dsPIC30F3014 using PLL

[MCUprogrammer]

CCS C 5.113


Hi
How many MHz crystals on the dsPIC30F3014 processor can I clock at 120MHz?
e.g;
#use delay(clock = 120MHz,crystal=20MHZ) gives an error.
How will I do.

I see that the processor is crashing during operation. When I cut off the power and put it back on, the problem goes away. Is that why it locks because I'm using an internal oscillator? What would you recommend to fix this situation?
The locked processor is the dsPIC33EV series processor.
_________________
Best Regards...
MCUprogrammer
_______________________________
Work Hard

[Ttelmah]

120MHz can't be generated from a 10MHz crystal.

120/4 = 30
120/8 = 15
120/16 = 7.5

Then you can use this *1, *2 or *3 (up to the 25MHz max). Or an external
oscillator.

So you need a 7.5Mhz, 15Mhz, or 22.5Mhz crystal to actually get to 120Mhz.

On your 33EV, sounds like an interference or supply problem. Or possibly
something silly like overrunning the stack. Check what the listing file
says the stack size needs to be (remember size is 2* the size given).

[MCUprogrammer]

I have a 24MHz or 25MHZ crystal. Are these useless?

I don't understand exactly what you want to say here.

"something silly like overrunning the stack. Check what the listing file
says the stack size needs to be (remember size is 2* the size given).
_________________
Best Regards...
MCUprogrammer
_______________________________
Work Hard

[MCUprogrammer]

Is this the event you're talking about?

[temtronic]

I was curious about the 'clock configuration,so I downloaded the datasheet.

Using table 20-1..
a 25 MHz xtal can be divided by 2 or /3 THEN PLL will multiply by x4,x8 or x16

25/2=12.5 x4 = 50MHZ, x8=100MHz, x16= 200 MHz

25/3=8.33 x4=33, x8=66 ,x16=132

so the fastest legal speed using 25MHz xtal is 100 MHz

In the good old days ,I'd cut a QB45 program to figure all this out,but alas can't run it on a modern PC so....my numbers were done with paper and pencil.
There's probably a fancy spreadsheet or 'ap' you can get off the web to do the 'math'.

BTW be very careful with PCB traces and layout ! Running any micro at very high speeds MUST have well designed PCB and CLEAN. Also have LOTS of proper 'decoupling' caps.

[MCUprogrammer]

So, am I going to do that splitting?
Or is it enough to write the following code?
#fuses HS3_PLL8 //High Speed Crystal osc (10-25MHz) divide by 3 with 8X PLL
#use delay(clock=100MHZ)
_________________
Best Regards...
MCUprogrammer
_______________________________
Work Hard

[temtronic]

I don't think that will work....
25 /3 x8 = 66MHz, so 'clock=100M' cannot compute.

I'd see if there's an example program in CCS or on this forum....
I don't have that PIC so hopefully others will reply.
pretty sure it has a max speed of 120 MHz,so you need to 'work back' to see what combination of xtal divider and PLL setup works for you.

[Ttelmah]

Your original clock line does all the settings for you. The point is though
that the frequency you ask for has to be one that can be done with your
crystal.
With the 10MHz, you can do 80MHz.
With the 25Mhz, you can do 100MHz.
With the 24MHz, you can do 96hz.

I think the middle one is going to be the one for you.

So:

#use delay(clock=100MHz,crystal=25MHz)

Will merrily compile/work.

If you must have 120MHz, then you need to buy a compatible crystal.

Your stack looks OK. A little tight (I usually work on needing at least 10%
free, since extra stack is used when debugging).
You don't say what compiler version?. There was an issue a few versions
ago, with certain maths overflows crashing the system (in the known
faults forum).
If it isn't a version with that problem, then you need to be looking at
an electrical problem.

[temtronic]

re: 96hz.

hehehe...Sure glad I'm not the only one with fat fingers ! I KNOW it's a typo, after all it is Monday again....well here anyway.

Crazy thing is my Energy control system panels used a 24 Hz clock, yes Twenty four hertz, for the serial communication lines that went up to 15 miles away.

[MCUprogrammer]

thank you for the replies. I wrote the compiler version in the beginning. I guess you didn't notice. I am using version 5.113.

Could I be having a locking problem because I'm using an internal oscillator?
- I pulled down all the empty pins that I didn't use with 4K7.
- Since I did not use OSC1 and OSC2 crystal pins, I did pulldown with 4k7.
_________________
Best Regards...
MCUprogrammer
_______________________________
Work Hard

[Ttelmah]

Almost certainly not because of the internal oscillator, unless there is
some timing event (for example you could easily get a framing error
on serial I/O with the internal OSC, and if this is not handled this could
cause a problem).
Have you got the brownout enabled on the chip?.

I quite commonly miss things in what I type. I tend to be doing other
things at the same time, and with three computer screens, a couple of
keyboards and mice, it is easy to miss a character...

It was not plain from your post that the code in the EV processor was
using the current compiler. Could easily be older code showing the issue.

[MCUprogrammer]

#use delay(clock=100MHz,crystal=25MHz)
I am writing this code. output_toggle(PIN_C13); I am writing. I can't see a 2us pulse.
setup_timer2(TMR_INTERNAL | TMR_DIV_BY_1, 49); ///2 us overflow, 2 us interrupt

But #use delay(crystal=25MHz)
setup_timer2(TMR_INTERNAL | TMR_DIV_BY_256, 244); When I write this code, 10ms pulse appears. What is the problem
I toggle every time it enters the interrupt. shouldn't it work?
_________________
Best Regards...
MCUprogrammer
_______________________________
Work Hard

[MCUprogrammer]

No, I think BROWNOUT is disabled. Because I didn't write #fuse BROWNOUT on the line.
_________________
Best Regards...
MCUprogrammer
_______________________________
Work Hard

[newguy]

The .lst file will show you, right at the bottom, the configuration fuses for your project. Don't assume that, since you didn't explicitly set a fuse, that it isn't actually set by default.

[MCUprogrammer]