Analog use of CMOS logic chips

[Jim Thompson]

On Mon, 15 Nov 2004 09:49:47 -0800, Charles Edmondson
<edmondson@ieee.org> wrote:

Jim Thompson wrote:
On Sat, 13 Nov 2004 20:17:09 -0800, Charles Edmondson
edmondson@ieee.org> wrote:


Jim Thompson wrote:

[snip]

Did you ever know an applications engineer who actually knew shit ?
They're almost always hackers without any ability to think beyond the
typical.

...Jim Thompson

Hey! I resemble that remark...

--
Charlie


Charlie, You're software apps. BIG difference.

...Jim Thompson
Hi Jim,
Actually, I am an MS:EE in microwaves and optics too lazy to get out and
work for a living. It is so much easier teling everyone else how to get
the job done, and I don't like the hours if you go management!


--
Charlie

Yep, Never go into management...

I once had a manager that I despised so much that I walked into his
office and said (hoping to be fired), "You're the dumbest
piece-of-shit that ever walked the earth".

His response, "I'm sorry you feel that way".

I finally had to lay myself off to get away from such a turd.

...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona Voice480)460-2350 | |
| E-mail Address at Website Fax480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |

I love to cook with wine. Sometimes I even put it in the food.

 

[Joerg]

Hello Mr. Civility,

Thanks for sharing this part of CMOS logic history. It's kind of sad
that many of these devices never really made it. The CA3600 is nowhere
to be found, neither is its data sheet. Even my 1978 RCA "Linear
Integrated Circuits" book doesn't have it. When I look back, the only
instances of transistor arrays I saw being used by US or European
engineers were in high end test equipment. The Japanese still use a lot
in consumer equipment because it makes sense.

Interesting how the 4069 came in the footsteps of the 4049. However,
they 'wimpified' the low side device sizes so the 4069 cannot pull down
as hard as the 4049 can. I'd rather trade lack of symmetry for more
oomph. Along the lines of a bumper sticker I saw on an American muscle
car in Europe: "There is nothing that can replace lots of cubic inches.
Except more cubic inches."

Regards, Joerg

http://www.analogconsultants.com

 

[Jim Thompson]

On Mon, 15 Nov 2004 01:22:57 GMT, Joerg
<notthisjoergsch@removethispacbell.net> wrote:

Hi Jim,

Yes, that always makes me drool. But even a shared run costs as much as
a nicely equipped used car.


Doesn't cost me a dime ;-)



Can you sometimes hitch a ride on a prototype run from your designs?

That's called a "shuttle". A lot of my clients take their first ride
on a shuttle.

Totally different topic: Do you see a trend in chip design to farm some
of that work out to India, as it is seen with SW and digital ASIC
projects? IOW, if one of your grandkids would ask you whether it is a
safe career move to go into chip design what would you say?

Regards, Joerg

http://www.analogconsultants.com

Become a lawya ;-)

...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona Voice480)460-2350 | |
| E-mail Address at Website Fax480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |

I love to cook with wine. Sometimes I even put it in the food.

 

[Joerg]

Hi Active8,

An engineer who worked in HVAC control gear already felt that 'spirit'
more than ten years ago. He suggested to switch to an 8bit micro
controller because that would open up some more fancy applications.
Nope, was the answer. Not as long as the Asian 4bit uC costs a few
pennies less.


How many were purchased in a year?

A gazillion.

You need a selling point on those fancy apps and you want to perhaps
hold off on them until you need them to keep your share of the
market.

Today, for basically the same price as 10 yrs ago, you can get the
same big 'ol Fedder A/C. But now the thermostat control is an
embedded deal with a friggin' remote control. Refrigerators talk.

It was the thermostat he designed. Realtime clock, programmable, LCD,
the whole nine yards except for the remote. With a four-banger in there.

Regards, Joerg

http://www.analogconsultants.com

 

[Rich The Philosophizer]

On Tue, 16 Nov 2004 02:38:07 +0000, Joerg wrote:

Hi David,

What stopped people from lifting the cart across the loop. or tilting
it if only one wheel was active? Was signal necessary to keep the
wheel unlocked?



The way almost all of these systems work is that someone has to waive
the magic wand to unfreeze things. Even if someone were to lift the cart
across it would be pretty useless with one stuck wheel. Unless, of
course, the wheels were easily removable and that thief took two carts ;-)

I think it's not really to deter theft as much as the neighborhood

folks just borrowing them to cart the groceries home, and then "forgetting"
to take them back. In one sleepy little community I lived in, the grocery
store had a truck, and the guy would periodically go around town and pick
up all the carts and bring them back to the store.

Cheers!
Rich

 

[Rich Grise]

On Tue, 16 Nov 2004 01:43:51 -0800, Winfield Hill wrote:

Ken Smith wrote...

BTW: There really was a one bitter, the 14500.

I have a pdf files of Motorola's 6-page mc14500 datasheet, in case
anyone is interested.

This one-bit CMOS processor had 16 instructions, operated at 1MHz,
worked from 3 to 18V supplies, along with the rest of the mc14000
and cd4000 cmos families, and came in a nifty 16-pin DIP package.

I'd be happy to see it on, for example, a.b.s.e. I remember our
supervisor and one of the other techs noticing the product announcement:
"There's a lot you can do with one bit." "Yeah, you can, um ..."
[simultaneously]"complement it, ..." [or, you _could_ compliment it:
"Hey, nice lookin' bit ya got there!"<groan!>]

But I did see a suggested app with a 16X4 TTL stack, (7489?) where one
of the nops acted like a "return."

I enjoyed reading the data sheet because I was concurrently
reading about the AM2900 and learning about pipelining and state
machines and stuff. I even like the "LU" in the block diagram.

It would have been great for industrial control apps, like you'd
use an Allen-Bradley or Square-D boat anchor for in a real live
industrial control app, if the guys who designed with either
had both visited a universe where they have ever even heard of
the other. ;-)

I used hobbyist-grade parts once in an industrial setting, and
the whole little control lashup blew up almost _before_ I powered
it up, that's how hostile industrial control is. I wouldn't
expect to see a 4000 or 14000 series CMOS chip unless it was
protected inside military-grade armorplate. ;-)

Thanks!
Rich

 

[Ralph Barone]

In article <15pjw7in20btg.dlg@news.individual.net>,
Active8 <reply2group@ndbbm.net> wrote:

On Tue, 16 Nov 2004 02:30:24 +0000 (UTC), Ken Smith wrote:

In article <Ywbmd.44507$QJ3.37685@newssvr21.news.prodigy.com>,
Joerg <notthisjoergsch@removethispacbell.net> wrote:
[...]
It was the thermostat he designed. Realtime clock, programmable, LCD,
the whole nine yards except for the remote. With a four-banger in there.

Four bits luxury! We had to do our designs with only one bit and were
happy to have that.

BTW: There really was a one bitter, the 14500.

I wonder if that's the one that was used in that old Circuit Cellar
supercomputer article - the Mandelbrot Engine.

The company I work for was one of the first to even make a multiprocessor
system. The system had 3 4040s in it. They were over clocked so they
really zipped along at about 2.2MHz.

If'n that was the one that was also in BYTE magazine, I think he used
something from the 8051 family.

 

[Steve Kavanagh]

Thanks for the suggestions (Joerg, too).

bigcat@meeow.co.uk (N. Thornton) wrote in message news:<a7076635.0411151843.3fa6d3a7@posting.google.com>...

I'm still not figuring out where the problem was, or why you wanted
low z input on the audio.

The mixer was a passive switched-FET type with 50 ohms output Z. The
following audio amp was initially an inverter biased for linear
operation. An inverter operating in linear mode has truly awful
audio-frequency noise figure at 50 ohms source impedance. I don't
know what the number is but it is certainly huge.

Gates can be parallelled for MC speaker drive,

I used six inverters in parallel and a transformer to drive low-Z
headphones.

(3) a voltage regulator ?

Which bits are you running off the Vreg? The key q is what i, so the
less running off the reg the better, and what v?

Voltage needed is 5 V more or less. The only thing that NEEDS to run
from the regulator is the 7 MHz oscillator and a few buffers/inverters
- about 5 mA using a 74HC04, perhaps a bit less using an 'HCU04. In
this transceiver, most stages are regulated as I had plenty of
regulator capacity and needed to drop the 7.5 V supply voltage
somehow.

I don't plan to post the schematics for this transceiver anywhere (too
much trouble for what it's worth). Perhaps, if I get to the 100
percent digital CMOS target with good performance, that one will get
published somewhere. Some similar circuits using digital CMOS chips
survived into the follow-on, more optimized design which is described
at

http://www.arsqrp.com/ars/pages/back_issues/2002_text/0602_text/ss_special.html

For those who are interested, here is the active device lineup for the
almost-all CMOS original:

Local oscillator/buffers: 74HC04 (yeah, I know it should be an 'HCU04
but I didn't have one at that moment!)
Mixer: 74HC4066 (perhaps not a 100% digital chip but in 74HC series,
anyway)
Audio preamp: MMBT3904 (CB)
Audio amp/active filters: 4069
Audio "power" amp: 4049
Control circuit: 74HC02 (sorry, I used a digital chip for a digital
purpose!)
Regulator: LM2931C
Sidetone oscillator: 4001
RF power amp: 74HC240 (150 mW - I've seen them used up to a half watt
output)

There are also a few diodes. I'm sure with a little more work
conducting semiconductor junctions could be dispensed with entirely,
except maybe one zener !

Steve

 

[Kevin Aylward]

Jim Thompson wrote:

On Tue, 16 Nov 2004 05:15:16 -0500, Active8 <reply2group@ndbbm.net
wrote:

On Mon, 15 Nov 2004 17:48:14 -0700, Jim Thompson wrote:

[snip]
I started, just out of MIT, for $6760 per year... in 1962. By 1970
I was making my age in $K. By 1980 I was making 3x my age, $K.
After that I finally started getting paid what I'm worth ;-)


If you had to start over today, do you think you could do it again
based on trends you've seen over the years in the market/economic
climate? IOW, are we FUBAR?

These are not good times for EEs.

Indeed. I wish I had become a lawyer.

Kevin Aylward
salesEXTRACT@anasoft.co.uk
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.

 

[Rich Grise]

On Tue, 16 Nov 2004 08:04:29 -0700, Jim Thompson wrote:

On Tue, 16 Nov 2004 05:15:16 -0500, Active8 <reply2group@ndbbm.net
wrote:

On Mon, 15 Nov 2004 17:48:14 -0700, Jim Thompson wrote:

[snip]
I started, just out of MIT, for $6760 per year... in 1962. By 1970 I
was making my age in $K. By 1980 I was making 3x my age, $K. After
that I finally started getting paid what I'm worth ;-)


If you had to start over today, do you think you could do it again
based on trends you've seen over the years in the market/economic
climate? IOW, are we FUBAR?

These are not good times for EEs.

If what they report on TeeVee has any credibility at all, it would
seem that times are just now getting ripe for new grads - all of
us^H^Hyou old baby boomer farts are retiring. ;-)

Thanks!
Rich

 

[Rich The Philosophizer]

On Tue, 16 Nov 2004 15:36:11 +0000, Kevin Aylward wrote:

Jim Thompson wrote:
On Tue, 16 Nov 2004 05:15:16 -0500, Active8 <reply2group@ndbbm.net
wrote:

On Mon, 15 Nov 2004 17:48:14 -0700, Jim Thompson wrote:

[snip]
I started, just out of MIT, for $6760 per year... in 1962. By 1970
I was making my age in $K. By 1980 I was making 3x my age, $K.
After that I finally started getting paid what I'm worth ;-)


If you had to start over today, do you think you could do it again
based on trends you've seen over the years in the market/economic
climate? IOW, are we FUBAR?

These are not good times for EEs.


Indeed. I wish I had become a lawyer.

You could always play one on TV. ;-)

<aside re TV> I saw an edjamacayshunal show about stars and telly-o-scopes
and flying saucers and stuff that was hosted by Patric Stewart. I was
so terribly terribly disappointed when he didn't introduce himself thus:
"Hi. I'm not a starship captain, but I play one on TV..." <rimshot>
</aside>

;^j
Rich

 

[Rich Grise]

On Tue, 16 Nov 2004 15:42:40 -0500, Active8 wrote:

On Tue, 16 Nov 2004 14:45:35 GMT, Ralph Barone wrote:

In article <15pjw7in20btg.dlg@news.individual.net>,
Active8 <reply2group@ndbbm.net> wrote:

On Tue, 16 Nov 2004 02:30:24 +0000 (UTC), Ken Smith wrote:

In article <Ywbmd.44507$QJ3.37685@newssvr21.news.prodigy.com>,
Joerg <notthisjoergsch@removethispacbell.net> wrote:
[...]
It was the thermostat he designed. Realtime clock, programmable, LCD,
the whole nine yards except for the remote. With a four-banger in there.

Four bits luxury! We had to do our designs with only one bit and were
happy to have that.

BTW: There really was a one bitter, the 14500.

I wonder if that's the one that was used in that old Circuit Cellar
supercomputer article - the Mandelbrot Engine.

The company I work for was one of the first to even make a multiprocessor
system. The system had 3 4040s in it. They were over clocked so they
really zipped along at about 2.2MHz.

If'n that was the one that was also in BYTE magazine, I think he used
something from the 8051 family.

I think it *was* BYTE, now that you mention it. Bunch of cards with
multiple 1 bitters. The whole shebang was controlled by an AT PC.
What were they, 8085/6?

Um, the only 1-bit processor there ever was, AFAIK, was the MC14500.
I can't imagine using them for anything approaching arithmetic, unless
you're building a visual barrel shifter trainer sequencer or something.

There was something called "the connection machine" that used something
about the same heft as a Z-80, but something like 256 of them, and it
was astronomical(in cost). 8085/6 isn't usually referred to like that -
there's the 8080/8085 and the 8086/8088. I wrote some stuff for an MP-M/16
system with an 8085 and 8088, but it wasn't a real multiprocessor - the
8088 would launch 8085 programs and idle until it was done, a lot like a
coprocessor.

In any case, the 8085 or 8088 would be a logical choice for this,
other than a large array of dumb multiplier chips, with, as I said,
a uP for sequencing.

There was a chip out once, the 8X300, that was designed and touted
for nothing but speed - it had that non-Von Neumann two-bus style,
with separate data and instruction buses. That was supposed to be
ultra-fast, and I think it was the first chip I ever saw that they
recommended liquid cooling.

Cheers!
Rich

 

[Allan Herriman]

On Wed, 17 Nov 2004 08:38:12 +0000, Guy Macon
<http://www.guymacon.com> wrote:

Product developer wrote:

Since the carrier is electro-magnetic it cannot be shielded either.
One of our demos at trade shows was wrapping wheels in aluminum foil
and then dropping them into aquariums and being able to lock and
unlock them.

I remember back when I supervised the first installations back in '97
that when trucks arrived with an inventory of wheels you could hear
them all locking up inside the truck as it traversed the loop entering
the driveway.

What did you use for a power supply inside the wheel?

from an earlier post:

The all in wheel electronics and locking mechanism had to last four
years or 15,000 lock / unlock cycles on one Lithium 3.6 volt photo
battery.

Regards,
Allan

 

[Joerg]

Hello Steve,

Since the 5V doesn't need to be too accurate you could use a CMOS device
as a regulator if your 7.5V is predictable enough.



Unfortunately, this doesn't seem practical for two reasons:
(1) the battery voltage drifts down significantly over time (due to
the significant discharge from the high current in the RF PA)
(2) the battery voltage in key-down transmit is much lower than in
receive.
The LO needs much better regulation for frequency stability.

Ok, here is another suggestion: Why not use a CMOS logic based switcher?
I have designed a few for clients and it made them really happy for two
reasons. First, none of the PWM chips could beat the low cost. Not even
close. Then, there are no issues regarding parts availability at all
times. Other than that these are rock stable and run at efficiencies
comparable to the better PWM chips. Except for quiescent current where
the CMOS logic solution was far ahead of the game, quiescent was close
to zilch without any 'power-down' pins or stuff like that.

Usually you need one external transistor and one diode here but I bet
you could figure a way to use something like the CD4007 in a synchronous
design.

Regards, Joerg

http://www.analogconsultants.com

 

[Rich Grise]

On Wed, 17 Nov 2004 10:55:38 -0800, Joel Kolstad wrote:

Since the carrier is electro-magnetic it cannot be shielded either.

You mean that since the 'carrier' is at a low enough frequency it can't be
shielded with reasonably thin shields?

I mean, aluminum works just fine for shielding even at 60 Hz -- you just
need many inches of it!

I suppose that a better attack would be to try to overload the front end of
your detector, eh?

Or, save up all the money you'd have spent on schemes and designs
and circuits and stuff, and just buy a cart! ;-)

Cheers!
Rich