Hi

The 74HC series is an “old and slow” CMOS family. In some cases people use
their low speed to do filtering (they won’t follow a fast glitch …). Other than that
sort of thing, the AC and LVC should be fine replacements for them. The HC
might pull a little less power with nothing going on. If you are running on a coin
cell it’s worth checking.

The 74AC series was essentially the first group of fast CMOS parts to hit the market.
They were indeed fast. They also had some issues with lead frame induced glitches
(power and ground pin locations …). You see a lot of odd things done to try to take
care of this.

Starting about 20 years ago, people began to bring out “improved” versions of the AC
parts. Processes had gotten faster and people had learned some things. Within 10
years the number of different families became almost un-countable. I think these
guys spent a lot of time coming up with weird reasons to add a new line of logic
to their portfolio.

The good news is that faster is better in silicon CMOS (except in weird cases like above).
About all you need to check is:

1. Is it CMOS or TTL levels? (AC vs ACT , HC vs HCT, etc) get the right one.
2. Is it as fast as the stuff I’m replacing ?(check toggle rates or delay) simple answer
   is almost always yes if HC and AC are the “comparison standard”.
3. Will it handle my supply? Some of the newer stuff is 3.3V only.

There are some odd cases like buffer gates. If you are crossing one of those, just be sure
you cross it to a buffer and not to a simple inverter. Drive levels can matter.

====

Simple answer - sure they are better / lower noise floor / better ADEV / faster / cheaper
/ easier to layout than the old stuff. They probably are lower power as well.

Bob

On Sun, 31 May 2015 14:06:26 -0400
Dan Watson watsondaniel3@gmail.com wrote:

Yes, quite a few of those. After CPLDs and FPGAs replaced all of the
more complex 74xxx's, people realized that most projects do not need
4 NAND gates at one spot, but rather single ones here and there
(a schmitt-trigger for signal conditioning, an AND gate to couple two
enable lines,...).

I have never used any of the LVC in a timing circuit, but i would
guess they are not worse than the AC. Also they have the advantage of
having single gates per package, which helps minimizing cross coupling
between different signal paths.

BTW: [1] may contain some interesting data for you. Especially as it
compares different manufacturers too.

Looking at [2], the ALVC family would probably be also worth a look.

			Attila Kinali

[1] "Low Voltage Logic Designers Guide", Ti, 1996
http://www.ti.com/lit/ml/scba010/scba010.pdf

[2] "Logic Guide", Ti, 2014
http://www.ti.com/lit/sg/sdyu001aa/sdyu001aa.pdf

--
< av500> phd is easy
< av500> getting dsl is hard

Thanks for the replies! Very informative.

Dan

Sent from my iPhone

I am curious, how the integrated gates could be compared for those
created on discrete elements ? Let say simple gates like this:

http://hyperphysics.phy-astr.gsu.edu/hbase/electronic/trangate.html

Regards,
Vlad

On 2015-05-31 18:23, Attila Kinali wrote:

--
WBW,

V.P.

Moin,

On Mon, 01 Jun 2015 11:00:31 -0400
Vlad time@patoka.org wrote:

For the same implementation: Worse. There is much less control
over the exact building of discrete devices than integrated ones.
There hundreds of tiny things that are harder to control:

• You have wires going over the PCB. These have a long length, thus
  a high inductivity, and high capacity.
• They are on a substrate that is known to be quite hydrophile and
  thus has changing electrical parameters (dielectric constant,
  permeability, resistivity,...)
• Everything is slower, due to higher capacitance/inductance/resistance
  of the wires inbetween.
• Slower translates into less defined edges -> more jitter.
• Higher capacitance translates into more current flowing at the same
  speed, which again translates into more shot-noise (though less
  therma noise)
• Longer wires are more susceptible to EMI, or rather pick-up of tiny
  interferences from other circuits or even other parts of the same
  circuit.
• Longer wires are also more prone to emit EMI.

... just to name a few. This list can be continued for quite some time.

On the other hand: A well designed discrete circuit can beat a general
purpose integrated circuit in almost all performance measures.

These are resistor-transistor-logic (RTL) elements. Those even more
noisy due to their simple build up.

		Attila Kinali

--
It is upon moral qualities that a society is ultimately founded. All
the prosperity and technological sophistication in the world is of no
use without that foundation.
-- Miss Matheson, The Diamond Age, Neil Stephenson

The example circuits given on that webpage will be inferior in every way to
RTL and DTL logic from the 1960's, which itself is inferior to late
60's/early 70's TTL.

A good reference on how to do discrete transistor logic design from the
1960's, is TI's book "Transistor Circuit Design". I just poked around
bitsavers hoping it might be there, but didn't see it. Paper copies abound
in used bookstores.
http://www.amazon.com/Transistor-Circuit-Design-Instruments-Incorporated/dp/0070637377

There are a handful of level-shifting logic-type circuits for oddball
voltages that might still be done best using discrete transistors
especially if you don't need say the full quad of a MC1488 or MC1489.

Tim N3QE

On Mon, Jun 1, 2015 at 11:00 AM, Vlad time@patoka.org wrote:

I have something of a follow up question. How good is the isolation inside
these devices (74LVC, SOT-23 package) between gates?

Let's say I have a 20MHz TCXO. I want to square up the output signal and
divide by two. Easy, just a buffer or inverter and a flip flop. But looking
at the pinout of the 74LVC1G175 (D flip flop) it doesn't have a Q not
output. So now I need a second inverter to make it toggle. The 74LVC2G14
includes two schmitt inverters in the package, but will isolation inside
the device be good enough to use it for two separate functions at 20 and 10
MHz?

Just from a layout perspective using three devices instead of two would be
easier. However the thing will be battery powered, so I'd like to save the
power if possible.

Thanks

Dan

On Mon, Jun 1, 2015 at 6:13 PM, Andy AI.egrps+tn@gmail.com wrote:

One thing that is hidden in AC and later CMOS is very tightly controlled
edge-rate to combat ground bounce.  The original AC components were so
fast, the ground bounce could be measured in volts and they had to be
quickly redesigned.

For the D-FF function, you might consider using one section of the dual
74LVC74.  With the inputs of the unused section connected to ground or
Vdd, it will draw no power.

David N1HAC

On 6/8/15 8:30 PM, Dan Watson wrote:

Maybe it's just me ... but I wouldn't trust the isolation to be good.
Apart from on-die coupling, they share the same power/ground leads, however
short they are.  So just getting power and ground from the board onto the
die, they would be corrupted by what happens in the other gate and its load
(L*di/dt).

Andy

Hi

The isolation in the package is likely better than the (practical) layout you will
do to mate up with them. In fact, the single gate stuff probably does a better job
of isolation than the multi gate stuff, simply because you can spread it out on
the board.

In the case of dividing by two, there are single gate flip-flops that are Q bar output
rather than Q. That eliminates the second single gate package in this design. Yes, there
are far to many different numbering systems. Finding this or that can be a massive
pain.

======

If power is an issue, the real trick is to find a family that is happy running on a low(er)
supply voltage. Some of this stuff will toggle at 20 MHz with very low supply. Often
the inputs are “high voltage tolerant” even with those low supplies.

Bob

Hi

The question is always “good isolation compared to what?”.

If you are expecting 180 db of isolation on a SOT-23 package at 10’s of MHz, it’s
not going to happen. It’s also not going to happen with a practical pc board layout
even without the SOT-23 involved.

If something around 120 db is “good isolation”, then yes you can do that with a
pair of the SC-70 parts and a good layout. In this case the test was at 10 MHz.

Bob