Shop OBEX P1 Docs P2 Docs Learn Events
Dumb question: Would a heat-sink allow the Propeller to run even faster (and stable)? — Parallax Forums

Dumb question: Would a heat-sink allow the Propeller to run even faster (and stable)?

cbmeekscbmeeks Posts: 634
in Propeller 1
From everything I've been reading, 6.25 MHz crystal (100 MHz) seems to be OK and "sanctioned" by Parallax. I've built a few breadboard circuits at that frequency...seems to be OK.

But it seems that heat is the real limiting factor (unless I misunderstood) for pushing it even higher??

I'm not talking about 500 MHz here. I'd like to drive it with a 7.15909 MHz crystal. 2x NTSC at 16x PLL = 114.54544 MHz (28.63636 MIPS). That's probably not too far out of possibility I suppose.

So would a small heat-sink help with keeping things stable at that frequency? Could it go even higher?

Keep in mind all of my projects are hobby level and I don't plan on selling anything for a very long time.

Thanks!

Comments

  • Heater.Heater. Posts: 21,230
    I run props at 104MHz. Never noticed them make any heat.

    I think it the phase locked loop in the Prop clock that is the limiting factor here. So I don't think a heat sink will help at all.

    For sure many people have been attempting over clocking their Props for years now. 104MHz is about the limit.
  • cbmeekscbmeeks Posts: 634
    Would you say 104MHz would be safe in a production run?

    Not that I have anything near production. But I was thinking 100MHz was sort of the "safe" limit.

    Thanks
  • Cluso99Cluso99 Posts: 18,069
    I use 104MHz on most projects. There are no heat issues here so a heatsink will not help. Cooling the chip might help - I dont' think anyone has tried this. However, I am scaling back to 96MHz because I intend to emulate USB and this makes the emulation easier.

    But you have to be very careful in your pcb layout - proper bypassing and ensuring the voltage supply remains stable under all operating conditions. This means all operating software as it is what is functioning that causes the power requirement spikes.

    Just for reference, the DIP part can actually be clocked higher than the QFP part - not sure why though.
  • jmgjmg Posts: 15,173
    MCUs have temperature and voltage ratings and are spec'd to work in the worst corner case.
    You can gain extra by avoiding the highest rated temperature and or lowest voltage.

    Prop data gives 2.7~3.6 V, and a wide -55°C < Ta < +125°C
    for speed you could experiment with using higher precision regulators, and 3.4 or 3.45v
    The chip self-heats little in typical use so saving 5'C on bench use is not gaining much.

    You could also trial a device like Si5351A, which can clock any MHz up to 200MHz directly and so avoid any possible PLL issues.
    Adafruit have a $7? Si5351A board.




  • TubularTubular Posts: 4,702
    I revisited this the other week trying to speed up P2 FPGA load, using a function generator to sweep upwards, and yes around 6.5 MHz is where things start to play up on the P1.

    However it may be worth bypassing the PLL altogether and try one of these programmable oscillators (eg Silabs). Adafruit has a small module that goes to 160 MHz which I think would be a good place to start
    https://www.adafruit.com/products/2045

  • jmgjmg Posts: 15,173
    Tubular wrote: »
    However it may be worth bypassing the PLL altogether and try one of these programmable oscillators (eg Silabs). Adafruit has a small module that goes to 160 MHz which I think would be a good place to start
    https://www.adafruit.com/products/2045

    Snap, but note the Si5351A can now go to 200MHz with a rating revision.

  • TubularTubular Posts: 4,702
    Right, even better.

    Mind you we'd be doing mighty well to get to 160 MHz

Sign In or Register to comment.