Will the Prop run fine with a 5Mhz resonator instead of 5Mhz crystal?

RobotWorkshop

In all the examples i've seen using the Propeller chip I see that most people are using a 5Mhz crystal to run the prop. The only excepion was the module from DLP design that fed in 6Mhz instead. Has anyone tried using a 5Mhz resonator instead? If so, how well did it work? Just curious as to what all the options are.

Robert

Mike Green

Can't help you with personal experience, but the Propeller is designed to use either a crystal or a resonator. The reason for using a 5MHz crystal is that it's generally accurate and stable and the 5MHz value multiples to the maximum spec'd Propeller clock speed (80MHz) although the Propeller works just fine with a 6MHz crystal and a 96MHz clock. The Hydra uses a 10MHz crystal and a x8 PLL multiplier to get the 80MHz.

RobotWorkshop

Hello Mike,

I must have missed the note about the resonator. I'll probably stick with the crystal!

Now though, hearing that the prop does run ok with a 6Mhz crystal (96Mhz clock) that does bring up the question as to what other speed combinations are generally used and known to work. Has anyone here tried playing with this thing to see if it will run even faster. A while back there was a thread in the SX forum about successful overclocking of that chip. I'm sure someone is tring it with the prop and it would be cool to hear the results.

Robert

Post Edited (RobotWorkshop) : 2/8/2007 8:19:45 PM GMT

Mike Green

There was some discussion a while ago as part of the basis for the datasheet. Search for forum postings by Paul Baker and Chip. They were able to clock the chip quite a bit faster using refrigerant since power consumption and dissipation rise quickly much above 100MHz, but the chip really isn't designed for running much faster in the sense that the chip becomes much less robust with regards to supply voltage and die temperature. Sure, you can make it work faster, but it's not as reliable.

crgwbr

This thread caught my eye, so I set up a quick LED blinker program to see how fast I could get the prop to run.

400_000_000 Hz· that's right 400 MHz

The prop I was testing it on must be a really good one (I doubt parallax would keep us in the dark if all props could do this).

PS: I did this with a 50 MHz SX resonator and PLL8X

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
NerdMaster
For
Life

Harley

crgwbr said...
This thread caught my eye, so I set up a quick LED blinker program to see how fast I could get the prop to run.

400_000_000 Hz· that's right 400 MHz

By any chance did you let it run for a while (minutes) and note if it ran warmer than usual? (Time to get out the digital thermometer)

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Harley Shanko
h.a.s. designn

cgracey

That means that the PLL would have·been running at 800MHz, but it tops out at only ~229MHz (ambient room temperature and 3.3V conditions). What was probably happening was that the PLL was running at around 229 MHz, and since you selected PLL8X, you were getting half that frequency, or about 114MHz. This is at the limit of the Propeller's hub RAM speed. The cogs, themselves, can run up to 120MHz (actually up to 160MHz, if you don't care about the Z flag), but you won't be able to reliably read the hub RAM at those speeds.

crgwbr said...
This thread caught my eye, so I set up a quick LED blinker program to see how fast I could get the prop to run.

400_000_000 Hz· that's right 400 MHz

The prop I was testing it on must be a really good one (I doubt parallax would keep us in the dark if all props could do this).

PS: I did this with a 50 MHz SX resonator and PLL8X

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔


Chip Gracey
Parallax, Inc.

Post Edited (Chip Gracey (Parallax)) : 2/8/2007 8:37:32 PM GMT

RobotWorkshop

That's a nice description on what the realistic upper end of the propeller can handle.

So, at the moment it sounds like there are at least three common (safe and sound) speed combinations in use for the crystal and multiplier:

5Mhz x PLLx16 = 80Mhz (Propeller DEMO & PropStick)
6Mhz x PLLx16 = 96Mhz (DLP prop module)
10Mhz x PLLx8 = 80Mhz (Hydra)

Any others?

DogP

Mine's running at 104MHz... I ordered various crystals from Mouser, and it just happened that the 5MHz 20pf one ended up being backordered >_<. So, I put in the 6.5MHz 20pf and it's been running fine for the few days I've been messing with it (I've left it on for several hours straight w/ no probs). I haven't really tried anything crazy with it... just ran a few of the demo apps.

I have 3 prop chips, I can try the others if anyone's interested... or doubts that most would run at this speed.

Pat

Post Edited (DogP) : 2/8/2007 8:47:33 PM GMT

RobotWorkshop

Hello Pat,

Sounds like you just squeaked under the limits above. If you don't have any heat related problems and it stable let us know! It would be nice to build a small spreadsheet with all the known working combinations....

I think the list should be broken into two sections. The first would be all the speed combinations that fall completely within the specified limits of the Propeller chip. These would then be considered safe and valid to use for anyone planning on using the propeller in any kind of comercial design or industrial application.

The second half of the speadsheet can be for others that like to tinker with the Propeller and may have their own one off projects where a failure could be tolerated (no loss of life) but in most cases should work just fine.

Thoughts?

DogP

Is there a demo that makes good use of the prop (all cogs running, lots of memory access, etc)? Kinda like the graphics demos we used to run to make sure our overclocked Celeron 300A's were stable .

Pat

paulmac

Could someone please post the specs for an external crystal for the electronics newbies here? (me included)
I know digital/logic but this sudden foray out of the chip and into the analogue world is a bit confusing at first.

I need a bit more than "just get a 5meg crystal and hook it up". For instance I just found from this thread that there is a farad measurement of crystals as well as the frequency.

Also will a 4.9xxMHz crystal run things that are timed for a 5Mhz?

There's a lot I don't know and I have tried to find out by reading the fine manual and google and stuff.

Thanks.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
I stand on the shoulders of giants

KC8DKT

Nutz and Volts 132

"Note that 80 MHz is the maximum internal speed, so don't get the idea that you can drop a 20 MHz crystal onto the Propeller and wind it up to 320 MHz – this is not going to work."

Sounds like it will work but its not going to run the program code any faster???

Mike Green

paulmac,
Most of the crystals in the range of interest (2-20MHz) are specified either for series resonance or for a parallel load capacitance of 12-32pf. You want the parallel load type. If you look on page 29 of the Propeller Manual, you'll see that there are 3 choices for an external crystal or resonator based primarily on the load capacitance. Pick the choice closest to your actual crystal specification, but keep in mind that these are small capacitance values and the length of the wires connecting the crystal to the Propeller's pins may add additional capacitance. Most 5-6MHz crystals work just fine with the XTAL1 setting. The main thing is to place the crystal as close as possible to the Propeller with short leads.

The _xinfreq constant declaration defines the actual crystal frequency assumed by a program. If you use the _clkfreq declaration instead to set the system clock frequency, the compiler will compute the other (_xinfreq) based on the _clkmode value set in your program. If you specify _xinfreq based on the actual frequency stamped on the crystal, programs should accommodate to the change. Demo programs and I/O drivers written by Parallax (and pretty much everything contributed so far) are written to adjust based on the supplied (or compiler computed) system clock frequency.

Phil Pilgrim (PhiPi)

One thing to consider when choosing between a crystal and a ceramic resonator is the frequency accuracy. Propeller installations that implement color NTSC or PAL video output demand a more accurate frequency than a ceramic resonator is likely to provide. But for things like serial I/O timing, a resonator is more than adequate.

-Phil

Dennis Ferron

Phil Pilgrim (PhiPi) said...
But for things like serial I/O timing, a resonator is more than adequate.

-Phil

I wouldn't be too sure about that one either. I was unable to do serial communciations when my Propeller was running off it's RCFAST internal resonator. What would happen is that the lower (earlier) bits would come through, but the higher bits would be garbled. I don't know if the ceramic resonators are as sloppy as the RCFAST resonator, but asynchronous serial does require precision timing. Suppose, for the sake of argument, that 10 bits are actually transmitted down the wire each time (8 bits, parity, maybe a start or stop bit too). If the clock is off by just 1%, the first bit will be sampled 1% off of center, the second will be 2% off center, then 3%, and so forth. The errors add up. If your resonator is off by just 5%, by the 10th bit that adds up to 50% of a bit period, and if you are trying to sample the center of a bit, going one-half bit period off to either side means you miss the bit entirely!

Phil Pilgrim (PhiPi)

Dennis,

'Sorry for the ambiguity. By "resonator", I meant ceramic resonator — not the internal RC oscillator. The BASIC Stamps accommodate serial I/O just fine with their ceramic resonators. But, as you point out, an RC oscillator just won't cut it for serial I/O, unless there's a way to calibrate it.

-Phil

paulmac

Mike, Thanks for the info. Been doing some looking at crystals and after looking hard found a 6Mhz (either that or a 4Mhz). We're geographically challenged here. I'll put it right next door to the Prop, just like in the PE Kit piccys.

Almost got all the bits I need.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
I stand on the shoulders of giants