All the outlets in my garage are GFCI, and my lathe pretty frequently trips the circuit when i turn the speed up.
GFCIs are a good thing to have in a shop - there are many electrical hazards, and GFCIs are are specifically designed to protect people against electric shock. That's why the NEC requires them on receptacle circuits. However, there several things that can trigger GFCIs. And more importantly, the 'problem' may actually be a symptom of something else.
Potential issues include:
- harmonics - electronic devices on the circuit cause voltage harmonics, which in turn cause current harmonics to flow in the circuit. GFCIs, like all other electrical devices, are designed for the fundamental frequency (60Hz). In some cases, it is possible to predict what harmonics might be present (for example, a simple three-phase variable speed drive device will produce a significant 300 Hz component, but a single-phase variable speed device will produce other frequencies including a significant third harmonic (180Hz). And depending on the design of the variable speed controller on the lathe, the harmonic composition of the current to the lathe can change as you change speed. In this instance, the circuit feeds both the variable-speed lathe as well as fluorescent lights, both of which are known sources of harmonics. Conventional magnetic ballasts produced third-harmonics, while modern electronic ballasts produce other frequencies. The GFCI designer can't possibly know what harmonic frequencies will be present, so in general, the high-frequency response of the detection circuit in consumer-grade GFCIs is unknown.
- Neutral-to-earth circuit - in theory, all of the current that flows out the hot lead (the black wire) should return via the neutral (the white wire), and the main protection functionality in a GFCI compares these two currents - if they are the same, everything is fine. If they are not the same, that could be because a person has come in contact with the hot wire and current is flowing through that person's body. In that case, the GFCI is designed to trigger. Inadvertently grounding the white wire at the load can also upset this balance, so some GFCIs are designed to detect when the neutral is grounded They do this by impressing a high-frequency voltage between neutral and ground and looking for a corresponding high frequency current flow. But that high-frequency current detection circuit can be fooled by high frequency currents that flow in the circuit for other reasons.
- excessive leakage. Every circuit includes some unavoidable leakage - both resistive leakage through insulation, and capacitive leakage because the circuit parallel the earth. If that leakage is excessive, it can appear as an undesirable unbalance between the hot and neutral currents. The leakage increases with circuit length. This was a common problem with early GFCIs that were installed in the service panel. The current practice of putting GFCIs in the receptacle itself means that the circuit length beyond the GFCI can be shorter, thereby reducing the tendency for this problem to appear. But if there is too much circuit beyond the GFCI, it can still trigger incorrectly.
- a combination of these factors
Are the lights in the garage hard-wired or plugged into a receptacle? The NEC requires that there be a GFCI on receptacles, but not on hard-wired fixtures. I would consider modifying the wiring to make then hard wired (ie, NOT connected through a receptacle), and then connect them to the circuit ahead of the GFCI.
Also, I would bring the main feed in to the garage from the house, and then split it into three branches. One branch would feed the hard-wired lights. The second would feed general purpose receptacles though a GFCI, including any receptacles outside the garage. The third branch would feed a single receptacle for the lathe through a dedicated GFCI. That way, any harmonic currents caused by the lathe would affect only the one GFCI associated with the lathe.
If the problem persists after these modifications, then it is likely that the cause is the lathe itself - either simply a high variable magnitude harmonic current, or a high leakage current combined with a variable magnitude harmonic such that the combination is triggering the GFCI. You could always hard-wire the lathe (eliminate the receptacle) in order to get around the NEC requirement for the GFCI, but that would mask the fact that something is happening in the lathe that could be an incipient problem in the variable-speed box.