FoxWall-HV

Araknid

Araknid

Member
Joined
Jun 1, 2020
Messages
54
1) Project Goal

Power wall built with Sony 20Q/25R,Samsung VTC5(a) and some VTC6 cells harvested from drill battery packs. built into a high voltage stack to power a Solax Hybrid X3-8.0T inverter for a peak powerof 10kW

2) Current Phase:

Battery Capacity + IR/SoCTesting

3) ProjectSpecification:
Voltage: 180-240v[size=small](min)/360-470v(max)[/size]
Current: 25A (30A peak, 60s, fused 50A)
Capacity: 30kWh
Power: 3-5kW[size=small][size=small][size=small](min)[/size][size=small][/size][size=small]/[/size][/size]7-10kW(max)[/size]
Configuration: 12s40p5s(min) /12s40p10s (max)
Battery Management: foxBMS (foxBMS.org; Open Source BMS); PCB re-work to optimise IO
Inverter: Solax Hybrid X3-8.0T
Batteries:[size=small]INR18650-25R,[size=small]INR18650-20Q,[size=small]US18650VTC6,[size=small]US18650VTC5/5a,[size=small]US18650VTC3[/size][/size][/size][/size][/size]
Source:Powertool battery packs. Single source, stable supply and quality

4) SubProject:

4.1) Battery Capacity Tester

Arduino ATMEGA2560 with fixed resistor load (~1.8 to 1.4A)

https://github.com/HSBNE/atmega-18650-discharger/

Phase 1: functional updates complete. Calibrated sockets, time accurate and functional data logging system with barcode scanning for tracking individual cells and their test data. Documenting the phase 1 project before building phase 2.

Phase 2: will be a new online IR measurements over SoC to log both voltage, current, temp, IR over the effective 3.9 to 3.0 OCV "nominal" capacity range

4.2) Battery Charger

30-bay TP4056 + PC power supply in laser cut perspex frame. Simple, effective.

Phase 2:Build new 30-bay unit based on Tp5100. Buck converter for improved energy efficiency from the 12V ATX PSU Rail and faster charge times using 2A over ~0.5-1A (depending on how the wind blows... Tp4056)

4.3)Battery Management System

foxBMS based (foxBMS.org) Master/Slave BMS based on LTC6811
Optimized IO and component substitution
https://github.com/ArakniD/foxbms


image_fjmfwu.jpg


image_xkeyag.jpg


PCB with reduced IO as the source project supports far more configuration options than I need. Substituted some parts too. I promise to detail what i've substituted and any new features added..

4.4) picoReflow raspberyPi based PCB Oven

https://github.com/HSBNE/picoReflow

I needed a PCB over for this and other projects...

5) Research and Development:

5.1) 4-wire 18650 holder

Pivot and self-latching to enable testing of naked cells, safely. And enables IR measurements over SoC. IR is not a static measurement, and it depends on temperature, SoC, time since rest. Many factors. I have seen the effects of high IR at end of SoC through a number of discharge cycles on otherwise 'OK' cells. I would prefer to bin these than include them in my packs.

concept drawing.. Initial thoughts..


image_uwdewb.jpg


5.2) Lithium Battery cell off-gassing detection and fire Prevention

ccs6811 based chip to detect gasses from a potential lithium fire.. more details later

6) Reference Material:

6.1) Analysis of the Effect of the Variable Charging Current Control Method on Cycle Life of Li-ion Batteries
6.2)Timescale and SoC affects on IR measurement
6.3)On-line Electrochemical Impedance Spectroscopy for Lithium-Ion Battery Systems
6.4) foxBMS.org : OpenSource Battery Management System
6.5)A Comprehensive Approach for the Clustering of Similar-Performance Cells for the Design of a Lithium-Ion Battery Module for Electric Vehicles
 
Sony VTC5(a) and some VTC6 cells and Samsung 20Q/25R
12s is 43.2v nominal@3.6

12s40p9s?

later floyd
 
floydR said:
Sony VTC5(a) and some VTC6 cells and Samsung 20Q/25R
12s is 43.2v nominal@3.6

12s40p9s?

later floyd
Click to expand...
My bad:)Posting in haste. max voltage 500V. Keeping cells at 3.9v for longevity.

So the peak stack count is; 12s40p10s

Pack startup count is 12s40p5s.

Likely only build to 12s40p8s
 
4.1) Battery Tester: Phase 1

The simplest dis-charger configuration with emphasis placed on time keeping, mV accuracy and per slot calibration and temperature drift compensation. Voltage is within 1mV and current is within 4mA over the full discharge cycle.

Data is recorded in influxDB enabled by a complex node-red flow to filter, dashboard and log discharge data plus result summary.

Careful placement of zero-current ground connection at each negative terminal enabled minimal mV drop error caused by neighboring slots.


image_ovrici.jpg


This setup is limited in that it cannot measure IR during discharge because it cannot measure the instantaneous voltage change when current without the socket resistance dominating the signal. Using an external single-cell tester I have validated a method to capture IR during discharge to datalog IR change over SoC per battery.

4wire enables sampling of all components of internal resistance; R0, Rcf, Rct. Turning the FET OFF during discharge allows measuring the instantaneous change in V and I, thus dV/dI = R0 at 0-0.1s, Rcf at 2s and Rct at ~ 5s (Slope analysis required, time in seconds is just an estimate)

Below is a proof-of-concept output from an independant 4-wire discharger


image_nwefqu.jpg


An android phone app layout for TouchOSC which shows progress and results per slot, and a diagnostic tab to see slot voltage drop incase of base cell connection.


image_bemtey.jpg


An influxDB dashboard shows current discharge curves and results as they come in, recording mAh, mWh, Initial OCV, runtime, barcode, slot and unique ID(DTS + SLOT).


image_msehkr.jpg


Node-Red Dashboard-ui "Slot Detail" to enable diagnostics and calibration of each individual slot.


image_hclqvm.jpg


A basic fixed resistor capacity tester with some added Vref chips I had in my draw, rtc and wifi bridge via an esp32


image_aljvmi.jpg


There is also a Node-Red javascript basedbar-code scanner and slot management view to allocate discharge curve and result data for tracking individual batteries through the whole testing process, as this may take weeks.. and you need data to make good decisions about which battery goes where, and perhaps where each battery has gone. Tracking and tracing all cells.

5.2) Lithium Battery cell off-gassing detection and fire Prevention

Module used:
CCS811+SI7021+BMP280 Carbon Dioxide Temperature Humidity Gas pressure 3in1 module

Concept: small air pump samples air from across the entire pack via little tubes. Sample and alert of high VoC, and turn off or shutdown pack when VoC levels go high.

This is in addition to temperature and voltage safety levels.

foxBMS.org lead was involved with developing the CCS811 sensor FWIW...

only done pre-liminary tests to verify the sensor detects levels of gas... it does.. I need to quantify how small and how sensitive it is.

References:

 

Attachments

  • OffGas.MasterJohnsplass2017.pdf
    2.4 MB · Views: 247
  • Offset.Lithium-Ion-Battery-Monitoring-System-White-Paper.pdf
    524.5 KB · Views: 252
Updating 4-wire battery tester 18650 holder. Mech parts in a laser cut version for high speed reproduction using terminals from Ali express.

Two pivoting parts push the battery against the terminal and lock into place.


image_uodhlw.jpg



image_wqkawi.jpg


Negative terminal at the back end so one can load naked cells and not worry about short circuits.


image_omezsj.jpg


Get the idea?


image_hbbuut.jpg


A small slot under each cell is the perfect position for an NTC sensor , giving you 4 wire measurement in an easy load slot with temperature.

Da daaaa

Also an earlier 3D printer version. Does look super cool :) but does take hours to print and the gears arent as strong or accurate as the laser cut version.


image_lxzegb.jpg


See the clearance for the terminal? Allows easy removal before engaging

image_ioyfwu.jpg
 
Update to the laser cut 4 wire holder design to resolve the battery heatshrink catching in the cross bar. Changed it to a U shape to match the 3D printed version.


image_ssawss.jpg


Now with ample clearance and retention in the open position allowing a smooth insertion with one hand.


image_virznv.jpg


The crossbar version also snapped in the middle, not enough material.


image_jgceyi.jpg
 
Great work, interesting project :)
With the TP4056 modules watch out for heating & strange performance when powered from >5VDC, it seems the fakes don't like higher input voltages....
I had a few & was trying to power from 9VDC & 3 out of 4 got hot & let the magic smoke out...
Have ditched TP4056's in favour of the TP5100's or TP5000's (seems there's less fakes?)
 
New 18650 tester in development. 10A bidirectional sync buck boost , 8 slots.
 

Attachments

  • image.jpg
    image.jpg
    3 MB · Views: 223
New laser cut 4wire mech slot holder mech. This encases the 8 slot 10A tester.

image.jpg
 
Cell holder from above when pushed in. 4 wire mech holds cell in place and allows easy insertion and removal.

not fiddly like every other 4 wire holder I’ve seen...
 

Attachments

  • image.jpg
    image.jpg
    1.5 MB · Views: 146
Araknid said:
Cell holder from above when pushed in. 4 wire mech holds cell in place and allows easy insertion and removal.

not fiddly like every other 4 wire holder I’ve seen...
Click to expand...
Coming together looking good. (y) (y) (y)
 
Envious of all that laser cut goodness :)
Are you using an external ADC for voltage detection?
Or oversampling the onboard ADC for higher resolution.

How is the current being measured?
 
Current is measured by a 5mOhm shunt in the high side, amplified and offset to be +-1.20v at Vcm=1.25v.

INA821 with an active 4th order filter to remove the inductor ripple. Plus side chain and with a higher bandwidth for hardware comparator OCP in both charge and discharge.

voltage also has dual paths, both 4-wire and board shunt voltage for CV loop and regulator startup voltage detection.

reverse polarity protection to keep my caps happy.

As for adc it’s using simultaneous sampling 24b sigma delta converters churning out 32ksps for both CC and CC+CV control loops.

It’s certainly not cheap :)

It should in theory be capable of 1kHz AC impedance measurement also. At 80A... but hey, we’ll see about that!
 
nz_lifer said:
Envious of all that laser cut goodness :)
Are you using an external ADC for voltage detection?
Or oversampling the onboard ADC for higher resolution.

How is the current being measured?
Click to expand...
Also yes, laser cutter go brrrr brrrr. It does look pretty. It’ll be even prettier with the RGBW LEDs underneath each slot for indication of slot state.

waiting on some cool m3 anodised standoffs for each slot in red and black to complement the + and - ends.
 
Interesting article about the VCC charging algorithm. Cell longevity is definitely top of mind for me and probably most people.
I'll probably be diving into BMS programming and CanBus protocol in the next few months. Should be a steep learning curve with the coding experience I have!
Once you stop learning, you start dying!
 
Solax battery canbus communications between master t58 and solax x3 hybrid inverter.

902B5EF0-365E-4249-8F33-780FBA359B71.jpeg
1A8EEBB4-2DAF-4C49-9957-57DABBC67DC1.jpeg
Will provide a canbus message format soon :)
 
Araknid said:
Sorry for late reply :0 cost about 1.2k with all dev boards etc. still coding the sequencing and tuning thr loops. New parts to fit arrived today.
Click to expand...
Is that how much for the tester or the BMS or everything?
How much is the FoxBMS anyways?
 
You must log in or register to reply here.
Back
Top