Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
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 power of 10kW

2) Current Phase: 

Battery Capacity + IR/SoC Testing

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

4) Sub Project:

4.1) Battery Capacity Tester

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

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 ( Master/Slave BMS based on LTC6811 
Optimized IO and component substitution


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

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..

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) : 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
humbleONE likes this post
Sony VTC5(a) and some VTC6 cells and Samsung 20Q/25R
12s is 43.2v nominal@3.6


later floyd
(06-02-2020, 02:07 AM)floydR Wrote: Sony VTC5(a) and some VTC6 cells and Samsung 20Q/25R
12s is 43.2v nominal@3.6


later floyd
My bad Smile 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
floydR likes this post
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.

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

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.

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).

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

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

There is also a Node-Red javascript based bar-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. 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.

humbleONE, Wolf, Redpacket like this post
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.

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

Get the idea?

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 Smile but does take hours to print and the gears aren’t as strong or accurate as the laser cut version.

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

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

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

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?)
Running off solar, DIY & electronics fan :-)
Ye I got rid of the tp4056 for many of those reasons.

Latest laser cut and 3D printed 4 wire slot design.

I’ll be posting all cad files and output on GitHub

Forum Jump:

Users browsing this thread: 1 Guest(s)