ROTKEE Sensors on PicoScope

Read this first: three kinds of number

A scope only shows volts. Whether those volts mean something depends on what the sensor maker published, so every figure on this page says where it came from.

• Published by ROTKEE — from ROTKEE's own product pages. Only the PS16 has a published scale (checked against ROTKEE's store page in September 2026).
• WaveAssist convention — settings we chose so two technicians' captures differ by the engine and not by the scope. Not a ROTKEE specification.
• Not known — nobody has published it, so we do not guess. A number you invent here will read as confident and be wrong.

ROTKEE is a third-party sensor maker. Figures below come from its public product pages and from captures we have run.

PS16 — 16 bar in-cylinder pressure transducer

The PS16 is the one ROTKEE sensor with a real scale, so it can be set up to read pounds per square inch directly in PicoScope.

Why 49.3 and -14.5

ROTKEE publishes the conversion for other scopes: bar = 3.4 × volts - 1. At 0 V that is -1 bar and at 5 V it is 16 bar, which matches the stated range. One bar is 14.504 PSI, so multiply both numbers: 3.4 × 14.504 = 49.3 and -1 × 14.504 = -14.5. That is PicoScope's gradient and offset for PSI. Use gradient 3.4 and offset -1 if you would rather read bar.

The published sensitivity says the same thing three ways: 294 mV per bar (1 / 0.294 = 3.40), 20.3 mV per PSI (1 / 0.0203 = 49.3), and 17 bar × 294 mV = 4.998 V for the 0 to 5 V output. Across the whole 0 to 5 V range the 49.3 / -14.5 line stays within 0.06 PSI of ROTKEE's own formula. Check three points: 0.294 V reads 0 PSI, 1 V reads 34.8 PSI, and 5 V reads 232 PSI.

Check it before you trust it

• With the sensor open to the air and powered, the channel should read about 0 PSI (about 0.29 V without the probe). Reading about 14.5 PSI high means the offset was left out.
• A flat trace can just mean the battery clips were off. The PS16 needs its 11 to 25 V.
• A flat-topped compression peak is the 232 PSI ceiling clipping, not a plateau.
• Use it plug-out, cranking or running with that cylinder's fuel and spark disabled. Combustion pressure in a firing cylinder under load exceeds the 348 PSI overload.

A file exported with this probe active is already in PSI. WaveAssist reads the unit row and does not convert it a second time.

PDS and VS — dynamic sensors

Both are piezo sensors. They answer to a change in pressure or a knock, and give nothing for a steady one. ROTKEE publishes an output span for each (PDS -5 to +5 V, VS -2 to +2 V) but no sensitivity for either, and the output moves with the hose, the pipe and where you mount it. What they give you is timing, shape, event counts and cylinder-to-cylinder ratios inside one capture.

ROTKEE publishes the output span and, for the PDS, the pressure range and overload (150 kPa, 21.8 PSI). Every other cell is WaveAssist convention, chosen from real captures (a relative pulse sensor read -1.17 to +0.96 V at idle on one test car) and from the sample-rate table for the test it feeds.

Can the PDS read PSI?

ROTKEE's PDS page lists a pressure range of -100 to +100 kPa (-14.5 to 14.5 PSI), an output of -5 to +5 V, and an overload of 150 kPa (21.8 PSI). It says the sensor reads only dynamic pulsation, never static pressure. It publishes no sensitivity.

If the full pressure range maps onto the full output, the nominal scale is 14.5 PSI / 5 V = 2.9 PSI per volt: a custom probe with Linear scaling, gradient 2.9, offset 0, unit PSI, AC coupled. That is arithmetic on ROTKEE's stated ranges, not a ROTKEE sensitivity, and it has not been checked against a reference here. Use it to compare cylinders and read pulse shape, not as a gauge reading. Until it is checked, the volts-only probe above is the safe default, and it changes nothing about the comparison because every verdict is a ratio inside one capture.

What to expect

• Amplitude is not comparable between cars, or between lines on one car. PDS output changes with tube diameter (rated 4 to 7 mm), material and insertion depth. VS output changes with pipe material.
• Do not zoom the input to fill the screen. On a real fuel-rail mount the VS used about 6% of its range (-0.125 to +0.106 V). A tighter range clips injector bursts and the ratios lose their meaning.
• If a sample reaches the input rail, go up one range and capture again. A clipped pulse has lost the amplitude the comparison depends on.

The mistake that puts fake PSI in a file

Never leave the PS16 probe (y = 49.3x - 14.5) on a PDS, VS or any other relative sensor. The channel exports numbers that look like PSI, read about -72 to +25, and nothing below -14.7 PSI gauge can exist. If you see a file like that, put the channel on a volts-only probe and capture again.

ROTKEE VS — how it is used in the app

The VS is a passive piezo sensor (output -2 to +2 V, BNC, no power supply). ROTKEE publishes no sensitivity, so it never gives a pressure or a g figure. What it gives is when an injector event happens and whether it happened at all. In WaveAssist it is a confirming channel: it backs up a finding from another test and is never the only source of a verdict.

Mount 1 — the injector's high-pressure pipe (common-rail diesel, direct injection)

• Screw the sensor onto one cylinder's high-pressure injector pipe. It reads that injector's opening: a burst that starts at the injection event and rings down after it. The mounted cylinder's burst is the strongest; the others show smaller ones because the sensor also hears its neighbours.
• Trigger on a low-amp clamp around that injector's wire (rising edge), so every window is anchored to the electrical command. A four-cylinder review used ±2 V on the VS, ±20 A on the clamp, and 50 to 100 ms/div to see all cylinders, 4 ms/div to see one event.
• The test: the command is present but the VS shows no burst in that cylinder's own window while the other cylinders' bursts remain. That is an injector that is not opening. In the review, unplugging one injector made its burst disappear and left the rest.
• Weak is not absent. Cross-pickup means a nearby burst can leak into the window. Only present-or-absent has been shown; burst amplitude has no established meaning.
• Two sensors are for comparison: one on a known-good pipe, one on the suspect.
• This comes from one practitioner review (a VW common-rail diesel, a healthy engine with an injector unplugged), not from ROTKEE and not from a real failed injector.

Mount 2 — the fuel feed line on a port-injected engine

• Mount it on the metal feed line or tee close to the rail, no nearer than 2 to 3 cm from a fitting. On a low-pressure rail the sensor mostly hears the hydraulic knock of each injector closing: one burst per injection window.
• Loud and quiet windows are normal. On a V engine, injectors on the rail the feed line serves ring loud and injectors on the far rail ring quiet. Walking the firing order gives a loud/quiet pattern per 720°, and a healthy engine's pattern matches it at exactly one starting point. That one match names every window and proves the bursts are injector events, not the fuel pump.
• Measured on a 2013 V8 at idle (500 kS/s, 139 cycles): loud windows about 191 mV peak-to-peak, quiet about 113 mV. Those are that car and that mount, not a target for yours.
• What the math cannot catch: if you tell the app the wrong rail is the sensor's side, an engine like the GM LS V8 still matches and the windows get the wrong companion cylinders. Settle it physically: move the sensor to the other rail's feed and the loud and quiet groups must swap, or unplug one injector.
• Not yet shown on real data: a burst that goes absent on a dead injector. The measurement so far is a healthy engine.

Mount 3 — the rail body

The events are there but weaker. There is no source for interpreting a rail-body mount, so the app refuses it rather than guess.

In the Waveform Lab

Port-injected engine, sensor on the feed line: declare the VS channel as Vibration sensor and a coil or injector current clamp on one cylinder as Ignition / sync, pick the engine, then declare the mount and that the engine is port-injected. The Fuel Rail Vibration test reports, window by window, whether a burst is present. It measures amplitude within a group for follow-up but does not grade it. It refuses a direct-injected engine, and a sensor on the rail body stays measured with no verdict. Capture at 20 kS/s or faster (100 kS/s recommended), with each waveform holding whole engine cycles. The app also flags clipping when the trace reaches 99% of the ±2 V range.

Common-rail diesel or direct injection, sensor on the pipe: that mount is read on the Pulse Multi screen, where you declare the sensor as a vibration channel with its mount, injection type and which cylinder's pipe it is on.

ROTKEE's own reading is inverted from intuition: a lower pulse tip means more fuel delivered. Amplitude also depends on pipe material (rubber lower, plastic higher), so it never compares across cars or across lines.

What we do not know yet

• A verified PSI scale for the PDS, or any g figure for the VS. ROTKEE publishes neither.
• How a real failed injector looks on the VS. Both sources so far used healthy engines.
• PicoScope menu names. Custom probes live under the Probes menu, and the fields are scaling type, gradient, offset and unit. Labels can differ slightly between PicoScope 7 versions.

Frequently asked

Upload the capture on the Waveform Lab and declare what each channel carries. The app reads the unit your file declares and never invents one.