who has tried over clocking the at89c51 in KEIL C? how fast can it go? 10% 20% 100%? post your results.
"going nowhere fast", and rude...
Ah. I happened to put pepper in the coffe instead of sugar.
Easy solution: get a new cup.
More interesting solution: try to figure out some chemical compound that will neutralize the pepper without affecting the coffe taste.
What is cheaper? Admit that the wrong processor was used and switch?
Ignore the issues and invest lots of time and still fail? And still have to switch processor in the end - but now with lots of extra time wasted trying to work around the bad design decision?
Every step further in the development process increases the costs from errors with a factor ten. So it really is imperative to make a good pre-study based on the requirements and make sure that a suitable processor is selected.
That pre-study should have considered board space + pin count + cost for having enough I/O. Often cheaper with a processor with more I/O than with I/O expansion chips.
That pre-study should have considered guestimated needs for RAM, flash, and computation speed.
That pre-study should have considered the difference between maximum optimization using hand-written assembler. Or a larger (potentially a bit more expensive but possibly not so) processor able to run the same functionality implemented in C.
That pre-study should have considered the implications of debugging using a processor with special debug hardware contra solving everything by guessing, flashing LEDs or printing to serial ports.
That pre-study should have considered availability of chips. Availability of example software (for the specific requirements - like amount of TCP/IP solutions for the specific chip). Availability of application notes. How many TCP/IP application notes did you see for your processor?
That pre-study should have looked for reference customers - other people solving simmilar problems. How many have you been in contact with (or read web pages from) that have implemented TCP/IP or display solutions with your processor?
Onverclocking will only guarantee that you will not be able to know what is a bug in your code, or a bad result from the processor not being trustworthy. That is really something you would love when trying to debug your program.
going to another processor is not possible and i have no time for a big learning curve.
Actually what you are trying to do is near to impossible on that old 8051. You really should take a good advice and switch architecture. A big learning curve is the best that can happen to you. You are heading towards the wall of failing now.
you all want to doubt my reason.
And we're all correct, because your reasoning really is every bit as flawed as we guessed it had to be.
Right now you're behaving like the textbook figure who, when asked why they don't sharpen the utterly blunt saw they've been working so hard to fell trees with all day, answered: "I have no time to sharpen that saw because I'm already way behind on the number of trees I have to fell!"
My application will contain [lost of things...]
No, it won't, because the way you're going about it, there's zero chance you'll ever come anywhere close to completing that task.
So i need as much speed as i can get.
No. You need enough speed to do the job at hand. And that's way more than you can get --- from the chip you insist on, that is.
Don't include me in the typically rediculous Bumhard generalisations.
Of course it is possible to do what you want with the chip you want. It may be slow and you may have to add more support hardware to achieve it. You may even find that your at89c51 ends up being the controller of a multi-processor configuration, but it could be done.
Now the question really is whether you are the best candidate for the operation. What do you think?
I rather suspect that speed will be the least of your problems:
The AT89C51 has only 4K Flash and 128 bytes RAM - you are certainly not going to be able to do all that stuff without external memory expansion!
Possibly not even with the full 64K expansions - so then you'd have to be messing about with banking. The learning curve for banking is not trivial...
You'd avoid all of that (and save money) with an appropriate processor choice!
Or why not get a Raspberry Pi? http://www.raspberrypi.org/
The big issue here. Do you want to build one or many devices?
If one - get a development board with ethernet hardware and a prototype area.
If many - get a development board with ethernet hardware and a prototype area. When you are happy: make custom circuit using that CPU.
well a bit roundabout answer is "there has been several threads over the years "why does my chip fail when I overclock"
Erik
Overclocking is similar to running with badly stabilized supply voltages. It might go well. But don't be surprised if it locks up or you get a corrupted result computed.
It might go well. But don't be surprised if it locks up or you get a corrupted result computed.
not only that, but "It might go well" and then days later when some coincidence happens (e.g. two ISRs firing at the same time) some "unexplained bug" may show up.
Overclocking obviously means that what works in the winter can fail in the summer because of a higher ambient temperature. And it matters if the equipment is near a window and the sun is shining on the device or if it is in shadow.
Hey buddy, did you test it? I also want to know how much you can increment the frequency. Please tell me what you found out.