Shop OBEX P1 Docs P2 Docs Learn Events
Make your Propeller dissipate 1.5W all by itself! — Parallax Forums

Make your Propeller dissipate 1.5W all by itself!

cgraceycgracey Posts: 14,206
edited 2006-11-17 15:26 in Propeller 1
Paul Baker was characterizing Propeller current consumption·today and he·was using·several small test programs to set up various cases. We modified one program to cause the Propeller to consume as much current as possible. It enables all cogs and PLLs and swings all 32 I/O pins at 160MHz. This causes the chip to dissipate ~1.5 Watts, which makes it quite hot. You can try it yourself:

'' This is a silly little program which starts all 8 cogs
'' and makes each cog enable both of its CTRs so that 4
'' pins will be outputting 160MHz. This results in all 32
'' I/O pins outputting 160MHz. This creates ~1.5 watts of
'' power dissipation, ~75% of which is a result of driving
'' so many pins so fast. Note how hot the chip gets. It
'' gradually sags the Demo Board's 3.3V regulated supply
'' until a brown-out reset occurs. Be sure to use F10 to
'' run (not F11).

CON
· _clkmode······· = xtal1 + pll16x
· _xinfreq······· = 5_000_000
VAR
· long stack[noparse][[/noparse]80]
PUB go | i
· repeat i from 7 to 0
··· coginit(i, crank_plls(i<<2), @stack[noparse][[/noparse]i*10])
PUB crank_plls(i)
· ctra := $0F800000 | i | (i+1) << 9
· frqa := $20000000
· ctrb := $0F800000 | (i+2) | (i+3) << 9
· frqb := $20000000
· dira := $FFFFFFFF
· repeat

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔


Chip Gracey
Parallax, Inc.

Comments

  • Graham StablerGraham Stabler Posts: 2,510
    edited 2006-11-16 10:42
    8 processor hand warmers here we come
  • CJCJ Posts: 470
    edited 2006-11-16 12:58
    sounds like there are gonna be a few things in the datasheet that look like misprints, like the magic smoke pressure limit and the power consumption range.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Who says you have to have knowledge to use it?

    I've killed a fly with my bare mind.
  • Bill HenningBill Henning Posts: 6,445
    edited 2006-11-16 18:08
    Cool... err... Hot!
    I wonder if it would be possible to draw even more... hmmm...
    what if the video shift registers were also going full blast, being fed by tight asm code on all eight cogs?
    Bill
    Chip Gracey (Parallax) said...
    Paul Baker was characterizing Propeller current consumption·today and he·was using·several small test programs to set up various cases. We modified one program to cause the Propeller to consume as much current as possible. It enables all cogs and PLLs and swings all 32 I/O pins at 160MHz. This causes the chip to dissipate ~1.5 Watts, which makes it quite hot. You can try it yourself:

    '' This is a silly little program which starts all 8 cogs
    '' and makes each cog enable both of its CTRs so that 4
    '' pins will be outputting 160MHz. This results in all 32
    '' I/O pins outputting 160MHz. This creates ~1.5 watts of
    '' power dissipation, ~75% of which is a result of driving
    '' so many pins so fast. Note how hot the chip gets. It
    '' gradually sags the Demo Board's 3.3V regulated supply
    '' until a brown-out reset occurs. Be sure to use F10 to
    '' run (not F11).

    CON
    · _clkmode······· = xtal1 + pll16x
    · _xinfreq······· = 5_000_000
    VAR
    · long stack[noparse][[/noparse]80]
    PUB go | i
    · repeat i from 7 to 0
    ··· coginit(i, crank_plls(i<<2), @stack[noparse][[/noparse]i*10])
    PUB crank_plls(i)
    · ctra := $0F800000 | i | (i+1) << 9
    · frqa := $20000000
    · ctrb := $0F800000 | (i+2) | (i+3) << 9
    · frqb := $20000000
    · dira := $FFFFFFFF
    · repeat
  • Paul BakerPaul Baker Posts: 6,351
    edited 2006-11-16 18:09
    The measured temp of the Propeller on the Demo board prior to reset was 153F, the measured temperature of the voltage regulator, 223F. The exact cause of the reset has not been fully determined, it's one of three possible causes:
    • Reset due to overcurrent voltage sag, initiated by the Propeller.
    • Reset due to overcurrent, initiated by the regulator.
    • Reset due to overtemperature, initiated by the regulator.
    I feel the evidence is inconclusive as to which occurs, it's basically a chicken and the egg scenario: Did the Propeller brown out because the regulator reset, or did the regulator reset because the Propeller browned out.

    Because of this, care should be used when using this program on any non-demoboard setup. With a beefier supply, or supplies which don't have shutdown protection,·this program can easily cause a thermal runaway condition and release the smoke, considered yourself warned.


    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Paul Baker
    Propeller Applications Engineer

    Parallax, Inc.
  • cgraceycgracey Posts: 14,206
    edited 2006-11-16 19:36
    Paul,

    We definitely saw the supply sagging lower and lower until the Propeller reset. With an adequate current supply, the Propeller might run indefinitely without things getting so hot that the chip slowed beneath the clock requirements. Could you please dig up an old Rev A Demo Board (with the HUGE regulators), and run this program again? It will either keep running, or the silicon will slow enough from the heat that it won't be able to keep up with the clock, and then maybe it will just hang, zeroing the current. My guess is that the cog processors will hang, but the CTRs, having been configured, will keep running. Either way, I bet the·Propeller will be unscathed. Maybe months at that temperature could hurt it, but I bet it's not a practical problem.
    Paul Baker (Parallax) said...

    The measured temp of the Propeller on the Demo board prior to reset was 153F, the measured temperature of the voltage regulator, 223F. The exact cause of the reset has not been fully determined, it's one of three possible causes:
    • Reset due to overcurrent voltage sag, initiated by the Propeller.
    • Reset due to overcurrent, initiated by the regulator.
    • Reset due to overtemperature, initiated by the regulator.
    I feel the evidence is inconclusive as to which occurs, it's basically a chicken and the egg scenario: Did the Propeller brown out because the regulator reset, or did the regulator reset because the Propeller browned out.

    Because of this, care should be used when using this program on any non-demoboard setup. With a beefier supply, or supplies which don't have shutdown protection,·this program can easily cause a thermal runaway condition and release the smoke, considered yourself warned.·


    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔


    Chip Gracey
    Parallax, Inc.
  • cgraceycgracey Posts: 14,206
    edited 2006-11-16 19:39
    Bill Henning said...
    Cool... err... Hot!
    I wonder if it would be possible to draw even more... hmmm...
    what if the video shift registers were also going full blast, being fed by tight asm code on all eight cogs?
    Bill
    Well, about 75% of the power is coming from just having to swing the pins so fast. The addition of video circuit currents would best be tested by not driving any pins, as it would only slow down the current 160 MHz transition rates. I imagine it would add less than 100ma, total, and only that much because of the crazy 160MHz clock.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔


    Chip Gracey
    Parallax, Inc.
  • Dennis FerronDennis Ferron Posts: 480
    edited 2006-11-16 20:14
    I accidently wired my PropStick's 3.3 volt line to 7.2 volts, and then left it on by accident. It ran until the 6 NiMH batteries discharged, and never even glitched the TV graphics demo that was running. It's now wired up correctly and running fine. Truly an amazing chip.

    Now we just need to put a heat sink on it and overclock it with overvoltage. Give a prize to the first person to break 1000 milliamps.
  • cgraceycgracey Posts: 14,206
    edited 2006-11-16 20:50
    Once at Parallax, we gave it 5V and continuously sprayed it with coolant, and we were able to run the internal clock at 160MHz. I don't know what the current was, but it was surely high.
    Dennis Ferron said...
    I accidently wired my PropStick's 3.3 volt line to 7.2 volts, and then left it on by accident. It ran until the 6 NiMH batteries discharged, and never even glitched the TV graphics demo that was running. It's now wired up correctly and running fine. Truly an amazing chip.

    Now we just need to put a heat sink on it and overclock it with overvoltage. Give a prize to the first person to break 1000 milliamps.
    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔


    Chip Gracey
    Parallax, Inc.
  • Cliff L. BiffleCliff L. Biffle Posts: 206
    edited 2006-11-16 20:50
    Dennis Ferron said...
    Now we just need to put a heat sink on it and overclock it with overvoltage. Give a prize to the first person to break 1000 milliamps.

    With a nice, solid solder bead between VCC and VSS, you could probably break an amp right now. smile.gif
  • Paul BakerPaul Baker Posts: 6,351
    edited 2006-11-16 21:17
    Ok, ran the test on the DemoBoard Rev D, replacing the on board supply with a bench top supply. The current was 550mA until the chip got to ~160F and preceded to drop the current to ~543mA and hit a quiesent temperature of 173F. Test was run for 5 minutes, no reset during the test. So the regulator was likely in thermal runaway and couldn't maintain the current requirements (I think its a 500mA part which would explain it supplying it for a while), so when it's output voltage dropped below 3V, the Propeller Brown Out reset the system.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Paul Baker
    Propeller Applications Engineer

    Parallax, Inc.
  • Jeff MartinJeff Martin Posts: 760
    edited 2006-11-16 23:24
    Hi Paul and Chip,

    I've been thinking about your serious problem noted in this thread.

    I've come up with the solution using the Prop Stick USB.· See attachment.


    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    --Jeff Martin

    · Sr. Software Engineer
    · Parallax, Inc.
    1024 x 296 - 38K
  • Beau SchwabeBeau Schwabe Posts: 6,568
    edited 2006-11-16 23:35
    Jeff,

    I also think that it's very important to use·one of these...tongue.gif

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Beau Schwabe

    IC Layout Engineer
    Parallax, Inc.
    450 x 446 - 66K
    fan.jpg 66.4K
  • Bill HenningBill Henning Posts: 6,445
    edited 2006-11-17 07:46
    Looks good :-)

    FYI, I've overclocked an Intel Core 2 Duo E6400 to 3.64GHz (from 2.13GHz), rock solid, air cooled with a Noctua 12 cooled by two 12cm 63cfm fans [noparse]:)[/noparse] [noparse]:)[/noparse] [noparse]:)[/noparse]
    Jeff Martin (Parallax) said...

    Hi Paul and Chip,

    I've been thinking about your serious problem noted in this thread.

    I've come up with the solution using the Prop Stick USB.· See attachment.


  • simonlsimonl Posts: 866
    edited 2006-11-17 13:49
    Jeff LOL. That's funny smile.gif

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Cheers,

    Simon
    www.norfolkhelicopterclub.co.uk
    You'll always have as many take-offs as landings, the trick is to be sure you can take-off again ;-)
  • Steve Hicks (N5AC)Steve Hicks (N5AC) Posts: 20
    edited 2006-11-17 15:26
    Just FYI, I have a pretty demanding application from a power perspective and I have a radio transmitting with 7W of output RF, the propeller, a GPS, etc. all off of one regulator, an LD29300.· Before significant heatsinking, I also had resets and as I looked into it, I found that the regulator has an internal thermal shutdown.· The external temp on the regulator as measured with an IR thermo was around 230F.· The regulator has also shut down when the external voltage hit 15V so there's a reasonable amount of protection circuitry in the regulator.· I now have a big "cheese grater" on the regulator and it is running at around 150-160F and is doing fine.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔


    Steve, N5AC
Sign In or Register to comment.