these batteries come from, so thanks for confirming what car yours are from. It's good that you got the whole battery pack, because assuming it's the same pack as mine, it should come with a really high quality set of fuses and contactors. Are you running the full 3 x strings of 6 modules?

It certainly makes sense for you to use my BMS if you're using these batteries, however this isn't really an "off the shelf" product, so there are a couple of things you will need to consider.

Firstly, I do not have any boards with integrated transformers. If you read through the recent posts on this thread, you will see that the BMS requires a transformer between the BMS and the first module in each parallel string, so if you'd like to use it, you have 2 options:
A) Use an older board I have in stock, but you will need to build transformer adapter boards as shown above. This is how my car works.
B) Order a new batch of BMS boards with integrated transformers. I can provide all the design files, but it'll probably cost around £250.

Secondly, you will need to consider how you will do charging. In my car, I have a second board that interfaces with the BMS, with the charge socket, and with an Elcon compatible charger (this includes Ovartech and Dilong chargers). If you're doing the same, you can use one of my boards, but if not, you will need to think about integrating the BMS into whatever charging setup you have using its CAN protocol.

[catphish]

Interesting and inevitable development - the BQ76PL455A is now in stock. You can buy the chip for £18+VAT from Mouser!

With this in mind, it actually seems senseless to continue to use my RP2040 design, as I have no doubt the official chip will do a much better job of the daisychain communication. It would be awesome to design a board with an STM32 and one of these chips! Unfortunately I now have little motivation to do this as my car is working just fine with the DIY design.

[Kelju]

Hi,

I finally got to the point in my build that I was able to test your board in my RX8. I have 40kWh worth of modules from a Mercedes EQC 80kWh pack installed. The EQC battery has the same bq-series cell instrumentation ASICs.
I made some slight addition/modifications to your code so that I have min, max and avg cell voltages reported in one message and temperatures and the pack voltage reported in another one.
I have the older version of your board with 3v3 as the serial comms supply. I managed to custom fit a LAN transformer on the board and also increased the inline capacitances, so apart from the supply, everything indicates a walk in the park...
I modified my Zombieverter embedded BMS-code to work with the aforementioned messages and was able to connect the battery with a successful DC-link precharge.
I then pursued to give it some "gas", but as soon as the Leaf inverter starts to spin, the battery voltage value reported by the Raspberry is jumping all over the place.

Was it so that you experienced issues only with charging and not on drive with the Tesla inverter?

[catphish]

Hi,

I finally got to the point in my build that I was able to test your board in my RX8. I have 40kWh worth of modules from a Mercedes EQC 80kWh pack installed. The EQC battery has the same bq-series cell instrumentation ASICs.
I made some slight addition/modifications to your code so that I have min, max and avg cell voltages reported in one message and temperatures and the pack voltage reported in another one.
I have the older version of your board with 3v3 as the serial comms supply. I managed to custom fit a LAN transformer on the board and also increased the inline capacitances, so apart from the supply, everything indicates a walk in the park...
I modified my Zombieverter embedded BMS-code to work with the aforementioned messages and was able to connect the battery with a successful DC-link precharge.
I then pursued to give it some "gas", but as soon as the Leaf inverter starts to spin, the battery voltage value reported by the Raspberry is jumping all over the place.

Was it so that you experienced issues only with charging and not on drive with the Tesla inverter?

There are a couple of things I'd suggest:

1) Make sure checksum verification is enabled in your code. This will cause corrupted data to be ignored rather then reported. It doesn't solve the underlying problem but it will prevent the corruption from messing up your display. https://github.com/catphish/ti-daisycha ... bms.c#L520
2) In addition to the Ethernet transformer, try with different types of cable run and try to keep the run to the first cell as sort as possible. I've heard that different types of cable can have wildly different results, and in particular shielded twisted pair doesn't seem to work very well (I think it has too much capacitance line to line).

[Kelju]

Thanks for the tips. I have to check that checksum verification.
The OEM slave boards with the bq-chip have a particular LAN transformer so I took it from one of the unused OEM boards and fitted onto your board.

It is quite interesting that on the contrary to your suggestion, the OEM wiring harness is a twisted pair shielded cable where the shield has been connected to one end of the cable with a third wire. So the shield is always connected only from one end to the OEM board. I have not yet checked where it actually connects there. The OEM board is packaged to a plastic housing so at least the shield is not connected to the vehicle chassis.
I am using one of the OEM shielded twisted pair cables to connect to your board. The length of the cable is maybe 20 cm long.

[catphish]

It is quite interesting that on the contrary to your suggestion, the OEM wiring harness is a twisted pair shielded cable where the shield has been connected to one end of the cable with a third wire.

Unfortunately, I've found this to be rather unpredictable. In my own experience the best results were achieved with a short twisted unshielded cable. In my own car I'm using the 5v version with an external transformer. I think there's some photos above of this above.

I'm very interested to hear you got this integrated with Zombieverter. Is the code for this available? Somebody else who bought a batch of these boards wants to integrate with Zombieverter so this could save some effort.

If you continue to struggle with interference, I will send you a 5V version. I don't have any with integrated Ethernet transformer, but I can send you the version that I'm using in my own car. It's still not perfect, but with checksum checking, the majority of the data is valid, and any bad frames are simply ignored.