Roman Black wrote... >Dave Dilatush wrote: >> >> Andy Fenstad wrote... >> >> >I have bulit a small circuit with a 12C508 that is >> >intended to run continuously. The chip runs an >> >endless loop. >> > >> >If I had built this with 'conventional' parts (ie. >> >555 timer and gates), I would not be worried about >> >it running for years and years. >> > >> >Does the fact that a PIC is a processor shorten its >> >life? >> >> Nope. >Dave, with the greatest of respect, i'd like to call >you on that one. I think there are significant reasons >the PIC circuit is inferior in reliability ALTHOUGH >I think PIC's are superbly reliable, compared to many >chips (and i've replaced most types of chips over the >years). The original poster asked a very specific question: does the fact that a PIC is a processor SHORTEN ITS LIFE (emphasis mine)? To this specific, narrow question, my answer is still "no": I don't believe there is anything inherent in the nature of a "processor" that necessarily makes it significantly less reliable than silicon which performs any other function. The complexity of the OP's "conventional" implementation is an all-important issue. Would it have involved a 555 timer and just one or two 74HCxx gate packages? Or would it have been more complex, involving a 555 timer and thirty 74HCxx gate packages? We don't know. =20 If the former, I would doubt there is any appreciable difference in reliability between the discrete implementation and a PIC implementation. If the latter, I would have to expect--unless shown VERY strong evidence to the contrary--that the PIC solution would be much, much more reliable simply because of the large disparity in component count. The above presumes--as did my original, rather glib "nope" answer to the OP's specific question--that both the PIC solution and the discrete solution would be free of designed-in weaknesses caused by design error, overstressing of components, problems associated with unconventional or inappropriate application of components, lack of attention to detail during the design process and so forth. And there I would acknowledge that performing a function with a processor usually involves a **LOT** more details--and more kinds of details--than doing that same function with discrete logic. And proper attention to those details requires a broader set of skills on the part of the designer. Your post addresses some of these, as well as some which I don't think are significant problems: >But, the 555 + logic chips solution is superior to >the PIC solution because: >* heat/load spread over more chips Is this really an issue? It seems unlikely to me that whatever the OP is building, it's going to consume more than a few milliwatts of power either way. (Of course, if he were to use a linear 555 chip instead of CMOS, and 74LSxx logic instead of 74HCxx, the discrete solution could consume quite a bit more power than the PIC; after all, some 74LSxx chips actually get slightly warm to the touch!) The OP didn't indicate any environmental conditions that would make power dissipation stand out as an important reliability issue, so I would question whether this consideration is important. >* 555 is very rugged linear chip with good protection > on all pins, Are PICs particularly vulnerable? I'm not familiar enough with their ESD ratings to make a comparison with 555s, but I would expect the ruggedness of PICs to be comparable to whatever logic family the OP would have used for his gates. Aren't they? >* PIC will need to operate at higher total clock speed > (faster/more supply surges, power is consumed in > violent pulses etc) Ah, violence. We all abhor violence, of course, in any form; hence the need for peacekeepers. Like the UN. =20 Or, in the case of violent current pulses, bypass capacitors. =20 Proper supply bypassing is crucial to a successful design, no less in designs using PICs than in designs using any other kind of chip--but no more crucial, either. The original, linear version of the 555 timer was notorious for being particularly violent in this regard, by the way: it had a slow-switching, high-drive totem-pole output on Pin 3 that drew an absolutely HORRIBLE amount of pass-through current during output transitions. When the CMOS version of the 555 was introduced, the drastically reduced current spikes of the new design were always prominently mentioned in the "Features" section of the datasheet. In sum, I would disagree with you on this issue: both designs, PIC and discrete, need proper supply bypassing. When properly bypassed, both can be expected to perform well; when not properly bypassed, both can be expected to have major problems. No matter what one is designing, supply bypassing should not be treated as a trivial matter or an afterthought. It needs to be done with care, to avoid a host of potentially hard-to-troubleshoot problems. >* PIC has many more devices on a similar sized chip > (smaller switching devices, smaller pin protection > devices etc) You might have an issue here, though if the total amount of silicon is roughly the same in each case I would not expect a big difference in reliability to result. In the end, it's the reliability data that speak to the issue. >* PIC has possible software faults as well as hardware I consider this a quality-of-design issue, not a reliability issue. This consideration points to the need for very careful software design, review and testing to avoid the kind of hidden "bugs" that rear their ugly heads long after a product is released. >* Micros much more likely to get jumped to wrong part > of code if there is a power spike etc They may, but on the other hand, I recall 555 timers being very susceptible to false triggering under the same conditions. This takes us back to the issue of supply bypassing again--and power system integrity, in general. A clean, glitch-free power system design is vital, no matter what kind of electronic device you're designing. >* Crystal and resonators known failure problems > (rc osc or 555 is more reliable) This is a very good point, and I would expect it to be especially important where there is a lot of mechanical vibration. >So although I do admit that a well designed PIC (or >other brand micro) app will work for 20 years and >the 555 app will also work for 20 years, there are real >reliability factors that people should at least >consider when designing the app. The words "well designed" are the key; and there is a lot more designing to attend to when you're dealing with processors than when you're not. Dave -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email listserv@mitvma.mit.edu with SET PICList DIGEST in the body