On Sun, 20 Mar 2005 "peng3002" <peng3002@...> wrote:
some datasheets.
They've got both the Motorola (as ON Semiconductor, which was "spun
off" from Motorola when the latter wanted to ditch the 'jellybean'
business) and Fairchild datasheets.
Motorola/ON's part shows a Schmitt trigger on the clock input, and
the specs say that the allowable rise and fall times are "unlimited".
Fairchild's part does *not* have the Schmitt trigger, so its
clock input must rise or fall within 5 microseconds. That's asking a
lot from an LM358 with a 15V swing! No wonder you were seeing a
100% failure rate with these parts.
As for why half of the Motorola parts failed, maybe they made some
without Schmitt triggers? Do you have the date codes handy?
Moral of the story: digital circuits like to have nice fast edges
(but not *excessively* fast, as that leads to ringing, power
supply noise, crosstalk, and so forth), particularly on clock lines.
And *especially* on clock lines which drive more than one chip,
as otherwise one chip will switch sooner than its neighbours, and
all kinds of mayhem will ensue. I've learned this the hard way.....
Oh yeah, the www.alldatasheet.com website doesn't really have *all*
datasheets, but they do seem to have 'em for all the common parts,
ICs as well as discrete semiconductors.
- Colin Hinz
Toronto, Canada
> I got the main sequencer board soldered up. Powered it up and weirdHmmmm. Let's go to the glorious www.alldatasheet.com and look at
> things started happening. Output 3 permanantly on, output 4 never
> turns on, and output seven is skipped (by skipped I mean it jumps
> from out 6 to out 8 with not even a pause on 7). Thought what the...?
> After much head scratching I decided to breadboard the circuit. Just
> the 4017 and half the LM358. No reset or End Out circuitry. Fired it
> up and this time I got nothing on output 2 and output 7 is still
> skipped over. Hmm...I replaced the transistor (bad one) for output 2
> and that problem was fixed. Tried different chips: seven different
> Fairchild CD4017BCN all had the skipping output 7 problem. Out of ten
> Motorola MC14017BCP five worked perfectly. Problem solved, now all 8
> outputs work properly.
> Out of 17 chips I only have five good ones? I found this hard to
> believe, so I re-installed one of the bunk Fairchild chips and
> started swapping out different LM358s. I tried Fairchild and National
> brands, and also LM2904. No change. Then I tried a TL072. Fixed it!
> Then I tried TL082, BA4558,MC33077. All worked. It seemed any split
> supply opamp worked. With the LM358 replaced by a TL082, I started
> swapping out 4017s. Every one of the seventeen 4017s I have worked
> perfectly!
> Can anyone explain this ? I'm just starting to dig into CMOS stuff
> and this really threw me. A nice thing about this discovery is maybe
> I can finally get my Gated Comparator working properly since it
> exhibits the same kind of skipping behavior.
some datasheets.
They've got both the Motorola (as ON Semiconductor, which was "spun
off" from Motorola when the latter wanted to ditch the 'jellybean'
business) and Fairchild datasheets.
Motorola/ON's part shows a Schmitt trigger on the clock input, and
the specs say that the allowable rise and fall times are "unlimited".
Fairchild's part does *not* have the Schmitt trigger, so its
clock input must rise or fall within 5 microseconds. That's asking a
lot from an LM358 with a 15V swing! No wonder you were seeing a
100% failure rate with these parts.
As for why half of the Motorola parts failed, maybe they made some
without Schmitt triggers? Do you have the date codes handy?
Moral of the story: digital circuits like to have nice fast edges
(but not *excessively* fast, as that leads to ringing, power
supply noise, crosstalk, and so forth), particularly on clock lines.
And *especially* on clock lines which drive more than one chip,
as otherwise one chip will switch sooner than its neighbours, and
all kinds of mayhem will ensue. I've learned this the hard way.....
Oh yeah, the www.alldatasheet.com website doesn't really have *all*
datasheets, but they do seem to have 'em for all the common parts,
ICs as well as discrete semiconductors.
- Colin Hinz
Toronto, Canada