Shop OBEX P1 Docs P2 Docs Learn Events
sx overheating trouble ?? — Parallax Forums

sx overheating trouble ??

jesse_in_venicejesse_in_venice Posts: 5
edited 2007-08-27 01:40 in General Discussion
I've created several circuits using the sx-48/52 proto pcb recently, clocked w/ the 50mhz resonator. I noticed an extremely hot sx chip by accident (ouch !!)... tried the code on a second proto pcb w/ out any circuitry attached to the ports. still very hot... reduced the clock to 4mhz & all is cool !!! is this a common prob ?? I added heatsinks to the processor for now, but had anticipated moving forward w/ all future projects @ 50mhz.. am I missing something, or can the speed alone cause heat issues ??

thanks,
jesse_in_venice

Comments

  • PJAllenPJAllen Banned Posts: 5,065
    edited 2007-07-15 00:50
    You hit the nail square on the head.
    SX48currentVSclk.JPG
    645 x 442 - 45K
  • Desy2820Desy2820 Posts: 138
    edited 2007-07-15 06:20
    Self-deleted post.
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2007-07-15 07:33
    The SX can get very warm when run at 50MHz, but I've never had to heatsink one except at 75MHz and inside a sealed enclosure. If it was painful to touch at 50Mhz, there might be something else going on. For one, check Vdd to make sure it's no more than 5V. Also verify, if you can, that your true clock speed really is 50Mhz and not something higher. If everything checks out, and in lieu of a heatsink, you could run it at a lower voltage.

    -Phil
  • jesse_in_venicejesse_in_venice Posts: 5
    edited 2007-07-18 16:55
    I'll check the voltage & see if it's OK... The problem only manifests on the 48/52 proto pcb.... I suspect there is a significant difference in the dissipation between package types.. the 28/28 & tech pcb both behave OK @50Mhz, but the 48/52 actually becomes unreliable &·crashes. I don't HAVE to run @ 50, but since I'm playing & learning, I thought I'd go large... I will get several 20Mhz resonators & try running with them on the 48/52 pcb.

    thx,

    jesse_in_venice
  • CCraigCCraig Posts: 163
    edited 2007-07-19 02:52
    OK, you got me interested, so I pulled out a board and ran it to see. I have no IR meter so here's what I have used over the years.

    -Warm: just that. warm when you touch it
    -Hot: hold your finger on it and after a second or two, uncomfortable. Almost pain but not. No burns. No ouch.
    -Very Hot: unable to hold your finger on it for more than two seconds. Pain. Maybe burns. Ouch.
    -Really Hot: instant burns, power transistors etc. You can smell them burn fingerprint oil etc.

    My proto board, running @ 50, turning two LEDs on and off as fast as it could (SXB), got Hot. It was just uncomfortable.

    Not a good test as my finger dwarfs the tiny chip and probably acted as a heat sink.

    Did yours crash by itself? I did notice twice when I touched mine, it stopped running. I'm sure I loaded down the oscillator or something. I was pressing pretty hard.

    Just thought I'd pass it along,

    HTH, Chris

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    I'm not scared of your robot. I'm covered by Old Glory (youtube)
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2007-07-19 04:12
    Here's another thought: Digital inputs that are left floating can sometimes cause excess power consumption and dissipation. An easy way to deal with this in the SX48/52 is to program all unused inputs with pull-ups. This will hold them in a known state. However, the fact that your heating is clockspeed-dependent leads me to believe that this is not the main culprit.

    -Phil
  • CCraigCCraig Posts: 163
    edited 2007-07-19 05:51
    OK, I did some more late night testing. I'll confirm your problems.

    I have two blank 48 proto boards. One was more heat tolerant than the other, but both would fail running @ 50. One took about 15 minutes to fail and the other took over a half hour. So out came the hair dryer and I stress tested them pretty good. I didn't have any freeze spray so I couldn't get them to snap back on, but it was pretty clear. Under the gun, they would fail in about 30 seconds. If you have never used a hair dryer or worst a heat gun on circuit boards.... DON'T. It's something I used to do in my 'tech days' and it can be very hard on parts and boards.

    Now for the good news. Running the same boards @ 20MHz, I got no failures. I did get it to fail with extreme heat but it was way, way above normal operating temps. Above Very Hot. Much more and I would have melted the plastic parts. Hey, it's only a $10 part and in the name of science. Back at normal temps and everything is running fine.

    Some thing's happening in there. The board can run much hotter at 20MHz without failing. I don't think that's unusual. What it does mean, is that we'll have to run them at 20MHz. If you have to run at 50MHz, it will mean more work. i.e. fans, heat sink, whatever.

    Note: I ran all my tests with OSCHS2

    HTH, Chris

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    I'm not scared of your robot. I'm covered by Old Glory (youtube)
  • metron9metron9 Posts: 1,100
    edited 2007-07-20 18:34
    I was just thinking the way I heat sink my LuxeonIII three watt leds is to put a hole in the circuit board and then I use Tputty504, a thermal compound. I then use a piece of double face tape with a hole in it fill the hole and void with the putty and attach it to my aluminum housing. I guess you could do the same for a microprocessor as well as a heat sink on the top side. I have tons of this putty (I had to buy minimum order) It is like thick toothpaste in a syringe.

    I can run three watts with this method, the putty stays put as well under what must be very high heat under that led.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Think Inside the box first and if that doesn't work..
    Re-arrange what's inside the box then...
    Think outside the BOX!
  • CCraigCCraig Posts: 163
    edited 2007-07-20 20:54
    Metron9, thanks for the ideas.

    I've been Googling around to see what the story is with super glue. Some say no, others say ok but try to get some heat sink metal flush with the chip. I'm not sure what that means. I'll try to get my hands on some of the Artic glue stuff. All my stuff can run at 20MHz (or slower) with no problem, so it's no big thing. BUT I think it was Gunther that said in his book: "You have a Porsche there, why are you driving around in first gear?"

    For my stuff, I have plenty of power available. However, I'm in a hot climate, so heat is bad.

    I saw in the Prop forum that they (Parallax) have a new chamber. Maybe we can talk them into testing the boards.

    Chris

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    I'm not scared of your robot. I'm covered by Old Glory (youtube)
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2007-07-20 21:52
    Guys,

    I've used J-B Weld, a metal-filled epoxy, to attach a small finned heatsink to the SX. It worked great.

    Another app involved an extrusion, where it was possible to slide a piece of sheet aluminum in above the SX. A small piece of Bergquist Gap-Pad squeezed between the SX and aluminum conducted the heat away without using any other heatsink compounds.

    -Phil
  • CCraigCCraig Posts: 163
    edited 2007-07-21 00:06
    Great idea Phil,

    I have some J-B weld. I think that would be much better than super glue.

    Chris

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    I'm not scared of your robot. I'm covered by Old Glory (youtube)
  • jesse_in_venicejesse_in_venice Posts: 5
    edited 2007-08-27 01:40
    I gave in and superglued a hacked ( as in hack-sawed) heatsink to it. I've also resigned myself (sigh) to ordering some 20 Mhz resonators. These devices live in a box outside in florida & heat is a real concern.

    thanks for all the input,
    jesse_in_venice

    (I attached a couple snapsohts of the guilty pcb's)
    Picture 696.jpg
    352 x 288 - 22K
    Picture 697.jpg
    352 x 288 - 20K
Sign In or Register to comment.