| ??? 03/08/07 18:44 Read: times |
#134589 - Erik, I don't know where your head's wedged ... Responding to: ???'s previous message |
Erik Malund said:
Richard Erlacher said:
If it happens 1 time in 1000 it's far too often, if it's flash corruption
If it was 1 in 1000 or even one in 10000 I would hear about it. My equipment does not have a power switch, you have to turn the bus off and restart it to 'restart' the uC. Also, if it lost its memory, we would have a service call and asking the service techies "do you ever 'rescue' a failing unit by reloading the program," the answer is "not since we pulled the JTAG TCK pin to ground". Some older boards - not my design - with Xilinx chips do not have a pullup on the TCK, but the TCK pin is next to gnd on the connector and a 'nugget' across these 2 pins solved the problem. Now Richard, you just learned that Xilinx CPLD chips ALSO 'lose their memory', how abot that? I believe you've fixed problems serious enough to be elevated to your level by now. I also have no doubt that you've seen JTAG interfaces incorrectly wired. I've not had to deal with "memory loss" in XILINX CPLD's or, for that matter in LATTICE or ALTERA CPLD's. I've not had to deal with FPGA-booters that were corrupted. If you say it happens, I'll look into it. This is the first time I've encountered such a claim. If you could get rid of your supervisorfobia we could end this.
No, you're imagining things, Erik. I've got no problem with supervisor chips. If you want to use one, then go ahead. Clearly there's some problem, and if using a supervisor IC helps, then, by all means ... go ahead! As for "ending this," it's YOU who turned what I hoped would be a productive discussion of a perennial problem into an argument with apparently only one "correct" solution. Since you're so firmly persuaded that there is only one solution, that being a "proper" supervisor, please tell me, and everyone else, which supervisor IC that would be, and why, exactly, you believe it to be the perfect solution. Be careful, though, because you frequently criticize others for saying, "Well, it works for me ..." so you have to provide some basis in fact, as provided by an informed third party, and not just random conjecture based on your limited sample. The argument you have pushed with various 'attachments' can be condensed to one thing "the bad, bad chip makers have made chips that require a supervisor. Since that is so I will push the argument by adding a lot of extraneous crap to make my point" All I've attempted to do is to isolate the various potential problems so this could be addressed systematically. Instead, you want it to be an argument over who's right and who's wrong. If that's all that matters to you, then go ahead! Trumpet on about how clever you are. I've merely proposed that an attempt be made, mostly by me, but with some input from others on this forum, to isolate the various potential faults that could contribute to this problem. I don't pretend to have all the answers, but I do have the desire to DO SOMETHING to expand the available information by means of long-term, carefully designed, stringently controlled and monitored, testing. Since you're unwilling and/or unable to do anything along this line, why do you go on and on to oppose it? Are you afraid that the facts that arise from such a study will prove you wrong? There's nothing to win or lose here ... no prize. I'm just offering to generate some information that you're free to use or to ignore. Why is it you want to turn every simple discussion into an argument? Don't you get enough attention at work or at home? Were you raised by parents who constantly chided you for being "wrong" or stupid? Can't you be a constructive participant in a discussion rather than simply being a spoiler? The fact is that when you use a proper supervisor there is no problem Again, I have to wonder on what documented facts you base this blatant assumption. Why not tell us, as I asked before, which supervisor is "proper" and what published basis you have for making such a statement. I see the current use of "supervisor" IC's as a band-aid for a problem not addressed by chip makers all too anxious to field FLASH memory in their, by now, static MCU design based on a reset intended for the previously dynamic architecture. If they'd been clever, they'd have redesigned the reset and clock circuitry. However, they were probably in a hurry. The original Intel design apparently had this reset flaw from Day 1, but it had no impact until program memory became volatile. By that time, Intel was no longer concerned with the part, since they were pushing what they thought was a better series. So why do you keep adding more stuff? is it trying to win an argument you have already lost? What is this obsession you so frequently demonstrate? Can't you have any kind of discussion without someone winning and someone losing? I was proposing that I add a function in order to establish whether it helps with the problem of ensuring that the MCU doesn't run away and "tinkle" on the program store during the power-off transient. Since we have no published data on which to base an evaluation of whether the reset IC is impacted negatively by the same stimuli that cause the MCU to misbehave, the observed "improvement" is clearly heuristic and not rigorous. It is lent considerable credence by the wide range of observation that support it, however. Since, as Lynn pointed out, the "reset" is really an NMI, holding the MCU's RESET pin high may very well not generate the effect one might otherwise assume. AS a result, I fear that the typical supervisor doesn't protect against the random write or erase of program memory during the power-down transient, which is its most vulnerable phase of operation. That some of the modern chips (oh you would not know about that, would you) have overcome this by having a functioning reset circuit internally is progress, not "showing how bad the others are". If you didn't blare on about your random opinions all the time, perhaps people would be able to go ahead and suggest MCU's for the test I've proposed to perform. Your favorites are, of course, inelegible because they're not offered in the most popular packages and supply voltage. Perhaps eventually, I can put a set of them on an adapter to the most popular footprints, but for now, I can't use them. Several popular manufacturers offer current DIP-40- and PLCC-44-packaged 805x's and, since it is easy and these packages are popular, I want to start with them. Erik RE |
