48 MHz oscillator problem

russk2txb · Joined: 30 Mar 2008 Posts: 109 Location: New Jersey

Hi all. Wonder if anyone has any suggestion for this problem. I had a new board made, using a PIC18F4553. This is a PIC that can do USB. To do fast USB it needs a 48 MHz clock. I used a 20 MHz crystal, 15 pf caps, and a 330 ohm resistor. The clock statement is:

Ttelmah · Joined: 11 Mar 2010 Posts: 19552

HSPLL, does not mean 'use the PLL' on a USB chip. The PLL is always enabled for USB. It means take the CPU frequency from the PLL.
Have you tried:

HSPLL, PLL5, CPUDIV1, USBDIV, VREGEN, NOXINST
#use delay(clock=48M)

In fact USB _will still work_, with HS, but you still have to add PLL5, to say that the USB section should use the /5 divider.

So HS, PLL5, CPUDIV1, USBDIV, VREGEN, NOXINST
#use delay (clock=20M)

is also legal, running the CPU at 20MHz, and the USB at 48MHz. Both assuming you do want the USB VREG enabled internally (you are not providing 3.3v externally). Obviously with MCLR, or MOMCLR, according to what you are doing with the MCLR pin.

Obviously, cap and resistor, depend on _your_ crystal, and it's gain. Try reducing the resistor, if the gain is not high. 50R, is fairly common for some 20MHz crystals.

Best Wishes

russk2txb · Joined: 30 Mar 2008 Posts: 109 Location: New Jersey

First off, thank you Ttelmah. I used:

gpsmikey · Joined: 16 Nov 2010 Posts: 588 Location: Kirkland, WA

Assuming you installed in the standard location, in the C:\Program Files\PICC directory, you will find a text file called "fuses.txt" that lists the fuses and a brief description. Thanks to whomever it was that pointed that out some time ago when I had the same sort of question Laughing

Hope that helps

mikey
_________________
mikey
-- you can't have too many gadgets or too much disk space !
old engineering saying: 1+1 = 3 for sufficiently large values of 1 or small values of 3

FvM · Joined: 27 Aug 2008 Posts: 2337 Location: Germany

Basically, you'll find all informations about fuses in the Microchip datasheet of your processor.
In some cases you may have difficulties to refer CCS fuse names to Microchip ones, as a last resort,
you can utilize MPLAB to figure it out.

russk2txb · Joined: 30 Mar 2008 Posts: 109 Location: New Jersey

Thanks gpsmikey and FvM. I didn't know about the fuses.txt file either, and glad to find it. Very useful.

Now I am wondering about NOXINST. It disables the extended instruction set. Why do this? It seems to me that if a program uses the extended set, then it is used and of not, not. And is there a disadvantage to having it enabled?

FvM suggested MPLAB, and I have been wondering if I should invest in that. I only use the CCS tools so far, but I find the debugger very problematical. It is tantalizingly good, but suffers from a lot of bugs that sometimes drive me crazy. I have been wondering if everyone else notices the same problems. Is MPLAB a lot better, easier to use, or what?

(For an example of the problems in the CCS debugger, try reading the value of a global structure, or it's address. Usually results in an error message. The thing is that it also sometimes works, so I can never know whether I can depend on it. Often when I try to read the value of a variable it gives wrong value. Sometimes with mouse over I will get a execution error when the mouse happens to be over something it doesn't understand. Often the stepping does not work, the debugger steps to the start of the program. If you try to step again, it either just runs, or else will step into one of the interrupt routines. Just frustrating!)

Thanks again, Russ

FvM · Joined: 27 Aug 2008 Posts: 2337 Location: Germany

MPLAB is free, and it would be already worth installing it, if you only use the simulator for some tests.
I'm using MPLAB with Microchip programming/debugging adaptors.

My assumption for XINST would be, that they are unsupported by ´CCS C.

Ttelmah · Joined: 11 Mar 2010 Posts: 19552

Enabling XINST, changes the behaviour of some standard instructions. CCS, does not use XINST, and if it is enabled, the code will almost certainly fail to work as expected. Do not enable it, if you want CCS code to work.

On MPLAB, it's simulator, while limited in some ways, is the best one around. There seem to be some oddities/bugs with the latest version, but probably 95% of people here will have MPLAB, and most 'old hands', _will_ use it in some way, even if only to look at fuses etc..

Best Wishes

bkamen · Posted: Sat Jan 29, 2011 8:02 am

russk2txb · Joined: 30 Mar 2008 Posts: 109 Location: New Jersey

Thanks Ttelmah, I wonder if not having the ex inst set disabled has anything to do with the unstable operation of the CCS debugger. I will certainly have to try that.

And I guess I better get MPLAB and give it a try.

Thanks to all,
Russ

bkamen · Posted: Sat Jan 29, 2011 8:28 am

CCS doesn't support the extended instruction set - period.

NOXINST is what you want and is enabled by default (from what I've seen)

-Ben
_________________
Dazed and confused? I don't think so. Just "plain lost" will do. :D

temtronic · Posted: Sat Jan 29, 2011 8:34 am

If you don't already know when using MPLAB as a 'conduit' to burn PICs with a Picstart Plus programmer(or others), you'll have to change the build option from 'debug' to 'release' and recompile the code THEN download into your PIC.
Currently the default is 'bebug' and cannot be saved as 'release'.
MPLAB dev. team has told me this will be a user option in the next release of MPLAB.When he called , he never knew that anyone programmed PIC the way I do.