RX-8 Rotary Compression Testing
Mazda limits, test procedure, waveform timing and a compact result sheet for the two-rotor 13B-MSP.
On this page
Quick limits
| Measurement at 250 rpm | Limit |
|---|---|
| Standard | 830 kPa / 8.30 bar / 120.4 psi |
| Minimum | 680 kPa / 6.80 bar / 98.6 psi |
| Maximum spread across one rotor | 150 kPa / 1.50 bar / 21.8 psi |
| Maximum front-to-rear difference | 100 kPa / 1.00 bar / 14.5 psi |
Don’t compare an uncorrected low-speed reading with the 250 rpm limits. Record all three events, the actual cranking speed, raw and corrected values, and the test conditions.
Mazda’s rule: if compression is at the minimum or lower, or the difference between chambers or between rotors is at the maximum or higher, the engine needs an overhaul.
Test procedure
- Check the battery is fully charged (Mazda’s wording; a load test is good practice). Warm the engine, shut it down and wait about ten minutes for exhaust temperature to fall.
- Remove one leading or trailing plug from each rotor. The tester can go in either plug hole; keep the other holes sealed and label the removed leads.
- Disconnect the eccentric-shaft position-sensor connector. Mazda uses this to inhibit both injection and ignition.
- Fit the rotary compression transducer to the chosen front-rotor plug hole. Mazda specifies a plug-hole adaptor and tester as special service tools. My view, not Mazda’s, is that a piston-engine gauge with a one-way valve is the wrong instrument because it keeps only the highest event.
- Hold the accelerator fully open and crank for 5-10 seconds, long enough for stable repeating events.
- Record front A/B/C peak pressure and the actual rpm. Repeat once if timing or amplitude is unstable.
- Move the transducer to the rear rotor, keeping battery state and temperature as consistent as possible, and repeat.
- Correct each result to 250 rpm with Mazda’s correction graph, or a tester implementation validated against it. Don’t use an invented universal linear factor.
- Compare corrected pressure, within-rotor spread, front-to-rear difference and repeatability.
- Refit the plugs at 12.8-17.7 N·m, reconnect the eccentric-shaft sensor and put the leads back in their labelled positions.
Waveform explanation
A, B and C are the three repeating waveform groups. Don’t assign them to physical rotor faces unless an external phase reference identifies crank/rotor position.
For a time interval in seconds:
RPM = 60 / adjacent peak interval
RPM = 180 / same-face interval
At 250 rpm, adjacent groups are 0.240 s apart and the same group repeats 0.720 s later. One consistently weak group suggests a local sealing problem. Three similarly low groups more often mean general wear, low cranking speed, temperature or test setup. A single noisy sample isn’t a diagnosis; repeat it under matched conditions.
Result sheet
| Rotor | Raw rpm | A raw | B raw | C raw | A @ 250 | B @ 250 | C @ 250 | Spread | Notes |
|---|---|---|---|---|---|---|---|---|---|
| Front | |||||||||
| Rear |
Record units, coolant temperature, battery voltage, tester and date with the table. Compare the two rotor means only after correcting all values the same way.
Devlin Rotary tester
A functional prototype under development. It uses 0-300 psi pressure transducers, records pressure through time, finds valid peaks, calculates cranking rpm, groups the three repeating events, reports kPa/bar/psi and corrects to 250 rpm. Its schematic, BOM, PCB, filter constants and firmware aren’t published while the design is being validated.
References
Technical review: