| ??? 10/12/06 20:47 Read: times |
#126343 - to reset or not to reset Responding to: ???'s previous message |
Richard Erlacher said:
I found, just casually trying a few things, that the reset circuits on my 805x's which work great and have never given me a single reason to curse 'em, using linear regulators, ALL fail to reset reliably when used with small commercial switchers. Huh, welcome to real world... (no offence). Richard Erlacher said:
My 805x's normally run from linear supplies in existing hardware or from "wall-warts" which, of course, demand local regulation, which is provided by means of an LM340 or the like. Steady-state power-gnd noise seems to be in the 30 mV range with a linear supply, while the statically loaded switcher produces a minimum of 50 mV of about 5 KHz switching noise (actually 50 mV of ripple with a 200 mV spike at the peak of the ripple. Actually, I suspect that it's not only the noise/ripple which makes the micros unhappy, but also the switchers tend to do various weird things at startup. Try to look at such startup of various switchers under various load conditions, you might be unpleasantly surprised. Richard Erlacher said:
The spec's arent terribly useful in setting up a reset circuit for the 805x, as they generally refer to "two clock cycles after the oscillator is running" and don't tell you how long before the oscillator starts after Vcc is valid. Yeah, that's true. A few internal resets incorporate a counter from the oscillator, but I know of none external which does so. However, if a crystal oscillator won't start within 100 ms or so, it's most probably broken, isn't it (or running on the edge, which is almost the same). If you require faster startup, sorry, special treatment is needed. Richard Erlacher said:
They also don't tell you much about required rise and fall times on RESET It's usually Schmitt anyway, so it shouldn't really matter. A reset IC will produce you nice and smooth Richard Erlacher said:
nor do they discuss thresholds on Vcc. Yeah. Sure. Seldom they do. Richard Erlacher said: Don't do that. Get a reset IC - 7705 would maybe appeal to you, it's a classic - and relax. That piece of black asphalt does not deserve more... :-)
I suspect that one could learn a lot about this in an afternoon with a comparator on Vcc and a GAL programmed to produce exactly two cycles-long RESET pulses. One of these days, when all my systems are tied up running simulations, I may just do that. Maybe a comparator that switches Vcc on only after it is at the requisite level is what's called-for.
Richard Erlacher said:
Jan, have you got a good treatment of RESET requirements from one or more of the chip-makers? How about the precise requirements of Vcc behavior on power-up? It appears to me, that these are exactly the topics most '51 makers want to avoid. ---- But, let's face it - '51-s are DIGITAL circuits, and reset and stuff is ANALOG. Mixing the two makes things complex and expensive. Yes it does. There is a huge pressure on the manufacturers to include them, they also try but obviously they often fail. So, simply, let's accept the status quo and DO use the reset IC. Jan Waclawek |
