Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Cheap, Different, very DIY
Excellent work. I am curious to see what you end up with for a BMS. I was going to use Arduino myself, but haven't had much time to make a start.
(04-17-2018, 06:36 PM)SuperBrainAK Wrote: YAY someome who cares about efficiency enough to not use TP4056! I am using LM2596 myself for my upcoming 6+6 bay charger + discharger
awesome build!! we need to see some wiring! haha

Wiring is all very ad-hoc at the moment, the system is still transitional from the original 12v lead batteries so the charge reg. is still far away (next job to move it closer for better regulation close to full). I have a side project for a 10-cell analyser that discharges back to the power source (battery assumed) when measuring cells Smile

(04-18-2018, 12:09 PM)Frnandu Martiński Wrote: Jeremy, can you say a bit more about your community? I'de love to know a few more details, how much off-grid (on several aspects) do you manage to get? how many people /houses?

I'm not sure I fully understand you question - there is no grid-power connection to the area so all of our electrical power is generated on-site. We are connected to internet by 3G, so I buy stuff from outside. We try to grow our own food as mmuch as possible, but this is still not much. There is a process of buying more land in progress, new families arriving, and new home projects in progress as well, but at this moment there are 13 dwellings (most are not full dwellings, many are just a bedroom) and our group house, which no-one lives in but everyone can use. Some of these have their own power systems, some have power and data supplied from the group house. Our population and social structure are in flux right now, but I last counted 24 people living here most of the time (including 7 children). Does this help?

(04-16-2018, 09:34 PM)DK100 Wrote: Nice writeup on your system and methods.

Ha! I am in Denmark this week! nice and sunny, unlike our mountain in italy...
(04-18-2018, 07:27 PM)Turtle Hermit Wrote: Here is the video about LM2596 CC-CV :

The full screen schematic (reverse engineering) is at the very end of the video.

To Jeremy : Apologies for hijacking your thread  Blush

you've hardly hijacked my thread ... the mod you suggest for end-of-charge shutdown would be just a resistor from the led drive (from the op-amp output?) to the nfb pin of the 2596, but I wonder why it is needed? instead of continuing maintaining 4.20V you allow the cell to float and internally equalise. is there some benefit to this?

(04-19-2018, 02:21 AM)Geek Wrote: Excellent work. I am curious to see what you end up with for a BMS. I was going to use Arduino myself, but haven't had much time to make a start.

The design is quite simple, the stm32 has a 12-bit dual 1Mhz ADC which I will feed with as many 5V-reduced-to-3.3V differential inputs as needed. It can sample these and send the results via serial (one-way) to the raspberryPI for logging and presentation as a rapidly updating webpage for viewing realtime data over our local network. I plan to have ~7 lines of control (2 PWM and 5 basic digital) for controlling when some things turn off at low levels and on again as it recharges (such as the fridge inverter, the 35V and 50V boosters etc). one of the PWM's would be for a heater loop, to stop charging from occurring at low temperatures by heating the cells with that energy till they can be charged. The balancing is being done by a 7-primary transformer operating single-ended at about 80% duty so that will want to be turned on and off according to balance state and load. This type of balancing circuit operates as if the 7 series cells were instead connected in parallel, so it becomes more difficult to monitor the amount of correction that is happening and detect faults in the cell packs - but with accurate cell voltage measurement the current that is flowing through from each cell can be calculated from the total resistance through each coil of the transformer (including mosfet and connecting wire). The second PWM operates the balancing to keep it's total dissipation low enough if the battery is under heavy load. I plan that all settings can only be made via USB so a direct presence is needed to make operational changes, making it unhackable. I haven't yet decided whether to make the serial connection to the rPI long and current-mode, or keep it short with the rPi in the same box and move the network connections (probably the latter). For the current sensing I'm planning on using ACS712 based hall-effect transducers in various places in the system so I can see what's going on. Where currents greater than the 20-amp range of the sensor are expected (everywhere) series-parallel arrangements of 1mΩ resistors can extend the range up to hundreds of amps. They have a 5V output range.

I posted a note about the multiplexed differential inputs on an stm32duino forum here.

(04-19-2018, 02:21 AM)Geek Wrote: Excellent work. I am curious to see what you end up with for a BMS. I was going to use Arduino myself, but haven't had much time to make a start.

So I know I've been very quiet for a long time - but I have been slowly working on battery things alongside the rest of what I must do...

in May I took the plunge and learned KiCad to get circuit boards made for my BMS and other electronic ideas. The BMS and some ridiculously high-power boost regulator boards (>2000W) were the result, but still after 4 months I've not built it, mostly because of the house I work in undergoing major rebuilding but also partly due to not having a firm idea of how the BMS will achieve balancing (it's primary purpose). It's not been a problem, my batteries are staying balanced without any work, the only thing that has been an issue is occasionally fuse-wires get broken (mechanically, not electrically) and then one layer will start cycling wider. The biggest issue is the location of the battery, there are rats living in the wall behind who come out from between the stones and have shat in between the cells. I'm fairly sure they're responsible for the broken fuses too...

I've been meaning to write an update on the BMS/balancing for a long time, but I wanted some positive progress or at least to have worked out why there is the issue with the design I described before (7-winding transformer) and come up with a next step...

The 7-winding transformer basically works, but with a cumulative offset of about 12mV per cell meaning the top cell of the 7 ends up "balancing" to 0.1V higher than the bottom one. The reason for this seems to be the one-sided use of the transformer (the circuit pulls it magnetically in only one direction, letting it reset back to zero between cycles). While switching to a push-pull circuit might work (doubling the cost and complexity in the process), while thinking about it I decided maybe a slightly different approach might be better, and came up with what I call "Boost Balancing".

There are two potential ways to make this work, a "cheap-and-dirty" self-balancing way, and a BMS-integrated cell-by-cell-controlled way. Both are based on using an isolated boost regulator to transfer energy from the whole battery to a single cell. The cheap way has 8 windings on the inductor core - one primary driven by the boost driver and 7 secondaries each connected to a single cell via a schottkey diode. Energy is taken out of the whole battery into the inductor. If a cell is lower than the rest then the flyback dumps that energy into that cell. The lowest cell gets the most energy, and as they equalise then the energy gets distributed to all of them. There may be issues though with component tolerances (differences in voltage drop over the diodes making different cell voltages) and thermal overshoot (the diode that is getting the most current gets hot and a hot diode has a lower voltage drop which makes the voltage of that cell seems lower to the circuit). The more expensive way is basically the same idea but there are separate inductors and boost drivers for each cell, activated when the BMS microcontroller deems it necessary.

In both cases there is no direct way to bring down a high cell, which is why I call it Boost Balancing. Boosting all the other cells would bring down the high cell, since energy is being taken from the whole battery and not given back to the high cell.

I will let you all know what happens, but I'll need to draw up some new circuit boards and delivery could take a while...
The 2596 schematic is missing the shut down pin.  This schematic is close to what I use for a pseudo MPPT charge controller that compares the panel voltage and ramps down current when panel voltage drops below a fixed voltage.  The reference and op amp can be eliminated by using a TL431 to monitor voltage. What I really wanted to talk about is how fast that shut down pin is and I use it to control the chip instead of faking the the feedback pin.  That pin is easy to lift up on these boards.
(04-18-2018, 11:51 AM)Turtle Hermit Wrote:
The LM2596 is a voltage regulator. How would you use that as a replacement for the TP4056? Or are you controlling the CC/CV with an external mcu?

No mcu required.

What you need is shunt resistor in the (-) side, a simple voltage reference with a pot and a comparator which output highers (not sure if this verb really exist but I hope you get what i mean) the feedback input of the LM2596 when then current is higher than the set limit.
There is even a way to detect when reaching a low current limit for end of charge.

I think the word you were looking for is raises. Smile
And thank you for the info on the LM2596.
Please excuse any errors introduced by autokorrekt.
All other errors are solely my own work.

Forum Jump:

Users browsing this thread: 1 Guest(s)