Ping module question

James Armstrong

Is there a way to modify the ping code to be able to start multiple ping sensor units at the same time and wait for all the responses and get the measured time before returning? I need to be able to monitor 4 or 5 sensors but don't want to use one cog per sensor or call one function per sensor in series because of the delays that would be added up. If the max wait time is ~50ms (max distance) and I am reading 5 of these in succession that would be about 250ms per ping cycle not allowing for more than 4 reads per second. I thought I saw a function somewhere that would start a timer that would measure a pulse on an input or multiple inputs, and wait for all of them to change before returning. I can't remember where I say that code.

- James

Chad George

James,

I've been working on a robot project that uses the SRF05 sonar unit which is very similar to the PING unit. I had the same problem when I wanted to drive multiple sonars from the same cog. I wrote a new assembly driver that will monitor multiple sonar units simultaneously. It is currently written to drive 3 sonars although the code should scale up to 5 easily. I've also included·SPIN code that demonstrates using the driver.

One word of·caution, I'm using the DLP-PROP units which run at 96MHz instead of the 80Mhz that the demo board runs at. It should be obvious where to make the appropriate changes.

Hope it helps,

Chad

Post Edited (Chad George) : 12/11/2006 6:06:11 PM GMT

Mike Green

' Assuming you have 4 pings starting at pin 4 with pulldowns (say 10K) to ground:
VAR
  long startTime, elapsed, started, pingCount

PUB timeIt | x
' Start 4 pings going and record the pulse width for each in elapsed[noparse][[/noparse]x].
' Units for pulse width are system clocks (normally 12.5ns).
  repeat x from 0 to 3
    started[noparse][[/noparse]x] := false
  pingCount := 4
  outa[noparse][[/noparse]4..7] := %0000
  dira[noparse][[/noparse]4..7] := %1111 ' make outputs
  outa[noparse][[/noparse]4..7] := %1111 ' start pulse
  waitcnt(clkfreq / 100_000 + cnt) ' 10us pulse
  dira[noparse][[/noparse]4..7] := %0000 ' change to inputs
  waitcnt(clkfreq / 2_000 + cnt) ' 500us hold off
  repeat while pingCount    ' wait until all pings respond
    repeat x from 0 to 3
      if started[noparse][[/noparse]x]    ' this ping's response pulse has started
        if ina[noparse][[/noparse]4+x] == 0    ' the pulse is now back to zero
          elapsed[noparse][[/noparse]x] := cnt - startTime[noparse][[/noparse]x]    ' record the elapsed time
          pingCount--    ' one less ping to wait for
      else
        if ina[noparse][[/noparse]4+x] == 1   ' not yet started, pulse now high
          startTime[noparse][[/noparse]x] := cnt   ' save the start time
          started[noparse][[/noparse]x] := true   ' mark pulse as started

acantostega

You probably already know this, but just in case, make sure the pings are positioned so the echoes of the respective pings don't interfere with each other.

John Abshier

Mike, won't all the elapsed times equal cnt for last sonor - their individual start times? I don't see anything to prevent multiple assignments to elapsed[noparse][[/noparse]x].

Mike Green

John,

You're absolutely right.· Easiest way to fix this is to initialize the elapsed array to zero in the loop where started is initialized, then test for·"ina[noparse][[/noparse]4+x] == 0 and elapsed[noparse][[/noparse]x] == 0".· Sorry

Mike