I asked Pyramid's engineer and he said
Doug and Matt,
I understand completely what both of you are saying. You are correct in that in normal everyday usage, the PB does not draw anywhere near the current called out in the specs (and as Matt pointed out, the Morex is able to supply less current at 3.3 Volts than the original supply so they can't be pulling the maximums).
I was simply pointing out that running at the worst case specs you would have these sort of dissipation numbers and that the difference is drastic. Perhaps it's not 100% valid, but I was simply trying to point out that by moving part of the power supply into an external brick, you have much better thermal conditions inside the chassis. Worst case specs are generally there for a reason. I know when I was an engineer I always tried to add up everything in the worst case. The times when I didn't caused trouble at some point.
your right , the comparison is not 100% valid.
looking at the worst case spec is like comparing cars by the speedometer, mine goes to 85 and yours goes to 120 (max power output) but both cars can do 75 mph all day. Seems we really want to know miles per gallon (efficiency).
So we put them on the track and the Morex and OEM efficiencies are very close.
Using the Morex to split the power supply and put some of the waste heat outside does not work as you would hope. The Morex configuration is actually TWO power supplies. 120vac to 12vdc on the outside and 12vdc to (3.3vdc and 5vdc) inside. the OEM supply is two steps also, 120vac to 180vdc and 180vdc to (3.3+5)vdc. But the OEM conversion of 120vac to 180vdc is almost 100% efficient so almost nothing is improved inside the HD1000. In fact the Morex conversion is LESS efficenent than (a properly working ) OEM if you start at the wall socet and add up ALL of the power consumed
Now to the original supplies when they start to fail:
When my Rev B supply failed, I took it out and to my surprise the output electrolytic capacitors had either unsoldered themselves or were a cold solder joint to begin with (they litterally fell off when I took the board out). Given that output capacitors on power supplies tend to either reduce noise or improve stability of the supply, I would argue that a failing supply has a horrible output. And if the supply is unstable due to bad caps, it could be oscillating somewhat which would cause more dissipation and heat and a gradual cascading failure...
what we have found is yes, as the capacitors dry up and loose their value the power supply runs poorly and hotter. the additional heat speeds up the capacitor failure. a downward spiral for sure.
Rebuilding the OEM PS with good caps and relocating the power resistor results in a reliable and only slightly warm PS
Roku3 and Roku HD1000 [Rev B] on a Samsung HLP5674W DLP in the living room; a Roku2 and two Roku XS and a few SoundBridges.Win7; Kubuntu and XP via RT-N66U, E2000 and a switch or two. I own stock in Roku, it's just all in the form of hardware.Viva la Roku