+ cx18_write_reg_expect(cx, 0x00020000, CX18_ADEC_CONTROL,
+ 0x00000000, 0x00020002);
+
+ /*
+ * The PLL parameters are based on the external crystal frequency that
+ * would ideally be:
+ *
+ * NTSC Color subcarrier freq * 8 =
+ * 4.5 MHz/286 * 455/2 * 8 = 28.63636363... MHz
+ *
+ * The accidents of history and rationale that explain from where this
+ * combination of magic numbers originate can be found in:
+ *
+ * [1] Abrahams, I. C., "Choice of Chrominance Subcarrier Frequency in
+ * the NTSC Standards", Proceedings of the I-R-E, January 1954, pp 79-80
+ *
+ * [2] Abrahams, I. C., "The 'Frequency Interleaving' Principle in the
+ * NTSC Standards", Proceedings of the I-R-E, January 1954, pp 81-83
+ *
+ * As Mike Bradley has rightly pointed out, it's not the exact crystal
+ * frequency that matters, only that all parts of the driver and
+ * firmware are using the same value (close to the ideal value).
+ *
+ * Since I have a strong suspicion that, if the firmware ever assumes a
+ * crystal value at all, it will assume 28.636360 MHz, the crystal
+ * freq used in calculations in this driver will be:
+ *
+ * xtal_freq = 28.636360 MHz
+ *
+ * an error of less than 0.13 ppm which is way, way better than any off
+ * the shelf crystal will have for accuracy anyway.
+ *
+ * Below I aim to run the PLLs' VCOs near 400 MHz to minimze errors.
+ *
+ * Many thanks to Jeff Campbell and Mike Bradley for their extensive
+ * investigation, experimentation, testing, and suggested solutions of
+ * of audio/video sync problems with SVideo and CVBS captures.
+ */
+
+ /* the fast clock is at 200/245 MHz */
+ /* 1 * xtal_freq * 0x0d.f7df9b8 / 2 = 200 MHz: 400 MHz pre post-divide*/
+ /* 1 * xtal_freq * 0x11.1c71eb8 / 2 = 245 MHz: 490 MHz pre post-divide*/