Underestimated standard clock :)

pik33

To do an experiment, I bought a 14 MHz quartz (they haven't 7 MHz ones) and put it into a demoboard.

KyeDos works
Chiptune player works
wav player works
my one cog nostalgic vga works @ 59 Hz

of course after recompiling with

_clkmode = xtal1 + pll8x
_xinfreq = 14_000_000


40% overclocking without any change in hardware, voltage, etc...

80 MHz standard clock is underestimated. This chip is not even warm.

Heater.

Indeed. Many of us have overclocked the Prop. Of course each chip is a bit difference in in tolerance to overclocking and a lot depends on your layout and decoupling arrangements. I have used 6.5536MHz xtals and 16 times PLL for 104.85Mhz clock rate a few times. That's 30% overclocked without any special care.

Circuitsoft

Microcontrollers don't generally get warm without serious overclocking (4V+ @160MHz?). What you will find, as you clock it up, is that certain operations will get less reliable.

Phil Pilgrim (PhiPi)

There are two frequencies to consider when overclocking the Prop: the net system clock frequency and the PLL's VCO frequency. The VCO frequency will always be 16x the crystal frequency, regardless of which PLL mode you choose. Although your net system clock is a moderately fast 112 MHz, the PLL's VCO is struggling to maintain lock at 224 MHz, which is perilously close to its free-running frequency. I would not count on your setup working over temperature or across different chips. If you want to run at 112 MHz, you will be much better off using a 7 MHz crystal and PLLx16.

-Phil

pik33

I didn't read anything about this pll settings before I put this 14 MHz quartz into a board. They haven't 7 MHz ones in the shop. Now I have bought what they have, something about 7.38, but didn't tried it yet. It gives a crazy 118 MHz main clock, so I will not be surprised if it will not work.

The demo board played all kind of chiptunes and wav files with this 14 MHz crystal several hours without any problems.

I thougt I can use for example 40 MHzx2 to get 80 MHz, then set _clkmode = xtal3+pll2x. Now I know this is not possible with the Propeller.

This PLL thing is strange. When I tried to set high (65 MHz) pixel frequency when experimenting with video driver, I set PLL too high. (@ 270 MHz) There was no picture and I didn't know why. Then when I slightly lowered frqa, the picture reappeared but it was not stable. Then I start to think, what is going on, and realised that I set pll to 240 MHz, and they write, don't go over 128 MHz.

Something is wrong with these specifications (64..128 MHz) - I got better VGA picture quality with 140/4 pll settings, than with 70/2.

And now after all these experiments I can better understand and feel the Propeller chip

jazzed

pik33 wrote: »

80 MHz standard clock is underestimated. This chip is not even warm.

Yes, for individual chips and the current batches this seems very conservative.
However being too aggressive is not good for continuity of supply in a production environment.
For many who post here the need for thousands of over-clocked chips doesn't seem likely.

hover1

pik33 wrote: »

I didn't read anything about this pll settings before I put this 14 MHz quartz into a board. They haven't 7 MHz ones in the shop. Now I have bought what they have, something about 7.38, but didn't tried it yet. It gives a crazy 118 MHz main clock, so I will not be surprised if it will not work.

The demo board played all kind of chiptunes and wav files with this 14 MHz crystal several hours without any problems.

I thougt I can use for example 40 MHzx2 to get 80 MHz, then set _clkmode = xtal3+pll2x. Now I know this is not possible with the Propeller.

This PLL thing is strange. When I tried to set high (65 MHz) pixel frequency when experimenting with video driver, I set PLL too high. (@ 270 MHz) There was no picture and I didn't know why. Then when I slightly lowered frqa, the picture reappeared but it was not stable. Then I start to think, what is going on, and realised that I set pll to 240 MHz, and they write, don't go over 128 MHz.

Something is wrong with these specifications (64..128 MHz) - I got better VGA picture quality with 140/4 pll settings, than with 70/2.

And now after all these experiments I can better understand and feel the Propeller chip

The Data Sheet is your friend:

While 5.0 Mhz is specified as a max, 6.0-6.5 Mhz have played well. Over that, your in uncharted territory using PLLx16. 10 Mhz @ PLL8x works fine, but 10MHZ@ PLL 16 has some problems with some instructions as mentioned before

Parallax sells the 6.25 xtal, so they are comfortable with that at 16PLL.

Jim

pik33

It was an experiment, to learn, not to make a production thing. I think results of this experiment are good: not only overclocked chip works, but I have better understanding of it now.

hover1

pik33 wrote: »

It was an experiment, to learn, not to make a production thing. I think results of this experiment are good: not only overclocked chip works, but I have better understanding of it now.

Experimentation is good! And we appreciate you sharing your experiences!

Jim

pik33

7.37 - works
8.00 - don't work

Somewhere about 120 MHz is upper limit of this chip.

Toby Seckshund

Of the five, DIP40, Props I have one would run like the wind (14.3MHz)and the other four were a bit slower. All seemed happy at 6MHz x pll16 (also 12MHz x 8pll) and a couple took 6.55MHz 16 x pll (all at 3.3 Volts)

I have ramped up the volts on one experiment, to 4.2 Volts without it curling up its toes, but the 9 Volt" test" didn't last very long :-(.

Cluso99

pik33: I am presuming you are using the DIP part??? We have had the DIP running at 120MHz but the QFP seems to top out at 14.318MHz (highest xtal to work, but 7.3x would not work). All tests done at 3v3 and room temp (20C-25C)

pik33

No, this is a Propeller demo board. Now it is playing a wav from SD with a 7.372 quartz put into its quartz socket There is 22C in my room. Wave player was of course recompiled with

_clkmode = xtal1 + pll16x
_xinfreq = 7372_800


Also tested "nostalgic vga" (it is 63 Hz now) , kyedos and chiptune player.

Toby Seckshund

There will always be a few good chips that will do something better, but as stated you couldn't base any form of production certainty on it. The gauling thing, with my lot, was that ***'s law stated that it was my fast one that suffered the 9 Volts "test" :-(

Cherish that one you have.