clock 328 for MWS

SZTRAD · Joined: 30 Dec 2019 Posts: 165

Since the topic is closed, I will answer my MWS colleague here. I think it's quite important and for some people it could solve unexpected problems.

From MWS
Flash is prepared to run with higher clock, after all the controller runs up to 20MHz.
If executing single instructions only, that's 20 million commands per second, the flash must be able to fetch instructions in that speed, 14 MHz will do no harm to flash.

Yes you are right flash is able to work up to 20MHz (even a little more), but this does not apply to the internal signal source. The catalog sheet quite clearly states that I quote "Note that this oscillator is used to time EEPROM and Flash write accesses, and these write times will be
affected accordingly. If the EEPROM or Flash are written, do not calibrate to more than 8.8MHz.
Otherwise, the EEPROM or Flash write may fail. "
Since the fuses are part of rewritable memory, you may not be able to write to it again from the program. I didn't write anything about the processor being damaged. It just freezes and you don't know why.
Again, I'm probably a little paranoid, but these attempts don't work well.
I'm not saying it's not possible, I'm just pointing out that it may not be reliable.
Just for fun, about 5 years ago I tried it and the 328 runs up to 25MHz on an external crystal and without any problems at 11.5MHz on an internal RC. I think it's not so much due to the speed of the memory but rather the edges of the clock signal, where the RC time constant will be so small that it will not be able to keep the input values of the schmitters (or whatever they have there) to match the edges of the signal.
RS

(BASCOM-AVR version : 2.0.8.3 )

MWS · Joined: 22 Aug 2009 Posts: 2262

SZTRAD · Joined: 30 Dec 2019 Posts: 165

So we really don't understand each other.
I understand how it works and I do not argue that after the reset everything returns and the processor works normally.
I'm just saying that if you do it repeatedly while the program is running, it can freeze. And a reset is not the solution in case of unexpected outages. We differ in our point of view, that's all. Just as the solution is not to go to parameters outside the manufacturer's recommendations just to the point that I need to generate the SPI bus faster. If the speed is not enough for me, I reach for something faster or I solve it differently. I hope I agree on that.
I have experience because I test the limit catalog values of the used components before their deployment.
How it works in tests. Buy 10 samples from different sources to have different batches. And they are tested for the given parameters. The clock is currently being tested in great detail for external interference and stability, including the temperature profile. A statistical sample is obtained from the 10 measurements and the given parameters must not be exceeded in the projects under any conditions.
I do the test in reverse reverse the frequency up and generate the output signal per port. since it works in interrupt so i can see when it dropped out.
I apologize for my English, the problem of the translator and my description seems to be known here.

MWS · Joined: 22 Aug 2009 Posts: 2262

SZTRAD · Joined: 30 Dec 2019 Posts: 165

In our language, this is exaggeratedly called "tearing off an identity card"
Yes you are right in my posts no but in yours yes.
I quote
Can make the difference between a 'tad too slow' vs. 'good enough'.
Only make sure you switch it back to default before any EEProm access.
If I understand correctly you mean switching at runtime. Or am i wrong?

And I just pointed out that he may not be able to change it back because he won't be allowed to write.
Yes, I didn't point out that it could reset from this, but I also didn't say that the chip would be damaged.
You know I do a lot with children and young people (I return to the community what was provided to me a long time ago) and since they are learning I try from the beginning to forget about the internet and the vast majority of arduino constructions where they just put it together and copy the program and it somehow works but they were thinking about what the manufacturer wrote. If it goes well and at least a few of them understand that the manufacturer wrote it there for some reason and it is necessary to follow it. I followed the master of the original interviewer and I know that he wrote that he was an older person (he wrote it somewhere) so I pointed out that a problem could occur.
Thanks to the Internet, it is read by more people and many of them buy a module from China or a processor from a store, use a scheme from the Internet, read that they can hack it and then are surprised by the result.
That's why I'm used to saying b) when I say a).

I quote
Means also I would avoid harsh changes in OSCCAL, as the R / C may stutter then. I would not do a hard switch from 255 to 128 and back, I would step up and down in terms of 1s or 10s in a loop.
You didn't say that in the original post.

You have to keep in mind that not everyone knows the catalog sheet to the last comma. Quite often I hear the opinion that the catalog sheet or instructions are read last because they have a lot of pages.

Before it degenerates into an exchange in Mark's previous closed post, I'd end it. It is probably clear to the two of us what is going on, and in my work I will not allow myself to do something that someone might find in the future as a mistake of judgment and processing.

From my point of view, I would not choose this display for these series of processors. If I were already forced to solve this AVR, I would choose xmeg with external memory and I would copy the memory image using DMA and SPI. When I ever have time (now it looks like I'll have enough of it) and I will rewrite it from c to bascom. I've been forced to use c for a long time now, and I have a lot of libraries written in c.
Since I'm slowly returning to bascom now and I have to say that Mark has worked a lot on it over the years and I expect your help. I used bascom for simple purposes and never lost its beauty of simple language where one does not have to worry about what it does but how it works as a result.

MWS · Joined: 22 Aug 2009 Posts: 2262