LOCK-STEP question
Sleazy - G
Posts: 79
I remember when chip realized that you could run more than one prop in lock step.· So im thinking to myself now.
Instead of running lock-step you run lock-half step? Could you hook the 5mhz·oscillator to one propeller correctly and connect a second propeller to the same oscillator in reverse?? One prop operating on the clocks falling edge and the other on the leading?
just an idea.· I figure if it works instead of enumerating props and have them synchronize their execution in an array, you could chain props together with one clock, each consecutive prop in the chain running in link - half step with the adjacent propellers in the array.· This could give a propeller array·made of half-step pairs could allow more data speed.
If 2 propellers shared an identical·link - step clock,·and shared their INA ports,·INA samples would be·identical between the two props, but in link - half step you should be able to double the INA data rate by having both props "take turns" at writing INA samples·to their local ram.
To share the INA port, you just solder two props together, like a sandwich.· Salami.
Am I on to anything here?
It feels good....
Post Edited (Sleazy - G) : 12/30/2007 11:39:04 AM GMT
Instead of running lock-step you run lock-half step? Could you hook the 5mhz·oscillator to one propeller correctly and connect a second propeller to the same oscillator in reverse?? One prop operating on the clocks falling edge and the other on the leading?
just an idea.· I figure if it works instead of enumerating props and have them synchronize their execution in an array, you could chain props together with one clock, each consecutive prop in the chain running in link - half step with the adjacent propellers in the array.· This could give a propeller array·made of half-step pairs could allow more data speed.
If 2 propellers shared an identical·link - step clock,·and shared their INA ports,·INA samples would be·identical between the two props, but in link - half step you should be able to double the INA data rate by having both props "take turns" at writing INA samples·to their local ram.
To share the INA port, you just solder two props together, like a sandwich.· Salami.
Am I on to anything here?
It feels good....
Post Edited (Sleazy - G) : 12/30/2007 11:39:04 AM GMT
Comments
Crystals cannot be driven simulatanously by two Propellers, instead you must use an active oscillator with complmentary outputs and drive each Propeller on the opposing signal. Doing this creates a DDR-like scheme and by using 4 cogs on each Propeller, a INA sampling rate of 160MHz is possible. There is a slight wrinkle in that some step will be needed to align the data streams from each Propeller since each Propeller's inital CNT value is indeterminate (IOW random), so the two Propellers need to synchronize thier system counters.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Paul Baker
Propeller Applications Engineer
Parallax, Inc.
As long as your signal isnt completely arbitrary this should work, right?
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Paul Baker
Propeller Applications Engineer
Parallax, Inc.
···········
Astonishing.