First record the decoder through the installation guide, preserve recoverable changes with the programming-track guide, and diagnose stalls before tuning with the troubleshooting guide.
Three control layers that must not be mixed blindly
| Layer | What it controls | Boundary |
|---|---|---|
| Basic curve | When supported, CV2 sets start, CV6 shapes the middle and CV5 limits the top. | Optional support and special zero/maximum behavior vary by decoder; read its manual. |
| Loadable speed table | NMRA CV67–94 provides 28 forward drive entries, normally selected through the decoder’s configuration. | It replaces rather than supplements the basic curve in ordinary implementations. |
| Manufacturer motor controls | Back-EMF, trims, kick start, momentum and decoder-specific algorithms can change measured response. | Names, ranges and interaction are not cross-brand constants. |
What official and first-party sources establish
| Source | Verified point | What it does not prove |
|---|---|---|
| NMRA S-9.2.2 | Defines Vstart, Vhigh, Vmid, forward/reverse trim and the 28-entry speed table as configuration-variable roles. | That every decoder implements every optional feature identically. |
| NCE | Recommends matching start, top and then middle, and separating locomotives before changing them to a shared final address. | A universal target speed or CV value. |
| Digitrax | Documents the three-point curve and notes that table selection and speed-step behavior depend on decoder generation. | That Digitrax values transfer to another decoder family. |
| JMRI DecoderPro | Exposes only the controls represented for the selected decoder and distinguishes basic control from a loadable table. | That software removes the need for a mechanical baseline or measured release. |
| FRA distributed power | Describes real locomotive consists receiving synchronous or independent commands and the operational importance of managing train forces. | A model-railroad CV recipe. |
Four gates before the first CV change
1. Mechanical gate
GO when both locomotives run reliably alone without binding, slip, pickup faults or abnormal heat.
2. Identity gate
GO when exact decoder models, firmware if available and readable baseline values are recorded.
3. Method gate
GO when the selected basic curve, speed table or manufacturer method is explicitly supported.
4. Measurement gate
GO when distance, throttle step, direction, route, load and run state can be repeated.
Exclusive tool 1: measured drift calculator
Enter the same measured distance and each locomotive’s elapsed time. The result reports relative speed difference only; it does not invent a universal pass threshold.
Mismatch fault matrix
| Observed mismatch | Do not assume | Next controlled evidence |
|---|---|---|
| Different start step | That top-speed CV is the remedy. | Clean mechanics, pickup, supported start control, kick-start and load compensation. |
| Matches low, diverges high | That start voltage should be changed. | Supported maximum or upper-table entries and achievable reference speed. |
| Matches forward, not reverse | That one shared curve proves both directions. | Mechanical asymmetry plus supported forward/reverse trim behavior. |
| Matches solo, fights coupled | That timing evidence is sufficient. | Back-EMF interaction, momentum, load sharing, wheel slip and coupler force under duty load. |
| Result changes between passes | That more CV editing improves it. | Warm state, clean wheels/rail, pickup intermittency, binding and timing repeatability. |
Exclusive tool 2: bidirectional consist-release ledger
| Record row | Minimum evidence | Release condition |
|---|---|---|
| Passport | Locomotive, decoder, firmware if known, wheel condition, consist role and chosen reference. | No unidentified decoder or mechanical defect. |
| Baseline | Readable motor CVs, curve-selection state, BEMF, trims, momentum, CV19 and solo behavior. | Recovery values preserved before writing. |
| Measured grid | Low/middle/high step, forward/reverse, distance, elapsed time, repeats and relative gap. | Chosen duty range is stable and reproducible uncoupled. |
| Coupled release | Both orientations, starts/stops, route, load, coupler behavior, slip, sound/function response and heat. | No persistent fighting or unstable response in the intended duty. |
| Reopen list | Hardware, firmware, wheel, lubrication, curve, BEMF, momentum, consist or duty changes. | Any trigger returns the pair to uncoupled measurement. |
Six steps from solo baseline to consist release
1. Stabilize and document both locomotives
Clean and inspect both locomotives, confirm reliable solo running, identify each decoder and record every readable motor, momentum, trim, speed-table and consist setting before changing anything.
2. Choose one achievable reference
Select the slower or operationally suitable locomotive as the reference, define the same test route, direction, throttle steps, warm-up state and load condition, and keep the locomotives uncoupled.
3. Measure the baseline drift
Time repeated passes over the same measured distance at low, middle and high operating steps in both directions, then record elapsed time and separation change rather than judging by eye.
4. Change one supported speed-control layer
Use the exact decoder documentation to choose either its basic curve, loadable speed table or manufacturer-specific method, change one region at a time and preserve the previous values.
5. Repeat uncoupled validation
Rerun the same low, middle and high measurements in both directions, checking starts, stops, stability, heat and any interaction between back-EMF, momentum, trims and the selected curve.
6. Build and release the consist
Configure the consist only after the uncoupled pair is acceptable, test both orientations and load cases, observe coupler compression or stretching, and retain the final roster plus reopen triggers.
Railway-culture answer: one command does not mean identical force
The FRA describes distributed-power locomotives receiving synchronous or independent commands from the lead unit. It also connects distributed power with managing slack action and coupler forces across a train. The modeling analogy is limited but useful: shared commands do not prove matched physical response. Measure the actual units and observe train forces after coupling; prototype MU or radio-control rules do not specify decoder CVs.
Official and first-party sources
Checked 2026-08-27. No generic CV values or universal drift tolerance is claimed.
- NMRA — approved 2012 S-9.2.2 configuration variables
- NMRA — documents under revision
- NCE — locomotive speed matching for consisting
- NCE — matching different decoder manufacturers
- Digitrax — mobile decoder manual
- Digitrax — three-step speed table
- JMRI DecoderPro — basic speed control
- JMRI DecoderPro — loadable speed table
- FRA — locomotive consists and distributed power
FAQ
What is DCC speed matching?
It is adjusting supported decoder motor-control settings so locomotives cover the same distance in nearly the same time at the throttle steps that matter for their shared duty.
Do I have to speed-match identical N scale locomotives?
Measure them first. Even nominally identical locomotives can differ through motor, gearing, lubrication, wheel condition, decoder and production variation, so identical catalog numbers are not proof of matched speed.
Should speed matching happen before consisting?
Yes. Establish and verify each locomotive separately, then run the pair uncoupled at the same commands before creating the final consist. This keeps consist addressing from hiding which decoder needs adjustment.
Are CV2, CV5 and CV6 universal speed-match controls?
They define the NMRA basic three-point speed curve when implemented and selected, but support and behavior vary. Some decoders use a loadable table or manufacturer-specific controls instead.
What are CV67 through CV94?
NMRA S-9.2.2 defines them as the 28-entry forward speed table. The decoder must support and select that table before those values govern motor output.
Can I use both CV2/5/6 and the 28-step speed table together?
Do not assume so. NMRA configuration selects between the basic curve and the loadable table, and JMRI notes that writing one control mode can replace the other. Follow the exact decoder instructions.
Which locomotive should be the speed reference?
Choose an achievable reference, commonly the slower healthy locomotive or one whose operational speed profile must be preserved. A decoder can reduce output, but it cannot guarantee extra mechanical capability.
Should the locomotives be coupled while matching?
No for the measurement phase. Run them uncoupled on the same route or equivalent parallel routes so one locomotive cannot push, pull or mask the other; couple only for the final consist release.
Does back-EMF need to be turned off?
There is no cross-brand rule. Load compensation can interact with a consist differently by decoder and firmware, so use the exact manufacturer guidance and record the chosen setting during both solo and coupled validation.
How close is close enough for a speed match?
There is no universal percentage in the cited standards. Define acceptance for the actual duty, route, load and run time, then confirm no persistent coupler compression, stretching, wheel slip, instability or overheating.
Can DecoderPro automatically make every decoder match?
No. DecoderPro exposes only features represented for the selected decoder and helps manage curves and roster values; mechanical condition, measurement quality, unsupported features and final consist behavior still require verification.
When should a speed-match release be reopened?
Reopen it after changing a locomotive, decoder, firmware, motor, gearing, wheel diameter, lubrication, speed-control mode, back-EMF, momentum, consist method or operating duty, and after new drift, slip, heat or coupler force appears.
