FS#12094 - RDA5802 tuner: clean up register 0x4 handling, drop implicitly enabled soft-mute
Opened by Stephan Grossklass (keyb_gr) - Tuesday, 03 May 2011, 20:51 GMT
Last edited by Bertrik Sikken (bertrik) - Saturday, 04 June 2011, 20:18 GMT
The attached patch does two things:
1. The initial value of register 0x4 is changed to make sense according to the RDA5802E datasheet. Originally, it sets bit 10 (value 0x0400), which is undocumented and does not seem to have any effect. However, I believe that the original intent was to set bit 11, which switches deemphasis to 50 µs. This would be in line with the other initial settings for frequency range (87.5 .. 108 MHz) and tuning steps (100 kHz), typical for Europe etc. (which is 50 µs territory). Therefore, the new value of of this register is 0x0800.
2. There was a line in the muting functionality which directly set register 0x4 bits 9 and 10, effectively causing them to be always enabled. (This is presumed to predate the availability of a datasheet and might even be a remnant from the Si4702 driver.) Bit 9 controls the "soft-mute" feature (volume reduction on weak stations, not associated with regular muting), which is enabled by setting it. Therefore, this was always on, without being set explicitly. The line in question is removed.
At this point it is unclear to me whether we do or do not want soft-muting (or whether it should be user-selectable even) - this needs further discussion. If we do, it should be enabled explicitly, either via the initial value of register 0x4 bit 9 or using the SYSCONFIG1_SMUTE define already present.
Two versions of this patch (against r29814) are attached:
* The standard version (std) contains the changes as outlined above.
* The other one additionally increases LNA working current from 2.5 mA to its maximum of 3.0 mA for maximum sensitivity. This is done by setting register 0x5 bits 5:4 from 10 to 11. More testing is needed to determine whether this really has the desired effect of improving reception on weak stations (it ought to give a few dB more signal, and I do think RSSI readings have increased a bit) or whether SNR does not improve enough to be worth the additional current draw.
Practical effects of this patch:
Soft-mute function is disabled, so volume levels on weak stations will remain constant and noise is heard in all its glory.
Battery life during radio operation is likely to be somewhat reduced on the "more gain" version, as LNA operating current increases by 0.5 mA. The effect on frontend dynamic range is unclear (the chip does seem to have LNA AGC, so nothing bad should happen).
Saturday, 04 June 2011, 20:18 GMT
Reason for closing: Accepted
Additional comments about closing: Committed as SVN r29959