Internal resistance and DCIR: what the number predicts
Impedance is the parameter that decides how much of a cell's capacity you can actually use under load — and it is the one most often quoted in the wrong form.
A cell's internal resistance is not one number. It is the sum of several mechanisms that respond on different timescales: the electronic resistance of the current collectors and tabs, the ionic resistance of the electrolyte and separator, and the polarisation that builds up at the electrodes as a current is drawn. A measurement that finishes in milliseconds sees the first two; a measurement that holds current for ten seconds sees all three.
ACIR and DCIR are not interchangeable
| ACIR (1 kHz) | DCIR (current pulse) | |
|---|---|---|
| What is applied | Small AC signal, ~1 kHz | Sustained DC current for seconds |
| What it captures | Ohmic resistance only | Ohmic plus polarisation |
| Typical value | Lower | Higher — often 1.5× to 2× the ACIR |
| Why manufacturers publish it | Fast, non-destructive, comparable across labs | Closer to what a pack experiences |
| What it predicts | Relative cell health, grading | Voltage sag and heating under real load |
Both are legitimate measurements. The error is using the ACIR figure to predict DC behaviour, which understates the sag a pack will actually see.
DCIR is derived from the voltage step: apply a known current, measure the voltage change after a defined time, divide. The time matters — a 10-second value is not comparable with a 1-second value, and a data sheet that omits the pulse duration has published a number you cannot reproduce.
What the number costs you in a real pack
The arithmetic is simple enough to run on the back of an envelope, and worth doing before the first prototype:
- A cell with 25 mΩ DCIR at 8 A loses 200 mV.
- Ten in series lose 2.0 V — from a 36 V pack, that is the difference between 3.6 V and 3.4 V per cell under load.
- The same current through a 50 mΩ cell loses 400 mV per cell, twice as much, for identical capacity.
- Heat is I²R: 8 A through 25 mΩ dissipates 1.6 W per cell. Through 50 mΩ, 3.2 W — which is a thermal problem, not just an efficiency one.
That is why two cells with the same capacity can behave completely differently in the same pack. The higher-impedance one reaches the low-voltage cut-off first, the BMS trips, and the user concludes the pack is empty when a meaningful fraction of its capacity is still there.
What moves the figure in service
- Temperature. Cold electrolyte raises resistance sharply; a pack that performs well at 25 °C can lose a large share of its usable power at −10 °C. This is the single largest variable.
- State of charge. Resistance is lowest in the middle of the charge range and rises at both ends, which is why the last few percent of charge deliver much less power.
- Age. Impedance rises with cycling and with calendar time; a pack at end of life may have half again the resistance it had when new.
- Weld and interconnect quality. A poor weld adds resistance outside the cell, and it is invisible in the cell's own data sheet.
How much spread a pack can tolerate
Cells in a series string carry identical current, so a cell with higher resistance dissipates more heat and reaches its cut-off earlier. Both effects accelerate: the hotter cell ages faster, its resistance rises further, and the imbalance grows.
The practical rule used in pack assembly is to group cells by measured capacity and impedance before building a string, and to keep the spread within a narrow band — commonly a few percent, tighter for high-current packs. The exact threshold is a design decision, not a standard, but the direction is not negotiable: the tighter the group, the longer the pack stays balanced.
Measuring it without fooling yourself
- Use four-wire measurement. Contact resistance at the probes can exceed the cell's own resistance if you use two-wire clips.
- Fix the state of charge and the temperature before measuring, and record both. A resistance figure without them is not reproducible.
- Fix the pulse length and the current magnitude. DCIR depends on both.
- Measure the same cell twice to see your own repeatability before comparing cells.
- Compare like with like: an ACIR from one lab and a DCIR from another are not a comparison.
From measurement to pack: the grading step
Impedance measurement only pays for itself if it changes what goes into a pack. In practice that means a grading step between goods-in and assembly, with three outputs: cells that meet the band and are grouped, cells that fall outside the band and go back to stock for lower-current use, and cells that fail a sanity check and are quarantined.
- Measure every cell at a fixed state of charge and a recorded temperature.
- Sort into bands that are narrower than the pack requirement, not equal to it — the band has to absorb measurement repeatability as well as cell variation.
- Build strings from a single band and, where possible, a single batch.
- Record the band on the pack serial label or in the build record, so a later fault can be traced to a cohort.
- Re-measure a sample of finished packs to confirm the assembled resistance is what the cell data predicted.
The last step is the one that catches assembly problems. If the finished pack measures materially higher than the sum of its cells, the difference is in the welds, the busbar or the sense wiring — and it is much cheaper to find at that point than in a returned product.
What impedance does not tell you
A low-impedance cell is not automatically a safe cell, and a high-impedance cell is not automatically a failing one. Impedance says nothing about separator integrity, contamination, or behaviour under internal short. It also does not predict cycle life on its own — a cell can hold its impedance and still lose capacity.
Treat it as one axis: it tells you how much voltage the cell gives up under load, and how much heat it makes doing so. Both matter for design, neither is a safety assessment.
Questions this guide does not answer in passing
Should I design against ACIR or DCIR?
DCIR, where it is available. ACIR is useful for grading and for tracking a cell population over time, but it understates the sag under a sustained load. If only ACIR exists, treat it as a floor rather than the design value.
Why is DCIR higher than ACIR?
Because it includes polarisation as well as ohmic resistance. A sustained current builds up concentration gradients at the electrodes, which an AC signal at 1 kHz does not have time to produce.
Does lower impedance always mean a better cell?
No. Impedance is one axis. A very low-impedance cell may have traded capacity, cycle life or safety margin to get there. It is better for high-current duty and often worse for energy duty.
How often should impedance be checked in production?
At goods-in for every batch, as part of cell grading, and again on a sample basis for stock held long enough to age. Incoming impedance that differs from the previous batch by more than the agreed band is a reason to hold the batch, not to sort around it.
More guides
Keep reading
Want a second opinion on a shortlist?
Send the candidates you are weighing up. We will tell you which ones we can supply with documentation and which we cannot.