Method group NG-CYC
Cycling and lifetime
Four standing methods cover almost every cycling request we receive. All of them run on the same channel hardware and in the same chambers, which is what makes results from different jobs comparable at all.
- Channels
- 96 independent channels in 12 units, each channel ±5 A over 0–5 V
- Current resolution
- 50 µA on the low range, settling within 8 ms of a step
- Voltage accuracy
- ±2 mV at the cell terminals, four-wire sensing on every channel
- Logging
- Fixed interval from 50 ms to 60 s, plus event logging on current change and on 5 mV voltage delta
- Chambers
- Six chambers, −40 °C to +90 °C, stability ±0.5 K at the cell holder measured with an independent logger
- Holders
- Coin 2032 and 2016, pouch up to 5 Ah under 0.2 MPa stack pressure, cylindrical 18650 and 21700
Each chamber carries its own calibrated temperature logger that is not part of the chamber controller. The report quotes the logger, not the setpoint, because those two numbers disagree more often than chamber datasheets suggest.
Constant-current cycling at one temperature
The default request. A cell is formed, then cycled between fixed voltage limits until it reaches a retention threshold or a cycle count, whichever comes first.
| Parameter | Default | Range we will run |
|---|---|---|
| Voltage window | 3.0–4.2 V | 1.5–4.9 V, stated by the client |
| Charge | CCCV, C/2, cut at C/20 | C/20 to 2C |
| Discharge | CC, C/2 | C/20 to 5C |
| Temperature | 25 °C | −20 to +60 °C |
| Formation | 3 cycles at C/20 | Client protocol accepted |
| Reference test | every 50 cycles | 25, 50 or 100 cycles |
| Stop criterion | 80 % retention | 70–90 %, or a fixed cycle count |
| Replicates | 3 cells | Minimum 3, we decline fewer |
The reference performance test interrupts cycling for a C/20 capacity measurement and a short impedance sweep. It costs about four hours of channel time per occurrence and is the only way to tell a genuine capacity loss from a rate limitation that appears as one.
We will not run a single cell and call the result a lifetime. Channel-to-channel spread on nominally identical coin cells is routinely 1–3 % of capacity; with one cell there is no way to know which side of that spread you are looking at. The methodology page gives our repeatability figures.
Rate capability
A short, cheap screen that answers whether a material is transport limited before anybody commits a chamber for six weeks. Five rates, three cycles each, then a return to the lowest rate to show whether the loss was reversible.
| Step | Discharge rate | Cycles | Reported |
|---|---|---|---|
| 1 | C/10 | 3 | Reference capacity, mAh g⁻¹ |
| 2 | C/2 | 3 | Retained fraction of step 1 |
| 3 | 1C | 3 | Retained fraction, mean voltage |
| 4 | 2C | 3 | Retained fraction, mean voltage |
| 5 | 5C | 3 | Retained fraction, polarisation |
| 6 | C/10 | 3 | Recovery against step 1 |
Charging stays at C/2 throughout so that the discharge comparison is not confounded by a different state of charge at the start of each step. If the question concerns fast charge rather than fast discharge we invert the protocol and say so on the method sheet.
Four-temperature lifetime matrix
Twelve cells, three per temperature, at 0, 25, 45 and 60 °C. This is the method behind the fade curves on the methodology page and the only one of ours that supports an ageing model with any honesty.
- All twelve cells are formed together in one chamber at 25 °C, so formation is not a variable between temperature groups.
- Reference tests at 25 °C every 100 cycles for every group, including the group that lives at 60 °C. Comparing a hot cell with itself at its own temperature hides most of what you paid for.
- Cells are redistributed across units, not kept as a block on one unit, so that a drifting channel cannot masquerade as a temperature effect.
- Raw data is delivered every two weeks during the run, not only at the end.
Expect 8 weeks if the stop criterion is a cycle count and up to 14 if it is a retention threshold at 0 °C, where cells are slow and the run is long.
Calendar ageing
Cells are brought to a stated state of charge, left at temperature, and checked on a schedule. Nothing cycles, which means the method occupies chamber space rather than channel time and is priced differently.
| Parameter | Default |
|---|---|
| States of charge | 30 %, 60 %, 100 % |
| Temperatures | 25 °C and 45 °C |
| Check interval | every 14 days, capacity and impedance |
| Open-circuit logging | hourly, continuous |
| Minimum duration | 12 weeks |
Between checks the cells rest on open circuit and the voltage is logged through a high-impedance input, so the measurement itself does not drain them. A 26-week run is the longest we take on a fixed quotation; beyond that we bill chamber time monthly.
Every cycling job returns the same four artefacts, described in full under reports:
- Raw per-channel time series as CSV, one file per cell, with the column dictionary in the header.
- A per-cycle summary table: charge and discharge capacity, coulombic efficiency, mean voltages, energy, resistance from the reference test.
- A fade table at 50-cycle intervals with median and range across replicates.
- A signed method sheet stating the protocol as executed, including interruptions and their cause.
Interruptions are logged and reported rather than smoothed away. A chamber door opened for a neighbouring job, a 40-minute power event in February 2026, a channel swapped mid-run — all of it appears in the method sheet with a timestamp.