Powerwall safety

Announcement - Help us fight the BOTS! Please report all spam including stuff in your inbox!

Nathan

New member
Joined
Dec 5, 2016
Messages
28
One thing I am doing as part of my design, is ensuring there are temperature censors between every 4 cells, and using a arduino to monitor the temperatures directly and reporting the data through a Raspberry pi and setting up text alerts for any batteries that go out of range. Potentially if I can get enough darlington transistors at a good enough price, covering the correct ampage, I could have thermically challenged cells disconnected on the fly

Fairly simple to setup using off the shelf and cheap components, currently intend to release all the work im putting in open source including the board designs. Time will tell if I can make it work though..
 

hbpowerwall

Administrator
Joined
Oct 7, 2016
Messages
1,977
Sounds like you have a cool idea, be great if you could start a thread here and I/we can folow your progress. Pete
 

Rerouter

Member
Joined
Jan 1, 2017
Messages
71
For these low voltages, you would likely be better off with a P channel mosfet to do the switching, still your going to be hard pressed to get the board the switching device mounts onto handling the high currents, every solution i can come up with involve soldering busbar directly to the pcb 0.5mm from the edge of the switching device, then flowing solder directly to the pins,

so far for the solder busbar directly to approach i can dream up would involve
1-2x http://www.digikey.com.au/product-d...068P03L3GATMA1/IPD068P03L3GATMA1CT-ND/5353501
The switching P channel mosfet, you ground its gate and it connects the circuit, use 1 or 2 depending on current level (1 covers ~65A)
1x http://www.digikey.com.au/product-detail/en/texas-instruments/TLV522DGKT/296-44391-1-ND/6110659
Dual rail to rail op amp for temp sensing / switching, 1 handles the temp switching, 1 buffers the temp signal out to your monitor via a current signal (gets around sensor being floated at different voltages) so you just need a resistor on the monitoring side referenced to ground to get the temp signal.
1x http://www.digikey.com.au/product-d...-america/NCP15WF104J03RC/490-2404-1-ND/588618
Tiny Temp Sensor, will technically double as a self resetting fuse as being mounted on the board that is then mounted to the pack would shut power from a cell if way too much current was pulled and heated up the mosfet.

and likely a trimpot and a few passives to do deadband and the like.
 

Nathan

New member
Joined
Dec 5, 2016
Messages
28
hbpowerwall said:
Sounds like you have a cool idea, be great if you could start a thread here and I/we can folow your progress. Pete

Will be putting the progress into the Project Daft Idea Thread as I get to that stage, just at the stage of waiting for the 18650 containers to turn up so I can charge/discharge the cells.


mike said:
I would be very interested if someone could document how to set up monitors through the RPI for both voltage and temperature. This is something I would like to do but don't have time to learn the electronics/programming involved.

Keep an Eye on Project Daft Idea, I will be going into great detail on there and most likely share the code/pcb design on github

Rerouter said:
For these low voltages, you would likely be better off with a P channel mosfet to do the switching, still your going to be hard pressed to get the board the switching device mounts onto handling the high currents, every solution i can come up with involve soldering busbar directly to the pcb 0.5mm from the edge of the switching device, then flowing solder directly to the pins,

so far for the solder busbar directly to approach i can dream up would involve
1-2x http://www.digikey.com.au/product-d...068P03L3GATMA1/IPD068P03L3GATMA1CT-ND/5353501
The switching P channel mosfet, you ground its gate and it connects the circuit, use 1 or 2 depending on current level (1 covers ~65A)
1x http://www.digikey.com.au/product-detail/en/texas-instruments/TLV522DGKT/296-44391-1-ND/6110659
Dual rail to rail op amp for temp sensing / switching, 1 handles the temp switching, 1 buffers the temp signal out to your monitor via a current signal (gets around sensor being floated at different voltages) so you just need a resistor on the monitoring side referenced to ground to get the temp signal.
1x http://www.digikey.com.au/product-d...-america/NCP15WF104J03RC/490-2404-1-ND/588618
Tiny Temp Sensor, will technically double as a self resetting fuse as being mounted on the board that is then mounted to the pack would shut power from a cell if way too much current was pulled and heated up the mosfet.

and likely a trimpot and a few passives to do deadband and the like.

Why a P Channel over a Darlington setup, I am new to transistors and so thats why my choice may be poor.

My current design (subject to change as im only 6 weeks into what I forsee to be a 12 month minimum project) is a single sided circuit board with standard pin hole through which I will solder the fuse onto, which will then goto the Cell.

Each Bank of 16 Cells will have a 6cmx6cm PCB, which will control the mosfet/darlington transistor arrangement, pulling in the data, and the PCB will connect to the main positive busbar.
 

Rerouter

Member
Joined
Jan 1, 2017
Messages
71
Mosfets dont draw any current apart from when the state is changed, and in general there on resistance is much lower. though need genenally 1-2 extra components to make switching reliable. (you don't want it sitting half way on at these current levels)

The components i chose was because including the temp sensor the current draw will be about 115uA at idle, but using a low power micro (not a arduino328), and by only turning him on from time to time, and letting him measure the temp, could be much lower, the switching circuit consumes at an estimate 15-33uA.

6x6cm is gigantic for the amount the circuit would be doing, and if you wish, i could probably help do a proper 2 layer pcb with smd for less cost and effort over a through hole one, mainly things get much cheaper with smd stuff, because that is what everyone is using for low power systems.
 

Nathan

New member
Joined
Dec 5, 2016
Messages
28
Rerouter said:
Mosfets dont draw any current apart from when the state is changed, and in general there on resistance is much lower. though need genenally 1-2 extra components to make switching reliable. (you don't want it sitting half way on at these current levels)

The components i chose was because including the temp sensor the current draw will be about 115uA at idle, but using a low power micro (not a arduino328), and by only turning him on from time to time, and letting him measure the temp, could be much lower, the switching circuit consumes at an estimate 15-33uA.

6x6cm is gigantic for the amount the circuit would be doing, and if you wish, i could probably help do a proper 2 layer pcb with smd for less cost and effort over a through hole one, mainly things get much cheaper with smd stuff, because that is what everyone is using for low power systems.

Ahh - I have some IRF3205's from another project and kept getting them sitting half way, and so looked at an alternative. can you give me any info on what causes this sticking and what components I would need? No worries if not I will also be searching google now i know its my lack of skill. Its the learning of the project that im loving so much and not the total practicality.

I agree that 6x6 is gigantic, its just a quick mental calculation based upon a 8 x 8 cm bank of 4x4 batteries, leaving 1cm around the edge. Without having the components to hand to measure properly (there in the post to me from china), I cant go into any more detail with my design as yet than a outline block diagram.

However I may not have described what I am using the through hole for properly. The system will use SMD for all components with the exception of the battery fusewire connection and wires to the thermisters, which I will use through hole for. This allows me to solder the fuse wires onto the batteries first, then push them through the PCB to solder the fuse on, with correct stand off's I should still have good backup fusing, with the MOSFET's controlling the batteries getting switched off.

I am using the Atmega328 as there easy to program, will allow me to report the data back to the main data storage systems and initially, for convenience. I am interested in using other methods as well, however due to limited experience am trying to keep this the one area I know, my intention once I have the system working using the arduino bootloader, is to learn a step further and rewrite the code in assembly to increase my abilities. It is this learning component which will give me a business benefit, as all coding outside my comfort zone improves my abilities and as such a increased power consumption will have to be tolerated. Once I have gained what I can from this (im not a fan of simple make blinky lights projects, done a version of this over i2C for this project for the charge/discharger, took me 3 hours to code, build the prototype and design a PCB that will meet my requirements). Once I have exhausted this learning process, I will start to look at improving each component's electrical performance before using it as a production system.

I hope this makes sense ;)
 

Rerouter

Member
Joined
Jan 1, 2017
Messages
71
Nathan said:
Ahh - I have some IRF3205's from another project and kept getting them sitting half way, and so looked at an alternative. can you give me any info on what causes this sticking and what components I would need? No worries if not I will also be searching google now i know its my lack of skill. Its the learning of the project that im loving so much and not the total practicality.

I am using the Atmega328 as there easy to program, will allow me to report the data back to the main data storage systems, done a version of this over i2C for this project for the charge/discharger, took me 3 hours to code, build the prototype and design a PCB that will meet my requirements)

the IRF3205 is and N-Channel mosfet, without a dedicated gate driver with charge pump, it likely wont work unless your switching on the negative side (This n channel turns on when the gate is 2-4V greater than the source pin, this would work, you would just have to make it clear its for the negative side (N channels at this current level are generally cheaper)

The main thing to get them to never be in the linear area of it is to use some strong hysteresis for your op amp operating theswitch, e.g. switch open at 60C, switch closed at 55C, rather than oscillate around 60C, to do this, google images is your friend,

For data comms, the main thing to remember is that each cells board is at a different voltage, so from what i can see, Isolated canbus would be about $6.70 per board, and consume 3mA even when idle, I2C as long as the monitors are slaves, could work very well by one of these per board, yes the bigger one, as he has enable pins that drop the idle current to 50nA,
http://au.mouser.com/ProductDetail/...syD0wnx/ymMtzz4rkLS6uSbapWSvbpqlRUm6Mwn6mtA== , the other thing you could do with these chips would be 1 Wire

The only thing with any commsis you will need a good way to assign ID's before you connected it to the rest of the network, say 2 banks of 4 switches, 1 for bank, one for cell,or have a process where by a non ID'd unit appeared as ID 0, then the controller assigns the next avaible ID, That or use a 1 wire eeprom on board and use that serial number, small cost but reduces problems on micros that dont have unique ID's
 

daromer

Moderator
Joined
Oct 8, 2016
Messages
5,659
Regarding temp monitoring: Its very easy to just setup 1wire temp-sensors. No matter if you are using a Raspberry or a ESP8266 to transmitt the data or any other mikrocontrollers there are already premade libs for it. Its a single loop and send the temps and you are good to go :)

I did a quick video where i did add some sensors to a charger. Same thing on the packs. My current such setup is an ESP8266 running temp-sensors directly. Both shown on a small screen and sent to a grafing service.

And i have the same when it comes to the Voltages. 2*16 Chan muxes switches in all cells to a 16bit ad and there i have all the voltages :)

Just a tip on how I did it unless you already have it up and running.
 

Nathan

New member
Joined
Dec 5, 2016
Messages
28
Thanks, that is what I was thinking, ive just got the final parts for my 24 cell charger/discharger/datalogger designed, and so from tomorrow I am onto the design of this board.

But now I am also thinking about adding 1 wire sensors for the charger...
 
Top