Hysen Labspower-rail-probe
Low-noise oscilloscope power rail probe
Power Rail Probe: measuring ripple at the millivolt level
An open hardware probe that attaches to a power rail up to 20V, applies a vertical offset, and lets a normal oscilloscope see ripple at the millivolt level.
The idea in one line
The probe is open hardware for measuring millivolt level ripple on a power rail up to 20V using a normal oscilloscope, by applying a vertical offset. The README notes it is loosely based on an existing writeup, and that the open design costs less than a tenth of the price of a comparable commercial probe from Tektronix. A sample measurement shows 60 mV peak to peak ripple on a 20V rail captured with an inexpensive Picoscope.
How the offset trick works
You connect the probe directly to the power rail of interest, either through a 1X scope probe or a direct cable. AC passes while an adjustable offset applies to the DC voltage. The oscilloscope then sees 0V DC, which lets you view ripple and transients at a 1 mV per division setting instead of 5 V per division. The README calls it effectively a much more powerful DC offset function for your scope.
The v1.0.1 specifications
The spec sheet for version 1.0.1 lists a maximum offset of plus or minus 20V, input impedance of 50 kiloohms from 0 to 30 kHz and 50 ohms from 30 kHz to 1 GHz, and an average signal attenuation of about 1 to 1.2. The frequency response stays flat within 1 dB between roughly 3 MHz and 490 MHz. The probe is battery powered, using Li Ion 18650 cells recharged through USB C or a DC barrel jack, and it should connect to a scope set to 50 ohm input impedance.
Versions and where they stand
Three versions are tracked. v1.0 is the completed proof of concept, flat within 1.3 dB to 500 MHz. v1.0.1 is fully characterized and available, fixing bugs from 1.0 and adding over and under range indicators. v1.1 exists only in simulation, with a flatter estimated response to 1 GHz, but it has not been manufactured yet.
Noise, and how to buy one
The design keeps introduced noise low by running on battery and avoiding semiconductors in the critical path. Simulated total noise works out to about 78.7 microvolts RMS over the full instrument bandwidth, and the README includes a table showing how that drops as the measurement bandwidth narrows. Version 1.0 is available assembled for roughly 249 pounds, and blank PCBs, enclosures, and machined panels can be bought as a kit for anyone who wants to build it.
Community notes