Prop at 100 MHz !

Rayman

Got my 6.25 MHz crystals today.· Went to desolder the crystal off the proto board when it suddenly just popped out [noparse]:)[/noparse]· I forgot that it was just plugged in!· Anyway, didn't even trim the leads on the new crystal, plugged it in and:· Well, the TV screen was all distorted.· Tried the "VGA_Demo" and got a blank screen.· Then, just changed 80_000_000 to 100_000_000 and:· It works!

This opens up some possibilities for me because, with 4 cogs, I can get an even 10-ns timing resolution in some apps I have in mind...

Update
I'll attach some photographic evidence for any non-believers out there.· I'm using the counter module to toggle PIN7 of the demo board at the clock frequency (code posted here:· http://www.rayslogic.com/propeller/PulseGen.htm).· Just hooked up a scope to measure the response.· (The signals aren't very pretty, but you can see the frequency).

Post Edited (Rayman) : 10/15/2007 1:11:16 PM GMT

Oldbitcollector (Jeff)

Wow! Neat stuff.. Keep us posted on the results... Also, I never realized the crystal was just plugged into the protoboard?

Oldbit

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Buttons . . . check. Dials . . . check. Switches . . . check. Little colored lights . . . check.

— Calvin, of 'Calvin and Hobbes.

Bryan K.

Have you ever done any testing at 128MHz? I have so far topped at 80, but i am thinking of getting a new crystal for my Propstick. Any thoughts?

deSilva

This matter has been discussed in utmost detail some weeks ago. As you guys do not read older thrads I shall summarize:
(1) Everybody does this this kind of tuning; it works under certain circumstances you should understand
(2) This is definitely and absolutely out of spec
(3) When using the PLL you have to understand that "x4" or "x8" means "x16/4" or "x16/2": The PLL boosted frequency is ALWAYS x16 of the crystal. When this frequency is to high it does not help to reduce to PLL1X The basic output of the PLL is dirty in the first place; there is no "locked phase" and - if ever - will clock at the crystal frequency.
(4) Using PLL8X seems to have some advantage, as the PLL signal is refreshed by the devider.
(5) Power consumption increases quadaratically with the frequency. (see Dasheet, that gives values upto 100 MHz)
(6) It seems best to not use the PLL but the simple frequency e.g. from an external oscilator

Rayman

From a previous discussion, a Parallax person (I forget who) said that this 6.25 MHz crystal will work just as well as an external 100 MHz crystal. Plus, this is a passive device whereas I think the 100 MHz oscillators need power and a capacitor. Not to mention that this is literally a drop in replacement for the Proto board!

Rayman

Hmm... Hit a snag! The TV demos (graphics, graphics_palette) work, but come out as black&white!

Rayman

Bryan: I think somebody said that ~100 MHz is the limit when using the PLL. But, I'd guess if you use an external 128 MHz oscillator, you might have a shot.

(I'm not really terribly interested in making the Propeller go faster, I just wanted an even 10.0 ns period! It was so close at 12.5 ns that I had to try it...)

deSilva

You can try a 12.5 MHz crystal at PLL8X. Saphieha is sure that that works more reliably. And add a 100nF stabilizer near your Prop in any case. And even better one at each side!


The PLL works fine at 160 MHz, most likely in many cases - but far out of spec - also up to 200 MHz, which gives my 12.5 MHz crystal suggestion a fair chance

Post Edited (deSilva) : 10/13/2007 2:35:19 PM GMT

adrift

Is NTSC/PAL output color burst freq. related directly to this clock timing?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔


Regards,
Jim

Post Edited (adrift) : 10/14/2007 3:13:18 PM GMT

Duffer

The uOLED-96-PROP display from 4D Systems uses an 8MHz crystal and I run the PLL at x16 (128 MHz) for days on end without a problem. I'm not using it to generate video, so it may not be germain to·some of this discussion, but I just thought I would throw it out there for those that are interested in what's possible. The main consideration for choosing the 8MHz crystal was the size of the component (there just wasn't room on the back of the 96-PROP for the larger 5MHz part).

Duffer

deSilva

Duffer, have you checked that it really runs @ 128 MHz, e.g. by outputting a 1Hz signal e.g. using WAITCNT(64_000_000) ?

Duffer

No, I haven't done that. And I haven't put a scope on it either. I do know, however, that the display draws text (I'm working on character and text string functions) nearly twice as fast with a PLL16X setting compared to a PLL8X setting. Is it REALLY working at 128 MHz or 125.83287124763 MHz or some other number? I don't know, but it's a lot faster and it seems to work reliably.

Duffer

deSilva

Video is the most susceptible check... So when you said you didn't use video I became suspicious I do not doubt that PLL8X is slower than PLL16X, but at what clock?

Rayman

I just added some data to the original post!

deSilva

@Rayman: It is not a matter of belief or non-belief...
All propellers run @ 96 MHz, but not when you apply a 48 MHz crystal and use PLL2X
I have a Prop running @120 MHz but a second one does not.
At higher frrequencies some parts may work and some may not: the counter PLLs used also for video are quite susceptable.

As the internal co-operation of all parts is nor always well understood by some bold tuners, I always recommend reservation.

A good simple check is this

CON
  myFreq = 100_000_000 ' insert here the frequency you think your prop is running at; don't use CLKFREQ !

PUB main
 DIRA[noparse][[/noparse] 0 ] := 1
 REPEAT
    WAIT(CNT+myFreq/2)
    !OUTA [noparse][[/noparse]0 ]

An LED connected to pin 0 should be half a seond on and half a second off. No special equipment needed.

Rayman

deSilva: My "non-believer" comment wasn't aimed at you! (at least, not directly[noparse]:)[/noparse]

Ale

Rayman,

You show us a nifty, almost unaffordable (for hobby) LeCroy DSO ! How do you think that affects our sleeping ?, now I have dreams... I see things... (well I always did ). A prop at 100 MHz ?, well that is cool, no doubt, but so far... I have not much use for it... well maybe a cheap 10 MHz DSO... mmmm I could improve mine (simple propeller based 2 MHz DSO).

Rayman

I think it could make a 100 MSPS DSO !

PS: Can you believe we just got 54 of those scopes at work for free? (off government surplus...)

lnielsen

Do you have a spare scope you want to get rid of??? I always wanted a good data scope
Do you have a·link to the surplus?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
BioProp: Robotics - Powered by Bioloids and controlled by the Propeller

Ale

Rayman, I *also* always wanted a good data scope, makes me also eligible for one :-D ?

Rayman

I just noticed that if you use the ENC28J60 ethernet controller with the recommended 25 MHz crystal, you can get a 6.25 MHz clock output using the prescaler. One could use this to drive the Prop at 100 MHz...

Or, I suppose an old 25 MHz oscillator and some divider logic can give you a 6.25 MHz clock...