Propeller or the crystal

Thric

Currently I'm trying to overclock my propeller to 96Mhz using a 6Mhz crystal. I had to do some tinkering and I've switched one of my QFP44 propellers from one pcb to another as well as the crystal. I'm successfully able to download and run programs with a PLL multiplier of x8 but not x16. I was wondering if the PLL could have been damaged by using my hot air gun on the chip, if the crystal could have been damaged, or is the chip just not liking the 96Mhz run speed?
Thanks

P.S. The board has a bypass capacitor at every set of power pins as well as several capacitors near the voltage regulator. Before removing the crystal from the old PCB the propeller there worked like a charm.

Cluso99

Thric: Does your prop work at 5MHz? Since you say that it works at PLLx8=48MHz, then I would not think the problem is the prop is damaged. You must have a proper circuit to overclock successfully. This means the layout is very important. I regularly clock using a 6.5MHz xtal = 104MHz and have also tried 13.5MHz with PLLx8 = 108MHz. 14.318MHz seems to be my successful limit but I have by no means checked everything in the prop works at this speed.

You may need a 10uF bypass cap at the prop chip to help this - try soldering a tantalum across the smt cap. Your xtal pins also need to be short and you should have a proper ground plane.

Thric

The prop did work with a 5Mhz crystal before I moved it to its new PCB. Unfortunately I don't have a 5Mhz crystal that has the form of the 6Mhz one. I tried adding some more/higher value capacitors (4.7uf) but the affects seem negligible. There are both ground planes and pours on my board and I knew that long xtal traces wouldn't be a good idea but I couldn't seem to find the space for it. I suppose some propeller chips have a higher tolerance for overclocking, should I just try replacing the prop with another one or is there another fix that I could try?

Phil Pilgrim (PhiPi)

If the Propeller works at PLLx8, there is no problem with either the PLL circuitry or the crystal oscillator. The reason I can say this is that the internal PLL always runs at crystal frequency x 16, regardless of the PLL mode. The "x8" pragma simply causes that output to be divided by two. So I will have to agree that your problems stem from power supply and/or bypassing issues. What kind of caps are you using, and where are they placed on your PCB?

-Phil

Thric

I'm using ceramic capacitors 0.1uf and 4.7uf.

A pair (one 0.1 and one 4.7) lie right next to the xtal pins. Another pair lies on the bottom supply pins (pin 1 is the top, xtals are to the right, etc.) and a single 0.1uf lies to the left. There are other caps scattered around the board trying to stick with the one IC one decoupling cap rule.

I had the same setup before on another board and that functioned properly (no pun intended )

Cluso99

I haven't found a prop chip that doesn't function at 104MHz and I have shipped approx 100 overclocked boards. Of course, not all my boards are overclocked but I don't check them for overclocking.

You may have to post some photos of your pcbs or the layouts.

Thric

Well I'm not really sure what the problem was but I swapped out the propeller and now everything is running up 96MHz . hurray for hot air! Thanks for all the help.

jmg

Thric wrote: »

Well I'm not really sure what the problem was but I swapped out the propeller and now everything is running up 96MHz

You could check the margin this has, by slowly lowering Vcc until it stops working, and/or warming the device.

Some vendors 'bin' grade their parts, perhaps Parallax do not quite have the volumes for that ?
Others have a 'sweet spot' spec, that tightens Vcc and sometimes lowers Tj, to also nudge up a speed-grade.
This recognizes that FreqMax is actually a Linear Equation combining Vcc,Tj, f

Then there is always Peltier cooling for the serious over-clocker..