henry@utzoo.UUCP (Henry Spencer) (07/04/84)
Has anybody had reasonable success using the 74S124 as a crystal
oscillator? The TI data book claims it can be done, and has a few
sentences on how to do it, but... When I tried it, I found it quite
impossible to convince the thing to start up cleanly on power-up.
It persisted in starting up in a messy mode that confused my frequency
counter, but from other evidence must have been a very high frequency.
The crystal I was using was 6.5536 MHz. I was using only one of the
oscillators in the S124, with the other disabled. I complied with
the databook recommendations on frequency-control input etc. I was
fairly lavish about decoupling, and my power supply was on voltage and
reasonably clean.
I could generally succeed in convincing the thing to run at 6.5536 by
fiddling with small capacitors and such *after* the thing was going,
but no combination of capacitance would make it start up cleanly. I
don't think I got a single clean startup.
I fixed the problem by switching to the MC4024 (second choice because
it's a little harder to find hereabouts), which -- with the same crystal,
same power supply, and same load -- starts cleanly every time, with no
extra capacitors present. As a minor bonus, it eats much less power.
--
Henry Spencer @ U of Toronto Zoology
{allegra,ihnp4,linus,decvax}!utzoo!henryrpw3@fortune.UUCP (07/08/84)
#R:utzoo:-402300:fortune:28000044:000:425
fortune!rpw3 Jul 7 23:49:00 1984
Don't use a 74S124 for ANYTHING, if you can avoid it. It EATS power,
puts spikes in the power supply, experiences lots of "pulling" from other
noise in the system, etc. Even TI agrees.
Check out some of the newer parts like the '624-9 series. The 4000-series
MOS stuff is also good.
Rob Warnock
UUCP: {ihnp4,ucbvax!amd}!fortune!redwood!rpw3
DDD: (415)369-7437
USPS: Suite 203, 4012 Farm Hill Blvd, Redwood City, CA 94061david@bragvax.UUCP (David DiGiacomo) (07/08/84)
Maybe I'm old fashioned, but I always use 1/3 of a 7404 + 2 resistors + 1 capacitor (OK, if you want it buffered it's 1/2 of a 7404). Total cost = $0.50, low power, reliable startup, etc. If you really want to spend more on it you could use a 74HC04.
henry@utzoo.UUCP (Henry Spencer) (07/10/84)
Various people replied to my problems with the 74S124; a few extracts
will sum up the substance of the replies:
"yeah, you found one of TI's embarassing mistakes..."
"I have tried using the S124 as a VCO, with very poor success...
the chip wasn't very accurate, and it had a tendency to oscillate
based on the parasitics."
"You found the right solution, abandon the part!"
"Don't use a 74S124 for ANYTHING, if you can avoid it. It EATS power,
puts spikes in the power supply, experiences lots of "pulling" from other
noise in the system, etc. Even TI agrees."
Several people observed that the LS624 and friends are TI's preferred
alternative to the [L]S124 these days, and that the 624 works quite
well and is readily available. I may try one.
--
Henry Spencer @ U of Toronto Zoology
{allegra,ihnp4,linus,decvax}!utzoo!henryhenry@utzoo.UUCP (Henry Spencer) (07/11/84)
David DiGiacomo observes:
Maybe I'm old fashioned, but I always use 1/3 of a 7404 + 2 resistors +
1 capacitor (OK, if you want it buffered it's 1/2 of a 7404). Total
cost = $0.50, low power, reliable startup, etc. If you really want to
spend more on it you could use a 74HC04.
The trouble with this is that TTL gate oscillators are *known* to be
prone to startup problems; my understanding is that you have to get a
good deal fancier than just 2R+C to get something that will start reliably
under all conditions of chip variation, temperature, voltage, transients,
etc. I agree that they usually work all right, and I've used them for
some things, but I would hesitate to design them into anything that was
intended for "production" use.
Dunno about 74HC. CMOS gates work much better as linear circuits than
TTL gates, and I seem to recall that gate oscillators are considered
respectable in CMOS circles, but HC is a slightly different animal and
I would have to look at specs. (I've given up keeping track of the
sixteen different kinds of fast CMOS, so I don't remember for sure just
what the logic levels etc. of 74HC are.)
Can some of the more knowledgeable hardware types comment on either of
these issues? I'm just a poor software man; hardware is my hobby, and
I lack formal background and "real" experience in it.
--
Henry Spencer @ U of Toronto Zoology
{allegra,ihnp4,linus,decvax}!utzoo!henrycrandell@ut-sally.UUCP (Jim Crandell) (07/19/84)
I've seen 2-inverter crystal oscillators in scads of commercial
products (usually with a 74LS04) and I've used them myself. They
ordinarily work fine, and if you can't make them start, then you've
probably selected the wrong resistor values, or else the crystal
has an effective series resistance which is too high. I don't know if
they're reliable over -55 - 125, but then 74 isn't spec'ed for that,
anyway. I've had excellent results with single-inverter CMOS
oscillators, too (CAVEAT -- different resonance mode!). A section
of a CD4049 works well, and another section will drive a standard
74-series input; 4069/74C04 won't!
--
Jim Crandell, C. S. Dept., The University of Texas at Austin
{ihnp4,seismo,ctvax}!ut-sally!crandelljones@fortune.UUCP (07/20/84)
#R:utzoo:-402300:fortune:28000045:000:1934
fortune!jones Jul 19 16:44:00 1984
About 5 years ago, when I was at Qantel Corp (before they became
MDS Qantel), we went through the engineering drill of deciding
whether to continue with gate/oscillator clock circuits. We
were pretty heavy into intelligent controllers and had Z80s
everywhere. We came to the conclusion that it was better to
hand off the problem to somebody else and buy hybrid oscillators.
One would think that a manufacturer with control of all the
variables would be able to solve all our problems. Well, they
did finally but in the early days we saw all the problems:
cracked substrates, poor solder seals, out of spec duty cycles,
temperature sensitivity, voltage sensitivity, and delayed
oscillation. Hybrid oscillators are great, but it pays to
test them.
As for gate oscillators, the following is from Jim Williams article,
_Basic Circuit-design Techniques Yield Stable Clock Oscillators_,
in the August 18, 1983 edition of EDN:
While gate oscillators are quite popular, they can
cause problems ranging from tempermental operation
to lack of oscillation. The gain elements are the
primary problem source-it's not possible to reliably
identify the analog characteristics of digital gates.
For example, there's no guarantee that gates from
various manufacturers will produce the same results
when plugged into the oscillator circuit. In other
cases, the circuit will work but the status of the
other gates within the package will affect its
performance. Finally, some circuits seem to favor
certain gate locations within the IC package.
Given these difficulties, gate oscillators are not
the best possible choice in a PRODUCTION DESIGN.
They do deserve mention, however, because they can
satisfy noncritical applications.
The emphasis is mine.
Dan Jones
UUCP: {ihnp4,ucbvax!amd,hpda,sri-unix,harpo}!fortune!jones
DDD: (415)595-8444 x 440
USPS: Fortune Systems Corp, 101 Twin Dolphin Drive, Redwood City, CA 94065david@bragvax.UUCP (David DiGiacomo) (07/20/84)
>>... the following is from Jim Williams article, >>_Basic Circuit-design Techniques Yield Stable Clock Oscillators_, >>in the August 18, 1983 edition of EDN: >> >> While gate oscillators are quite popular ... >> Wasn't this article referring to non-crystal oscillators (which I agree can be problematic) ? Also, someone mentioned using a 4049 as an oscillator; this is not a good idea because of the non-linearity of the part, and possible excessive power dissipation. The best CMOS parts for oscillators are CD4007s or CD4069U/74CU04s up to 1 MHz, 74HCU04s above that. The 74HCU04 makes a better oscillator than the 74LS04, but can dissipate more power (I think). I'm really amazed at the volume of traffic on this inherently boring subject.
jones@fortune.UUCP (07/20/84)
#R:utzoo:-402300:fortune:28000046:000:314
fortune!jones Jul 20 10:42:00 1984
Sorry, for clarity in my previous submission, you can replace each
reference to "oscillator" with "crystal-oscillator". Including
the quote.
Dan Jones
UUCP: {ihnp4,ucbvax!amd,hpda,sri-unix,harpo}!fortune!jones
DDD: (415)595-8444 x 440
USPS: Fortune Systems Corp, 101 Twin Dolphin Drive, Redwood City, CA 94065phil@amd.UUCP (Phil Ngai) (07/21/84)
I don't understand why people continue to design osc circuits
when the xtal clock osc you can buy are about the same price
as xtals, at least if you are buying a reasonable quantity.
--
Just another valley guy
Phil Ngai (408) 982-6554
UUCPnet: {ucbvax,decwrl,ihnp4,allegra,intelca}!amd!phil
ARPAnet: amd!phil@decwrl.ARPA