Video DACs - split from Blog

Comments

• jmg Posts: 15,148

  2013-12-04 14:04 edited 2013-12-04 14:04

  Cluso99 wrote: »

  Obviously, Chip & Beau would need to chime in to say if this is even possible. However, with the savings made in just routing the video dacs the new way freed up so much space, this idea IMHO is likely to free up another large space.

  
  IIRC Chip said the DAC themselves came almost for free, as they are under the IO areas.
  So there is little gain, and these are already in the custom layout, so are 'done'.
  The any-any routing was outside the ring, and was significant.
  

  [*]Are there any other potential uses?

  
  Of course.
  I can see waveform/pulse generation, and ATE uses.
  Perhaps even Data transmission..
  

  Is there likely to be any use for more than 1, 2, 3... sets of Video DACs in a P2?

  
  IIRC Chip said FAST DAC was possible for 4 DACS per COG, for a max of 32 DACS (fast), if all COGs enabled theirs.
  Mapping was on a '4 of 12' basis, in pin groups.
  
  Not sure on any global sync of 32 FAST DACs, but some WAIT opcodes could manually sync DACs
  
  
  All chip DACs are available to any COG, on a 16 cycle slower path basis
  (I'm guessing ones mapped to Video, need to be unmapped, to use the slower path)
  
  Found some info here :
  http://forums.parallax.com/showthread.php/125543-Propeller-II-update-BLOG?p=1223120&viewfull=1#post1223120
• Roy Eltham Posts: 2,996

  2013-12-04 15:06 edited 2013-12-04 15:06

  The DACs are largely in the pin pads. There are 92 of them. Choosing to use them for Video is more of a configuring of how the cog drives the 4 DACs it has mapped than anything. There's aren't special Video only DACs.
• Tubular Posts: 4,622

  2013-12-04 15:46 edited 2013-12-04 15:46

  I think the "video" part of the term "video dac" may be more about the fast settling time. They probably also have a 75 ohm output impedance (like is achieved with the current dac resistors on the breakout board). This certainly doesn't stop you doing other things with them and they can be used for almost any purpose. You can buffer if you need a lower impedance for whatever reason.
  
  If you haven't already, since you have a comms background have a play around with generating and modulating sine waves using the counters coupled to the dacs, You can have dac0 happily outputting video and two phase shifted sine waves on other channels (eg dac 2 and 3), with 9 bit control over each of the amplitude, offset and phase shift. It is simply about three instructions and a constant or two to make all this work.
  
  Or, there is a mode where you can use both counters in a cog to add together before being sent to the dac(s), for modulation of a carrier, etc. I did a quick, successful test of transmitting to AM radio. It will be interesting to see whether we can do this for 27 MHz when we get 160~200 MHz silicon. P2 controlled RC cars here we come
• potatohead Posts: 10,254

  2013-12-04 15:55 edited 2013-12-04 15:55

  I was wondering if the Video DAC section could be removed from each cog and just 1 or 2 of them be placed on the die, and permitting one cog (any cog) to claim each Video DAC.

  
  I really would not like to see this. They can be used in combination for some tricks on component displays, and having a P2 able to drive multiple displays may well prove handy.
  
  As already mentioned, you don't have to drive video signals with them either.
• Cluso99 Posts: 18,069

  2013-12-04 16:07 edited 2013-12-04 16:07

  Thanks guys. Now I am confused as to how Chip saved that DAC routing space.
  
  I was thinking that we now have a restriction of what pins can be driven by the DACs connected to respective cogs. I was thinking there could be some benefit by tying the DAC/Video generation to the pin matrix (which could provide access to any of the pins) by decoupling the video/dacs from the cogs, and this might free some more dac routing.
  Obviously my understanding is flawed. I was just kind of hoping that there could be more dac routing saved if the implementation could be separated differently.
• Bill Henning Posts: 6,445

  2013-12-04 16:07 edited 2013-12-04 16:07

  Tubular wrote: »

  Or, there is a mode where you can use both counters in a cog to add together before being sent to the dac(s), for modulation of a carrier, etc. I did a quick, successful test of transmitting to AM radio. It will be interesting to see whether we can do this for 27 MHz when we get 160~200 MHz silicon. P2 controlled RC cars here we come

  
  Now that is seriously cool!
• cgracey Posts: 14,133

  2013-12-04 18:40 edited 2013-12-04 18:40

  Bill Henning wrote: »

  Now that is seriously cool!

  
  Any cog can still drive any pin's DAC with a static value, updated in 16 clocks.
  
  Each cog can drive half a side of the chip's DACs with video- or CTR-generated values which update every clock.
  
  75 ohms is very low-impedance by DAC standards, but it's necessary to drive fast signals accurately.