18F4610 at 100MHz ??!!
Moderators: David Barker, Jerry Messina
18F4610 at 100MHz ??!!
I connected an external clock running at 25MHz into the OSC pin, used the PLL multiplier and set the CLOCK = 100.... I completely expected it to either a) not power up or b) catch on fire. I was very surprised to see my entire program boot-up and run normally. This PIC interfaces with 4 SUARTs, ADCs, an ENC28J60 and external non-volatile RAM. So needless to say, there are plenty of places I would expect it to fail. But to the contrary, everything works... AT BLISTERING SPEED! I do a lot of string manipulation and data encryption with this PIC and all I can saw, WOW, it's like going from a C64 to a Core Duo 2!
The PIC doesn't even feel warm to the touch.
Does anyone else overclock their PICs?
The PIC doesn't even feel warm to the touch.
Does anyone else overclock their PICs?
I did a test with a PIC18F4620 with a 20MHz crystal and PLL (which gave a 80MHz clock). Ran it for 72 hours straight with no problems. Test was done just out of curiosity. NOT something I would do with an actual design - best to stay within the manufacturer's specs.
If I really needed the speed, I would run a PIC24H or dsPIC33 at the manufacturer's rated clock of 80Mhz (for 40 MIPS performance).
If I really needed the speed, I would run a PIC24H or dsPIC33 at the manufacturer's rated clock of 80Mhz (for 40 MIPS performance).
Likewise here - 16 MHz resonator with 4x PLL engaged for 64 MHz with an 18F4620.
No signs of overheating - chip just barely above room temperature. Functionality seems unimpaired, at least with the test programs I fed it.
Update -- a 20 MHz resonator fails with this particular 18F4620, but a 16 MHz resonator seems solid. That's 64 MHz or 16 MIPS, which is not a small improvement at all.
Jack
No signs of overheating - chip just barely above room temperature. Functionality seems unimpaired, at least with the test programs I fed it.
Update -- a 20 MHz resonator fails with this particular 18F4620, but a 16 MHz resonator seems solid. That's 64 MHz or 16 MIPS, which is not a small improvement at all.
Jack
Jack, Clifton VA
Maybe the 'K' version is just the same part re-branded. Remember when AMD used to do that? They would make a CPU work at 900 MHz then cripple it to work at only 600 MHz and sell it at a cheaper price. There were some interesting hacks you had to do to get some of those systems overclocked... I remember cutting traces on motherboards and hot-wiring CPU's by drawing traces on them with a pencil.
Kids have it too easy today, it's all done in software. I just saw an ad for a Dell gaming machine that is "factory overclocked". I'm not sure it's technically 'overclocking' if it came from the manufacturer that way.
Kids have it too easy today, it's all done in software. I just saw an ad for a Dell gaming machine that is "factory overclocked". I'm not sure it's technically 'overclocking' if it came from the manufacturer that way.
Last edited by JWinters on Fri Aug 08, 2008 6:10 am, edited 1 time in total.
Yes I remember those days well , the old pencil traces across the L2 and L1 bridges on the top of the chip , where they sued to laser cut a line depending on which sticker was goin on the top Hacking the PSU circuit to deliver a higher cpu voltage. Now it's all in bios ,, my mobo has a NOS interface (assume it means nitrous oxide buzzword boost) you can tweak the core voltage , front side bus , buss multiplier, cpu voltage , ram voltage ect all for windows nowadays.
Don't think K parts are the same re badged tho because they have a different internal oscillator section.
Don't think K parts are the same re badged tho because they have a different internal oscillator section.
If you can read this you are too close
Hello,
Sincerly fiddling with overclocking is not serious for any hardware project. Overclocking from 20MHz to 22MHz or from 40MHz to 42 or 43MHz in order to have a special requirement on frequency to be multiple of something (usually used to be multiple of RTC) can be done without fear. But boosting PIC using such artifacts is really not serious (at least for commercial projects).
If you need more speed, go for a PIC24HJ/dsPIC33 (about 40MIPS - far from the 16MIPS of PIC18) or to ARM or CORTEX-M3 mcu. These later are really cheap, about the same price as PIC18 series (and cheaper in quantities>10000 = less than $1 for Luminary chips), with all other advantages like linear memory, huge number of hard peripherals, Vectorised interrupts (Nested interrupts for CORTEX-M3), ...
If smt packaging is a problem for you, PIC24 exists in DIP28 packages.
Regards
octal
Sincerly fiddling with overclocking is not serious for any hardware project. Overclocking from 20MHz to 22MHz or from 40MHz to 42 or 43MHz in order to have a special requirement on frequency to be multiple of something (usually used to be multiple of RTC) can be done without fear. But boosting PIC using such artifacts is really not serious (at least for commercial projects).
If you need more speed, go for a PIC24HJ/dsPIC33 (about 40MIPS - far from the 16MIPS of PIC18) or to ARM or CORTEX-M3 mcu. These later are really cheap, about the same price as PIC18 series (and cheaper in quantities>10000 = less than $1 for Luminary chips), with all other advantages like linear memory, huge number of hard peripherals, Vectorised interrupts (Nested interrupts for CORTEX-M3), ...
If smt packaging is a problem for you, PIC24 exists in DIP28 packages.
Regards
octal
- David Barker
- Swordfish Developer
- Posts: 1214
- Joined: Tue Oct 03, 2006 7:01 pm
- Location: Saltburn by the Sea, UK
- Contact:
-
- Registered User
- Posts: 7
- Joined: Thu Oct 23, 2008 12:42 pm