I always considered GEs system of using the gated burst to ring a tank circuit (that would resonate for longer than a horizontal line) to be the best. No expensive special components (crystals cough) and I've never seen it need adjusting to achieve color sync unlike RCA and Westinghouse sets.

Not to pick nits, but it's actually the reference oscillator, supplying the signal against which the (demodulated) subcarrier (i.e., chroma signal) is referenced.

All the while, the burst signal, coming as as quick "burst" during retrace, locks the reference osc. to a steady 3.58 Mhz. The burst is only a small number of cycles in duration. Then the osc. simply "coasts" on thru the scan period.

The tint (or hue or whatever) generally works by varying the phase of the burst signal.

(Edit.) GE's "ringing tank" is a shortcut method of achieving the same result minus the dedicated reference oscillator.

Ok, let's get clearer on this:

1) Color demodulation for NTSC or PAL needs a reference subcarrier to do the demodulation.
2) The carrier can be generated by an oscillator or a ringing circuit.
3) In the ringing circuit, there is no feedback to make a continuous CW. Therefore, the ringing resonator has to be very high-Q to keep ringing through at least one line time. Thus, a crystal is needed. The ringing decreases in amplitude between burst injections and therefore needs to be followed by a limiter to provide constant amplitude output to the demodulators. If the circuit is high enough Q to ring for several lines, the amplitude of burst and accompanying noise that is applied can be reduced. The circuit then takes more than one line to reach full amplitude when first turned on, but the effects of noisy signals are reduced.
4) An oscillator can be synchronized with the burst by use of a phase detector in a phase-locked loop. This is the most commonly used carrier recovery circuit. The loop filtering has low bandwidth (in the hundreds of hertz) and therefore responds less to noise impulses than if it were wider band.
5)An oscillator can be synchronized by injection of the burst, without use of a phase lock loop. In this case, it is like a ringing circuit with an infinite ring duration. Motorola used this in its single-tube "SODPIL" (Self Oscillating Detecting Phase Injection Lock) color circuit. Performance can be very similar to a PLL.
6)An oscillator can be based on a crystal resonator or an LC resonator. The crystal oscillator can be designed so it is nearly impossible to make it run more than 1 kHz or so off frequency, so that side lock is impossible. (Side lock is when the oscilator is off frequency by an integer multiple of horizontal rate, so that the output is back in phase with the burst when it occurs.)
7)An LC oscillator has to be designed with as high a Q as possible and very low temperature coefficient to prevent it drifting out of lock or even into side lock. An accessible frequency adjustment is a good idea in case it does drift too far.
8) SECAM uses FM chroma carriers instead of quadrature amplitude modulation, so the detection of the chroma is entirely different from NTSC or PAL. For example, SECAM can be subject to high frequency luma components exceeding the FM noise reduction threshold, resulting in bursts of color on strong details.