Version 01.05, Nov. 06, 2005 - Additional probe notes.

Homemade Probes

This web page describes some fairly inexpensive voltage probes for Prius MG1 and MG2 monitoring. MG1, 18 kW (24 hp), operates the center sun gear as a combination clutch and variable ratio transmission. The engine drives and is connected to the planetary gear carrier via a long shaft that also serves as the bearing for the hollow, sun gear. MG2, connected to the ring gear, provides up to 33 kW (44 hp) to augment the 70 hp gas engine. We need sensors to understand which is doing what at any given time.

Each sensor uses a pipe clamp, wing-nut, plastic sewing machine bobbin and some #28 magnetic wire and cost less than $2 each. A 10 ft. section of Cat-5 ethernet cable terminates in a miniature audio jack in the cabin:

Each pipe clamp goes around a single MG1 or MG2 power cable with the carriage bolt passing through the wire-wound, plastic bobbin. The original nut is affixed and the wing nut used to lock the two together. This makes a simple transformer with the primary being one-turn of a power leg and the bobbin providing the secondary and generating the output voltage.

Earlier tests indicated a 100 turn bobbin would generate voltage in the 100 millivolt range for MG1. The MG2 bobbin has only 55 turns to keep it in the same voltage range as MG1. Experimental data using hard accelleration shows MG2 generated just under 500 millivolts. The goal is to plug the sensors into the microphone input of a Macintosh G3 Wallstreet which is rated at 2 V. max input into 6.8 k Ohms and use audio recording software to capture the data.

This example shows the detail available from an 11 minute trip. The route was from a parking lot; up a short hill to a stretch of 55 mph; an exit zone at 35 mph; another short hill and stretch of 65 mph; an off ramp; a 40 mph urban street; and ended at an electronics store.


We can use this data to compare identical routes at different speeds on a trip-by-trip basis instead of waiting for 2-3 gallons to burn off. We also need to calibrate the sensors to give engineering units and need to optimize the sampling rate. BTW, listening to the audio during the drive has revealed some audio signatures such as the MG1 'spinning down' as the engine starts and 'spinning up' as it stops:

Eventually, we'll have a Java application to record and analyze the audio data realtime and make it host independent.

Others have already shared a lot of data about the MG1/MG2 operation but this data will help develop an automatic "Pulse and glide" system to augment the cruise control. It could also drive a couple of small speakers and give audio clues for "Pulse and glide" driving to avoid watching the energy display.