On Mon, Apr 20, 2015 at 3:03 PM, Mark Brown broonie@kernel.org wrote:
On Mon, Apr 20, 2015 at 02:18:56PM -0700, Kevin Cernekee wrote:
On Sat, Apr 18, 2015 at 4:16 AM, Mark Brown broonie@kernel.org wrote:
On Wed, Apr 15, 2015 at 02:42:19PM -0700, Kevin Cernekee wrote:
+- VDD-supply: regulator phandle for the AVDD/DVDD/HP_VDD supply
This is clearly not correct - if there are three separate physical supplies there should be three separate regulators requested. They may all resolve to one physical regulator on the board you are working with but that might not be true on other boards.
In the "simplified diagram," TI shows a single AVDD/DVDD/HP_VDD supply:
Page 20 also suggests the use of a single 3.3V supply for AVDD/DVDD/HP_VDD.
But this combines a number of separate pins. On 5711 we have dedicated pins for:
PVDD_A PVDD_B PVDD_C PVDD_D AVDD DVDD
Yes, those are three separate supplies that are typically tied together
- it looks like PVDD has high current draw so is tied through multiple
pins.
Each half-bridge output OUT_{A,B,C,D} has its own dedicated PVDD_{A,B,C,D} supply on 5711:
http://www.ti.com/lit/ds/symlink/tas5711.pdf#7
I don't understand the statement: "those are three separate supplies that are typically tied together." AVDD/DVDD are 3.3v fixed but PVDD is 8V minimum (5711) or 4.5V minimum (5717/5719). So AVDD/DVDD can never be driven by the same regulator output as PVDD. Perhaps you were thinking of HPVDD/AVDD/DVDD on 5717/5719?
The simplified application diagram on page 2 does show that AVDD/DVDD are typically tied together, and that PVDD_* are also typically tied together. But as you pointed out, there may be situations where that isn't the case, so that leaves us with up to 6 separate supplies on 5711, or 5 on 5717/5719. The typical case is probably closer to 2 or 3, though.
I'd not be surprised to see systems with AVDD tied to a separate pin, analogue supplies often benefit from low noise supplies separate to digital ones. Indeed if you look at the pin descriptions the analogue and digital supplies even have separate grounds.
I didn't see anything in the datasheet suggesting it is OK to have different voltages or power states on the various supply pins (other than the special voltage on PVDD).
That's more likely to go wrong if they're not controlled separately than if they are since we will only be able to turn a single supply on or off..
I can add as many regulator entries as appropriate. What do you recommend?
Have the driver accurately reflect the hardware, request one regulator per supply.
So, request one regulator per VDD pin? Or group all of PVDD_* into a single regulator?