Dyson v7 Trigger cordless vacuum - TEARDOWN of battery pack

[tinfever]

Measuring ESR (equivalent series resistance) of a battery cell isn't a one step measurement with a multimeter. I believe one way to do it is to measure the change in cell voltage with respect to an increase in load current.

For example:

With 0A load (no-load) a cell measures 4.20V.
With 3A load, the same cell measures 3.87V.

Ohms laws is V=IR. Voltage = Current * Resistance

Thus:
4.20V-3.87V (our change in voltage) = 3A-0A (our change in current) * Resistance
330mV = 3A * Resistance
330mV/3A = Resistance
110mOhm = Resistance
Cell ESR is 110mOhm

There are also battery cell ESR testers you can buy I think, but I've never used one.

Regarding the fast cutout on high suction mode, if you wanted to, there is a setting in config.h that sets the low voltage cutout to 3000mV (3V). I think you could safely lower that to 2700mV. I'd be curious if that gives you any more usable high suction runtime.

For replacement BMS boards, there are some you can get on Aliexpress that I think I heard might have balancing. Not 100% on that though.

[diebog]

Regarding the fast cutout on high suction mode, if you wanted to, there is a setting in config.h that sets the low voltage cutout to 3000mV (3V). I think you could safely lower that to 2700mV. I'd be curious if that gives you any more usable high suction runtime.

Ill give it a try.  I found the file config.h do I just adjust the value here or do I need the MPLAB X IDE to change and create a new hex file?  I know a little on programing but not much as you can tell.   

[tinfever]

Regarding the fast cutout on high suction mode, if you wanted to, there is a setting in config.h that sets the low voltage cutout to 3000mV (3V). I think you could safely lower that to 2700mV. I'd be curious if that gives you any more usable high suction runtime.

Ill give it a try.  I found the file config.h do I just adjust the value here or do I need the MPLAB X IDE to change and create a new hex file?  I know a little on programing but not much as you can tell.   

You'd need MPLAB X to recompile the firmware. I think if you open the project in MPLAB X, edit the config.h file, and save. You should just be able to build and program. There is a chance you'd need to edit the project settings in MPLAB to use your PicKit though.

[diebog]

I downloaded the program, found the file and changed the value but haven't got a clue how to get it to work with the pickit 3.  Everything is so foreign to me.  I get message "Compiler toolchain "XC8" of type "XC8" does not exist. Please visit Tools/Options/Embedded to locate an installation of this toolchain or to select another compiler toolchain." when I try to compile and program chip.  So I manually downloaded MPLAB XC8 Compiler but again dont understated where I tell the MPLAB to use it      Any pointers?  Sorry for being suck a pain, im sure you fell like your 

[tinfever]

I downloaded the program, found the file and changed the value but haven't got a clue how to get it to work with the pickit 3.  Everything is so foreign to me.  I get message "Compiler toolchain "XC8" of type "XC8" does not exist. Please visit Tools/Options/Embedded to locate an installation of this toolchain or to select another compiler toolchain." when I try to compile and program chip.  So I manually downloaded MPLAB XC8 Compiler but again dont understated where I tell the MPLAB to use it      Any pointers?  Sorry for being suck a pain, im sure you fell like your 

You might need to download and install the XC8 compiler separately from here: https://www.microchip.com/en-us/tools-resources/develop/mplab-xc-compilers/downloads-documentation#XC8

Then try to restart MPLAB (also make sure your PicKit is connected for future steps) and see if it finds XC8. You should be able to go to Tools > Options > Embeddeded > Build Tools and see something like I show in the attached screenshot 1. If XC8 isn't showing there after separately installing, try clicking "Scan for Build Tools".

Then I'd suggest going in to the project settings by clicking on the tiny wrench and gear icon I circles in the second attached picture. Double check that the compiler toolchain of XC8 is selected. Then (you should already have your PICKIT connected to the computer) click the drop down arrow I also circled and select your PicKit. Then click OK.

Finally, you should be able to click the Hammer and broom icon for "Clean and build" in the upper toolbar (next to the play button). If it builds, then you can connect the programming wires to the BMS board, and click the button to the right of the play button for "Make and Program Device". With any luck that should work!

Also, MPLAB and the PICKits can be weird sometimes (that's not your current error reason though) so if things aren't acting right, it's always worth closing MPLAB completely, unplugging the PICKit, waiting a few seconds, plugging the PICKit back in, and starting MPLAB again. That's resolved more weird issues for me than it ever should haha

Hope this helps!

[diebog]

Thanks bud, after your help I got allot closer but still couldn't get the darn thing to recognize the pickit3.  After some playing around I had to first open the pickit3 software and "revert to MPLAB mode".  Then hold down the button on device while plugging in usb (dont know if this was really needed), and under "project property's>pickit3>option category's>firmware> uncheck "use latest firmware and then select the PK3FW_015609.jam file" went to compile and that manually forced it to download the correct firmware and finally it connected with the pickit3   Its odd because the MPLAB did recognize the pickit3 but when I tried to compile and program it kept telling me there was no device found,  after the firmware was updated it worked.  But all in all I did learn allot in the process though which was good.

Just want to make sure I adjusted the right spot, is this correct?  Also Is there a way to check to see if the data got changed on the BMS pic16LF1847?  I see there is a read function, its been awhile but if i remember correctly on the  Arduino IDE I was able to compile and program then reread and check that everything was compiled correctly on the Arduino itself.  With this do you just go off the fact if it complies correctly its good to go?  I noticed when I close out the project the "read device memory" function is no longer available, its greyed out.

[tinfever]

Thanks bud, after your help I got allot closer but still couldn't get the darn thing to recognize the pickit3.  After some playing around I had to first open the pickit3 software and "revert to MPLAB mode".  Then hold down the button on device while plugging in usb (dont know if this was really needed), and under "project property's>pickit3>option category's>firmware> uncheck "use latest firmware and then select the PK3FW_015609.jam file" went to compile and that manually forced it to download the correct firmware and finally it connected with the pickit3   Its odd because the MPLAB did recognize the pickit3 but when I tried to compile and program it kept telling me there was no device found,  after the firmware was updated it worked.  But all in all I did learn allot in the process though which was good.

Just want to make sure I adjusted the right spot, is this correct?  Also Is there a way to check to see if the data got changed on the BMS pic16LF1847?  I see there is a read function, its been awhile but if i remember correctly on the  Arduino IDE I was able to compile and program then reread and check that everything was compiled correctly on the Arduino itself.  With this do you just go off the fact if it complies correctly its good to go?  I noticed when I close out the project the "read device memory" function is no longer available, its greyed out.

Sorry for the delay in getting back to you.

You're definitely on the right track with the "revert to MPLAB mode". The PICKit has to be in a different mode with different software depending on if you are using the old PICKit standalone programmer software or MPLAB. Sorry I forgot to mention that earlier. It's really not obvious when MPLAB just says "Connection failed".

I think what you found with "use latest firmware and then select the PK3FW_015609.jam file" might be a red herring though. I think what you found as far as "revert to MPLAB mode" is all you need to do. Don't forget that closing MPLAB completely, unplugging and replugging the PICKit, and restart MPLAB can solve lots of problems.

If you adjusted "const uint16_t MIN_DISCHARGE_CELL_VOLTAGE_mV = 3000;" in the config.h file and changed it to something like 2700. That's the right spot. If you want to make sure your programming is changing something, you could change the line in config.h "#define FIRMWARE_VERSION 1;" to 2 or 3. Then when you use the procedure to check the firmware version (hold down trigger and then plug in charger, see Github page for more info) you should see 2 or 3 white blinks depending on the setting you used. That could be a quick way to tell your programming is working.

I think as long as you are getting to the message "Programming/Verify complete" in MPLAB, you are probably doing everything right though.

[zhoukevin]

For the V10 battery, I tried to use the EV2400 tool to reset the bq76930 flag bit successfully. However, since the ATSAMD20E15A in the system has been working, the flag bit is restored in the next step. So, do you have any way to let this single chip computer get away from the system before it can succeed!

[tude888]

Hello all,
I'm new to this forum but not new to repairing electronics, including fiddling with low-level programming.
First off, thanks to all who put in an incredible number of hours towards fixing the mess that Dyson created.

I am about to replace the cells in my V7, and will probably have to re-program the PIC (thanks tinfever for the video and write-ups)

That being said, a few questions came to mind as I browsed the thread:

- Has anyone tried to modify the BMS board in order to make it balance the cells?
- Second question: would it make sense to replace the original BMS with a cheap 6s BMS module that does the balancing on its own? (i know some functionality would be lost, but it seems the simplest solution)
Thank you all in davance!
T

[tinfever]

Hello all,
I'm new to this forum but not new to repairing electronics, including fiddling with low-level programming.
First off, thanks to all who put in an incredible number of hours towards fixing the mess that Dyson created.

I am about to replace the cells in my V7, and will probably have to re-program the PIC (thanks tinfever for the video and write-ups)

That being said, a few questions came to mind as I browsed the thread:

- Has anyone tried to modify the BMS board in order to make it balance the cells?
- Second question: would it make sense to replace the original BMS with a cheap 6s BMS module that does the balancing on its own? (i know some functionality would be lost, but it seems the simplest solution)
Thank you all in davance!
T

Unfortunately, right now my firmware doesn't support balancing even if you were to add the balancing resistors to the PCB. I highly doubt the stock firmware does either, although I guess I can't say for sure.

If you were to replace the BMS with a different BMS module, you'd have to figure out some way to connect the vacuum trigger to a switch on the new BMS to turn it on. Since the vacuum trigger literally just pushes a button on the battery to turn it on, the BMS has to handle turning the vacuum on when the button is pressed. I'm not familiar with the various BMS modules but if one of them had an output_enable pin you could attach to a button maybe it could work? It might be easier to attach the balancing module in addition to the stock BMS and let the balancing module do it's thing when it wants, although I'm not sure how it decides when to balance or if there would be any leakage current that could slowly drain the battery over time.

[tude888]

Thank you for getting back to me so quickly.

The BMS boards I was able to find are simple: the charging and output ports are the same. two wires (see attached).

My (very crude) solution is to connect the BMS directly to the charging port, and insert the trigger switch between the BMS and the vacuum motor. The problem with this solution is that the user will be able to  turn on the vacuum while it's charging. I don't think this could cause any major problems unless I'm missing something.

Sadly the vacuum charging plug has only two pins. If it had three, I could probably wire it so it would cut off the motor when plugging in the charger.

The other alternative would be to put together a simple circuit, maybe 3 - 4 transistors to cut off the load when the charger is plugged in. I'll put together a schematic, but first I want to test the BMS idea.

Will keep you all posted.

[KalleMp]

My firmware doesn't lock you out if the cells are far out of balance. However, it will shutdown with those 3 blue blinks if any battery cell goes too low. Of course, it can only charge the battery pack until the highest cell hits 4.2V, and only discharge the pack until the lowest cell hits 3.0V, so if the pack it out of balance, you'll get diminished usable battery capacity.

I wonder if you might like to save the time, cell voltages and temperature every time the trigger is activated and released, same for charger insertion.  It might give ideas on how the batteries behave and give warning of failing ESR.  I would be happy to sacrifice the bytes of EEPROM and the wear will likely never become a problem.  It could be used to snapshot the values when charging by pressing the trigger or plugging in the charger while doing load tests and then reading the data. Using IR-Da or RC5 codes could get data out without opening the case perhaps.

Your project has been phenomenal and just keeps on giving now that it supports V8 with just funny lighting effects.

[KalleMp]

- Second question: would it make sense to replace the original BMS with a cheap 6s BMS module that does the balancing on its own? (i know some functionality would be lost, but it seems the simplest solution)

The Dyson board also manages the power switching, a lot of amps and using a switch would need a beefy switch.  Also there is not much space to add a second board.  Connecting to each cell and leaving a multi-pin balancing charger connection on the outside is probably the easiest.  A possibly simpler balance capable board could be made that simply turns on the output when you press the switch but you would loose a number of protection modes.  Making a board that includes balancing support but is otherwise the same is also possible but might be more money than it is worth.

What is a cunning idea though is to do the design and order some boards etched, perhaps even a few prototypes populated at a service in the east, add the GitHub repo name to the silkscreen layer that includes the Hex File and BOM and some eastern knock-off shop will start making the boards at a price we could only dream of achieving, soon to be sold on Amazon and then callously ripped off by Amazon and sold under their own brand.

This may be the correct way to make open-source-hardware affordable.  Design and let it get copied in the east.  Instead of bemoaning the rip-off, expect and embrace it.

[tinfever]

- Second question: would it make sense to replace the original BMS with a cheap 6s BMS module that does the balancing on its own? (i know some functionality would be lost, but it seems the simplest solution)

The Dyson board also manages the power switching, a lot of amps and using a switch would need a beefy switch.  Also there is not much space to add a second board.  Connecting to each cell and leaving a multi-pin balancing charger connection on the outside is probably the easiest.  A possibly simpler balance capable board could be made that simply turns on the output when you press the switch but you would loose a number of protection modes.  Making a board that includes balancing support but is otherwise the same is also possible but might be more money than it is worth.

What is a cunning idea though is to do the design and order some boards etched, perhaps even a few prototypes populated at a service in the east, add the GitHub repo name to the silkscreen layer that includes the Hex File and BOM and some eastern knock-off shop will start making the boards at a price we could only dream of achieving, soon to be sold on Amazon and then callously ripped off by Amazon and sold under their own brand.

This may be the correct way to make open-source-hardware affordable.  Design and let it get copied in the east.  Instead of bemoaning the rip-off, expect and embrace it.

Apparently there is a Russian firmware version (not open source) that actually sends debug info out over the LED like you mention. It's a really cool idea. Now that I think about it, ideally you'd be able to have your smartphone camera decode it with no outside electronics required!

Although if the phone camera is limited to 60 fps, that might limit you to 60 bits per second of data absolute max if you only use on or off. I wonder how much data you could encode in the colors of the LED?

This is kind of insane but this paper makes it sound like the researchers managed to transmit 47 Kbps! It sounds like they are using a single RGB LED but I can't tell how they are getting 264 symbols/frame with a single LED.

"47 kbit/s (264 symbol/frame × 3 bit/symbol × 60 frame/second)"
https://opg.optica.org/oe/fulltext.cfm?uri=oe-27-23-33840&id=422856

it be clear, the odds I actually implement anything like this are pretty much zero. It'd be really cool but unfortunately being cool doesn't pay the rent 

[SilverSolder]

[...]It'd be really cool but unfortunately being cool doesn't pay the rent 

Being cool is much more important than paying the rent, though, so some kind of compromise needs to be achieved! 

[Steve1M]

Just joined this forum.
-
Such a relief to find this thread and your priceless youtube material, very much appreciated.
-
I have V8 with SV10 battery.
-
I sourced some Molicel 20700A cells, and have successfully changed out individual cells twice in the past, soldering, with no issue.
-
Poor performance yet again, tired of repeated strip down of pack, so decided to replace all 6.
-
Bought ten brand new Molicels, (expensive here, in UK) and and spent a long time performing discharge tests on quality RC iCharger, 2A down to 3V/cell, and managed to select 6 with only 40mAh between best/worst. Really lucky.
-
Charged all to exactly 3.85V each.
-
Spot welded them all in yesterday, which I was really pleased with, after having experimented for an eternity first.
-
End result -  4 x red LED flash if press trigger, or plug into charger.
If hold trigger, 4 x red, then 1 x blue.
-
Lots of web searching, to discover this Dyson PCB "kill feature", and additionally, was absolutely shocked to discover lack of BMS balancing function (an absolutely essential function of any BMS...unbelievable).....absolutely furious, after so much cell research/welding prep etc.
-
Anyhow, I only wish to be able to unlock the BMS, changing the 2 x values in the eeprom positions, and have a couple of questions:-
-
a) Can someone please tell me the correct pcb mapping (N7,N3,N4,N8,N9) to Vpp/Vdd/Vss/Data/Clock.
-
b) Also, if my pack is all assembled, do I need to check the Vdd box, or is the chip powered within the pcb?
-
Many thanks, Steve

[tinfever]

Anyhow, I only wish to be able to unlock the BMS, changing the 2 x values in the eeprom positions, and have a couple of questions:-
-
a) Can someone please tell me the correct pcb mapping (N7,N3,N4,N8,N9) to Vpp/Vdd/Vss/Data/Clock.
-
b) Also, if my pack is all assembled, do I need to check the Vdd box, or is the chip powered within the pcb?
-
Many thanks, Steve

On the GitHub repo (https://github.com/tinfever/FU-Dyson-BMS), part way down under "How to install it" there is an image showing the correct programming pinout. Edit: I'm 99% sure the pinout for the V8 is the exact same order, just with the ground pin in the same group. Reasons:

Just realized I never really thought about the V8/SV10 pinout. Given others have reported success with the V8's, not mentioning any pinout differences, another person (Aleksey1406) traced out some of the circuitry for the V8 and that also looks the same (https://github.com/tinfever/FU-Dyson-BMS/issues/20), and also someone else (dr-mark-roberts) reverse engineered part of the schematic long before I did any of this and their info matches (https://github.com/dr-mark-roberts/open-dyson-battery/tree/master/hw/docs), I think it is the same.

I recommend letting the microcontroller be powered by the PCB (Leaving the box unchecked, I believe.) You'll need to do something to wake up the BMS like push the button and apply a magnet (if applicable), then run the programming software while it is awake. I believe this is the safest method, although supplying VDD from the PicKit will work too, I'm just slightly more suspicious about doing it that way given how many times I've broken the VDD regulator portion of the ISL94208 IC for unknown reasons.

Hope that helps! Good luck!

[Steve1M]

 Yep, it's in the same order.....thanks.

Pic - bit concerned about how long it retains power if use magnet, press switch...will it corrupt if I mess up and it powers down mid-write?

Thanks for your intetest/time.

[tinfever]

Yep, it's in the same order.....thanks.

Pic - bit concerned about how long it retains power if use magnet, press switch...will it corrupt if I mess up and it powers down mid-write?

Thanks for your intetest/time.

Once it is in programming mode, the PIC isn't running normally and so it won't shut down on it's own. You'll notice that if you place the magnet and press the button to wake it up, then take away the button press and magnet, it stays awake for a bit. This is because the software inside decides when it is time to turn off again. I think just having the programmer connected and clicking Tools > Check Connectivity while it is awake is enough to get it into programming mode and then it will wait indefinitely for you to do whatever until you remove the programmer.

I wouldn't worry about it powering down mid-write. Since you are only editing the EEPROM hex to unlock your BMS, take very special care to configure the programming software correctly to only program the EEPROM. I believe the correct checkboxes and buttons are listed previously in the thread.

[Steve1M]

Excellent!!!

Many thanks for your concise, detailed explanation.

[Steve1M]

Followed your guide.
Silicon goop on pcb actually peeled away cleanly, using wooden toothpick to get under it, once I had cut around the perimeter really carefully using a fine-tipped scalpel blade, to avoiding shorting anything.
Residual silicon in the plated through holes popped through on insertion of pins.
-
Found magnetic switch only activates when placed at one end or the other. If too central it will not activate.
-
I confirm that all connected fine for my V8 SV10 pack using same pin designations as your V7 photo as follows (white arrow on Pickit being pin 1):-
-
N7>1   Vpp
N3>5   CLK
N4>4   DAT
N8>2   Vdd
N9>3   Vss
-
I just disabled Program Memory, took a read, then just changed EEprom position 1E to 00, as per dvd4me's guide....now activated with blue LED when hit switch!!!
-
If brave enough, in future, will test your full modded firmware on a spare tired pack, and see how it behaves, considering difference in number of LEDs between V7 and V8.
-
Many thanks again to yourself, and Mr "dvd4me" for your boundless generosity.
-
Steve

[dvd4me]

Many thanks again to yourself, and Mr "dvd4me" for your boundless generosity.
-
Steve

Yeah, dvd4me here is happy you managed to turn back on your "blue lights". Thanks to tinfever, to kindly support you over the path of "empirical" way of altering the eeprom.
Even though he tried to lure you to the "dark side"  you resisted 
Remember, once you go there there is no turning back. 
BTW, did you checked my other li-ion deep discharge recovery videos? It works for power tools battery or laptop ones too.
Or the PS3 de-lid and restore from RLOD, by applying direct pressure on the silicon die.
Cheers

[Steve1M]

I'll take a look at your other vids, thanks.
-
With regards to balancing the pack periodically, everything is so snug in the case, that I don't even have room to add wiring and JST-XH connector.
-
I have therefore ordered 7 x single pin male 8mm pogo magnetic connectors.
-
Will bridge pin to outer, (maybe this will marginally increasing contact integrity when attached), drill 9mm holes in sides of pack shell, and use these to tap off the cells to do a proper balance charge on one of my RC chargers periodically, plugging holes with 9mm blanking/blind  grommets when not in use.
(Note...I have plated steel strip. This will probably provide good contact for their non-magnetic pure nickel strip also, as 20700 have steel case, and magnets are strong)
-
This will save the frustration of repeatedly struggling with those battery shell tabs, yet again, which is always a cumbersome exercise.
-
I realise that all the above sounds a bit obsessive, but I have always had the maybe over-optomistic view that,  if you look after/maintain things properly, they will last forever.
(Have Black Decker dustbuster, 30 yrs old still going!!! Just keep the motor bearing lightly oiled, and added Lifepo4 cells many years ago....similar story leccy drills/screwdrivers etc...latter 30yr old BD9019 with lovely metal planetary gearbox..tiny dab molybdenum grease now and again, nicad replaced with 26650 salvaged Life cell years ago, all still good)
-
Gone off-topic I'm afraid..apologies.
-
Rgds, Steve

[tinfever]

I have therefore ordered 7 x single pin male 8mm pogo magnetic connectors.

If you wanted to be really slick, you could install a multi-pin magnetic connector like this (or equivalent) : https://www.adafruit.com/product/5468

I've never used these though. I might be a little worried about shorting the connections accidentally, but you could probably add some series resistors and a cover and be safe enough.

[SilverSolder]

I have therefore ordered 7 x single pin male 8mm pogo magnetic connectors.

If you wanted to be really slick, you could install a multi-pin magnetic connector like this (or equivalent) : https://www.adafruit.com/product/5468

I've never used these though. I might be a little worried about shorting the connections accidentally, but you could probably add some series resistors and a cover and be safe enough.

That's a pretty schmick connector...