Author Topic: Analog circuits are fast: But how fast?  (Read 1619 times)

0 Members and 1 Guest are viewing this topic.

Offline Beamin

  • Super Contributor
  • ***
  • Posts: 1011
  • Country: us
  • Beamin=She not He
Analog circuits are fast: But how fast?
« on: September 17, 2018, 12:07:01 am »
In a lot of posts and videos on the internet people always say "remember in analog circuits things work fast" Referring to how signals move around the circuit. But how fast? Or are they saying relative to clock speed in a micro the analog circuits work instantly (speed of light?)


What can slow an analog circuit down? Putting in caps that would have to charge up each time something happens but somehow not block signal?
 

Offline MK14

  • Super Contributor
  • ***
  • Posts: 1955
  • Country: gb
Re: Analog circuits are fast: But how fast?
« Reply #1 on: September 17, 2018, 12:12:41 am »
Not as fast as Beamin can create new threads.  ;D ;D

Digital circuits are usually considered to be way faster.
But analogue circuits, can do much more work, while it is settling on a new voltage, and anyway, you can have super fast analogue circuits (especially if you are allowed to include going to microwave frequencies, and even higher, probably), as well.

But there are all sorts of drawbacks to analogue circuits, which digital circuits tend not to suffer from (drift, tolerance, susceptibility to interference, cost, needing trimmers, etc etc). They can be very tricky to design, and there are not that many people, who are really good at very fast analogue design.
It can get very expensive as well.

In many things, high speed digital systems, is the solution and winner.
« Last Edit: September 17, 2018, 12:16:09 am by MK14 »
 

Offline Brumby

  • Supporter
  • ****
  • Posts: 7452
  • Country: au
Re: Analog circuits are fast: But how fast?
« Reply #2 on: September 17, 2018, 12:15:48 am »
If you want to get down to brass tacks, even digital circuitry operates in an analogue world.  Slew rates are one parameter that reflects this.
 
The following users thanked this post: NiHaoMike, joeqsmith, MK14, petert

Offline MK14

  • Super Contributor
  • ***
  • Posts: 1955
  • Country: gb
Re: Analog circuits are fast: But how fast?
« Reply #3 on: September 17, 2018, 12:20:00 am »
If you want to get down to brass tacks, even digital circuitry operates in an analogue world.  Slew rates are one parameter that reflects this.

But if you want a real challenge. Design this forum, to work, entirely made up of analogue stuff, such as analogue computers and things.

Good Luck!

Don't cheat and say that digital gates are really 100% analogue circuitry. I know what you mean, but that is NOT what I meant, here.
 

Offline Beamin

  • Super Contributor
  • ***
  • Posts: 1011
  • Country: us
  • Beamin=She not He
Re: Analog circuits are fast: But how fast?
« Reply #4 on: September 17, 2018, 12:24:10 am »
If you want to get down to brass tacks, even digital circuitry operates in an analogue world.  Slew rates are one parameter that reflects this.

But if you want a real challenge. Design this forum, to work, entirely made up of analogue stuff, such as analogue computers and things.

Good Luck!

Don't cheat and say that digital gates are really 100% analogue circuitry. I know what you mean, but that is NOT what I meant, here.


With through hole transistors that would probably be 100's of square feet. An iPhone out of vacuum tubes would be bigger then LA and consume more power too.
 
The following users thanked this post: MK14

Offline MK14

  • Super Contributor
  • ***
  • Posts: 1955
  • Country: gb
Re: Analog circuits are fast: But how fast?
« Reply #5 on: September 17, 2018, 12:25:45 am »
In a lot of posts and videos on the internet people always say "remember in analog circuits things work fast" Referring to how signals move around the circuit. But how fast? Or are they saying relative to clock speed in a micro the analog circuits work instantly (speed of light?)


What can slow an analog circuit down? Putting in caps that would have to charge up each time something happens but somehow not block signal?

They are slowed down by things like slew rate (mentioned above) and other things, such as capacitors (and stray capacitance), taking time to charge up and settle down to the final values.

That is why (there is more to it though), op-amps run at megahertz (but you can get some very fast ones), but logic gates can run at gigahertz and faster.
 
The following users thanked this post: petert

Offline Brumby

  • Supporter
  • ****
  • Posts: 7452
  • Country: au
Re: Analog circuits are fast: But how fast?
« Reply #6 on: September 17, 2018, 12:26:10 am »
If you want to get down to brass tacks, even digital circuitry operates in an analogue world.  Slew rates are one parameter that reflects this.

But if you want a real challenge. Design this forum, to work, entirely made up of analogue stuff, such as analogue computers and things.

Good Luck!

Don't cheat and say that digital gates are really 100% analogue circuitry. I know what you mean, but that is NOT what I meant, here.

Yeah... I know.  But somebody had to go there - and I was feeling a little impish tonight.   :D
 
The following users thanked this post: MK14

Online coppercone2

  • Frequent Contributor
  • **
  • Posts: 571
  • Country: zw
Re: Analog circuits are fast: But how fast?
« Reply #7 on: September 17, 2018, 12:26:37 am »
I think for an analog device thats not switching a logic level an important parameter you are interested in is capacitance. The lead in to the device combined with the capacitance to ground forms a low pass filter which impedes operation.

I mean this from a datasheet level view point.

For instance if a device  has 10pF capacitance to ground, you can calculate the equivalent resistance to see what kind of drive impedance you would need for that circuit, the drive impedance becomes absurd for higher frequencies, if your impedance is like 0.5 because of a capacitor shunting it to ground, your not going to get much energy into the device in the first place

from there on the specifics probably require you to pick up a transistor physics text book
« Last Edit: September 17, 2018, 12:33:15 am by coppercone2 »
 
The following users thanked this post: MK14

Offline MK14

  • Super Contributor
  • ***
  • Posts: 1955
  • Country: gb
Re: Analog circuits are fast: But how fast?
« Reply #8 on: September 17, 2018, 12:30:49 am »
With through hole transistors that would probably be 100's of square feet. An iPhone out of vacuum tubes would be bigger then LA and consume more power too.

Giant analogue circuits, don't really scale up too well. Because the drift and other effects, could mean that it never works, as engineers are forever re-tuning it and fixing it. (Especially old analogue designs, modern stuff can be pretty decent, and solves many of those problems).
Even old digital (valve/tube) based computers were like that as well.

I think I read that the really old valve/tube, massive computers, could only run for an hour or so (maybe a few hours or so), before another vacuum tube would burn out and need replacing.
I think they spent half their time, being repaired and the other half doing stuff (very approximately).

N.B. I'm talking about digital computers from the 1940s and 1950s.
Later ones became more and more reliable.
« Last Edit: September 17, 2018, 12:32:35 am by MK14 »
 

Offline rstofer

  • Super Contributor
  • ***
  • Posts: 4700
  • Country: us
Re: Analog circuits are fast: But how fast?
« Reply #9 on: September 17, 2018, 12:58:12 am »
If you want to get down to brass tacks, even digital circuitry operates in an analogue world.  Slew rates are one parameter that reflects this.

But if you want a real challenge. Design this forum, to work, entirely made up of analogue stuff, such as analogue computers and things.

Good Luck!

Don't cheat and say that digital gates are really 100% analogue circuitry. I know what you mean, but that is NOT what I meant, here.

You're on!

I want to display damped harmonic motion like the mass-spring-damper problem (sort of like the suspension on a car).  You start with your digital approach (first principles, not MATLAB) and I'll start patching my analog computer (actually, it's already patched).  Don't forget to allow for parameter changes on the damping coefficient and the spring rate.  It's fair to assume gravity is constant.

Or, change it up a bit.  You start with 7400 logic and I'll start with 741 op amps, we'll just breadboard the solution.  I need two op amps as integrators, one op amp as a summer and another op amp as an inverter.  I also need 3 potentiometers and two timing capacitors for the integrators.  Plus a power supply and a hand full of resistors.

How's that adder coming along?  Actually, it would be fair to use a uC and just write the code.  I'll still be up and running first.

In my old age, I am finding a lot of fun in playing with the differential equations I struggled with in college.

I didn't like all the math required for analog circuits back in school.  We had sliderules, the HP35 hadn't been invented and it seemed to me that analog design would go the way of the dinosaur as soon as we had a way to digitize the signals.  Not a bad prediction for '72 or so.  Even the uC was a pipe dream for a couple of years.

I'm not sure younger people realize just how much things have changed over the last 50 years.


« Last Edit: September 17, 2018, 01:00:12 am by rstofer »
 
The following users thanked this post: drussell, MK14, petert

Offline MK14

  • Super Contributor
  • ***
  • Posts: 1955
  • Country: gb
Re: Analog circuits are fast: But how fast?
« Reply #10 on: September 17, 2018, 01:20:59 am »
Sadly, I suspect that these (analogue) design skills are being lost. So if we suddenly need them again (and to an extent we do, as they are still needed), it is not so good to lose them.

My understanding, is that in the 1960s, there were (relatively) a huge number of Analogue Transistor design engineers, around the world.
Who could be charged (excuse the triple pun, as in electrical charge, £/$ charge and charge as in to do things), with making a brand new 20 watt, audio amplifier, completely out of transistors (no ICs or op-amps to be seen/used).

These days, people who can design such things, exclusively out of transistors (20 watt amplifier), are getting rarer and rarer.
 

Offline rstofer

  • Super Contributor
  • ***
  • Posts: 4700
  • Country: us
Re: Analog circuits are fast: But how fast?
« Reply #11 on: September 17, 2018, 01:22:21 am »
I was just thinking about the difference between an analog and digital approach to controlling the rocket nozzles on the Atlas missile.  During launch, the nozzles are used to keep the bottom of the missile under the top.  It's kind of important.

Which would be faster more responsive?  An all analog system where rotation is measured in real time, amplified, integrated, differentiated and/or multiplied and sent to the servos or a digital system where sensors are measured, results computed and finally converted back to voltage and then sent to the servos.  Analog integration is a natural result of analog circuits.  Differentiation introduces noise and is not often used.

That's not to say there aren't applications where digital is more appropriate.  What is interesting is the degree to which celestial navigation was used for course corrections.  Just a high dollar sextant and a bit of math (probably pre-computed tables).  Kind of the ultimate in analog.

Once you get outside the uC chip itself, the world is an analog place.  Digital doesn't keep that bridge up, analog does.  Speed down the highway is really analog.  It may show up as a digital indication (+-1 MPH) but the idea that distance traveled is the integral of velocity is actually analog.
 
The following users thanked this post: MK14

Offline rstofer

  • Super Contributor
  • ***
  • Posts: 4700
  • Country: us
Re: Analog circuits are fast: But how fast?
« Reply #12 on: September 17, 2018, 01:30:28 am »
Sadly, I suspect that these (analogue) design skills are being lost. So if we suddenly need them again (and to an extent we do, as they are still needed), it is not so good to lose them.

My understanding, is that in the 1960s, there were (relatively) a huge number of Analogue Transistor design engineers, around the world.
Who could be charged (excuse the triple pun, as in electrical charge, £/$ charge and charge as in to do things), with making a brand new 20 watt, audio amplifier, completely out of transistors (no ICs or op-amps to be seen/used).

These days, people who can design such things, exclusively out of transistors (20 watt amplifier), are getting rarer and rarer.

And more expensive, I suspect.  I don't know whether companies like Analog Devices have a recruiting problem or not.  To some extent, the math skills and basic circuitry are taught in college.  Everything else is just on-the-job training.  You come in as an enthusiastic new-grad and work with the more experienced engineers to learn the profession.  One day you are the experienced engineer.

I don't hear names with the recognition of Bob Widlar, Bob Dobkin or Bob Pease (that's a lot of Bobs).  There are a number of these guys who really invented the fundamental building blocks for analog design.  Now it is simply a matter of step-wise refinement as process capabilities improve.

FWIW, here is a brief discussion of the Big Three.  Hint, they all worked together at one time or another:

https://en.wikipedia.org/wiki/Bob_Dobkin
 
The following users thanked this post: MK14

Offline MK14

  • Super Contributor
  • ***
  • Posts: 1955
  • Country: gb
Re: Analog circuits are fast: But how fast?
« Reply #13 on: September 17, 2018, 01:33:30 am »
Which would be faster more responsive?  An all analog system where rotation is measured in real time, amplified, integrated, differentiated and/or multiplied and sent to the servos or a digital system where sensors are measured, results computed and finally converted back to voltage and then sent to the servos.

From a control point of view, that can become a very tricky question. Because the digital computing solution, could attempt to predict (without being 100% accurate), what the control variables are going to be (and need to be), a bit (or longer) in the future.

Taking this one step further. Let's say a major control servo, as part of the rockets control system, takes exactly 10 millseconds to respond to changes in requested position.
The analogue solution, will tend to respond to changes in the sensors and hence change its output. So the rocket gets its changes 10ms later (because of the servos mechanical delay).

But the digital solution, because of its predictions on the near future, can send the new position request to the servo, approx 10ms, before it needs to move, hence removing or reducing the mechanical delay of the servo.

But it is possible that a clever/innovative analogue computer solution, could do something similar. With feed-forward (possibly wrong technical name here, what I mean is you calculate it a bit in the future, like a d term in a PID controller, but more sophisticated than that) terms and stuff.
« Last Edit: September 17, 2018, 01:36:01 am by MK14 »
 
The following users thanked this post: analogo, petert

Offline MK14

  • Super Contributor
  • ***
  • Posts: 1955
  • Country: gb
Re: Analog circuits are fast: But how fast?
« Reply #14 on: September 17, 2018, 01:42:42 am »
Bob Widlar

I'm a big fan of his!
I would have loved to have met him. It is a pity he died so young (relatively speaking).
He was one of the main developers, of the early IC op-amps.
He came up with so much brilliant stuff!

If I remember correctly, he had a right hand man, who was hugely helpful for him (possibly a brilliant IC process engineer, which is extremely important). Whose name I have forgotten, and who is often left out of the history.

Similar claims are made for the invention of the transistor. Where the leader (William Shockley), took much of the credit. Despite the other two team members, having important roles as well.
« Last Edit: September 17, 2018, 01:44:13 am by MK14 »
 

Offline MK14

  • Super Contributor
  • ***
  • Posts: 1955
  • Country: gb
Re: Analog circuits are fast: But how fast?
« Reply #15 on: September 17, 2018, 01:53:59 am »
And more expensive, I suspect.  I don't know whether companies like Analog Devices have a recruiting problem or not.  To some extent, the math skills and basic circuitry are taught in college.  Everything else is just on-the-job training.  You come in as an enthusiastic new-grad and work with the more experienced engineers to learn the profession.  One day you are the experienced engineer.

I don't hear names with the recognition of Bob Widlar, Bob Dobkin or Bob Pease (that's a lot of Bobs).  There are a number of these guys who really invented the fundamental building blocks for analog design.  Now it is simply a matter of step-wise refinement as process capabilities improve.

FWIW, here is a brief discussion of the Big Three.  Hint, they all worked together at one time or another:

https://en.wikipedia.org/wiki/Bob_Dobkin

My understand, is that because in the 1960s, they were doing a huge number of (usually) analogue designs, using discrete transistors. There were lots of analogue design engineers, who day in, day out.
Were living and breathing in, analogue designs, and were amongst other brilliant analogue engineers.
It was also a time, when doing things like that was very exciting, and people were prowd of such work.

Hence the conditions to produce many quality analogue engineers occurred, and it did.

But these days, there is considerably less stuff like that done, these days. If you wanted an IC free twenty transistor, complicated analogue system designed from scratch. It would be very difficult to find (especially from the younger generation), people to do it.

My understanding is, that it is the design of complicated ICs (other areas as well), which still potentially needs a large quantity of quality analogue design engineers. So a shortage of them, could adversely affect the IC semiconductor industries.
« Last Edit: September 17, 2018, 01:59:58 am by MK14 »
 

Offline rstofer

  • Super Contributor
  • ***
  • Posts: 4700
  • Country: us
Re: Analog circuits are fast: But how fast?
« Reply #16 on: September 17, 2018, 01:58:19 am »
Bob Widlar

I'm a big fan of his!
I would have loved to have met him. It is a pity he died so young (relatively speaking).
He was one of the main developers, of the early IC op-amps.
He came up with so much brilliant stuff!

If I remember correctly, he had a right hand man, who was hugely helpful for him (possibly a brilliant IC process engineer, which is extremely important). Whose name I have forgotten, and who is often left out of the history.

Similar claims are made for the invention of the transistor. Where the leader (William Shockley), took much of the credit. Despite the other two team members, having important roles as well.

I had lunch with Bob and a bunch of his cohorts one time when I worked for National.  He spent a lot of time in Puerto Vallarta kicking back and thinking about new stuff.  And then there is the story about the lawn out front of National's first building.  They didn't have money to pay a gardener so Bob bought a sheep (not the rumored goat).  Problem solved!

Bob Widlar was a unique individual.  I didn't meet the other Bobs but I did meet Charlie Sporck one time.

https://en.wikipedia.org/wiki/Bob_Widlar

 
The following users thanked this post: MK14

Offline rstofer

  • Super Contributor
  • ***
  • Posts: 4700
  • Country: us
Re: Analog circuits are fast: But how fast?
« Reply #17 on: September 17, 2018, 02:04:37 am »

My understand, is that because in the 1960s, they were doing a huge number of (usually) analogue designs, using discrete transistors. There were lots of analogue design engineers, who day in, day out.
Were living and breathing in, analogue designs, and were amongst other brilliant analogue engineers.
It was also a time, when doing things like that was very exciting, and people were prowd of such work.

Hence the conditions to produce many quality analogue engineers occurred, and it did.

But these days, there is considerably less stuff like that done, these days. If you wanted an IC free twenty transistor, complicated analogue system designed from scratch. It would be very difficult to find (especially from the younger generation), people to do it.

My understanding is, that it is the design of complicated ICs, which still potentially needs a large quantity of quality analogue design engineers. So a shortage of them, could adversely affect the IC semiconductor industries.

But why would you use an all-transistor design these days?  We don't still use the CK-722 Germanium transistor and not many new designs will use a 741 op amp.  Things change...

The analog companies are all still in business making new devices all the time.  The engineers wouldn't design a circuit using all transistors because there is no motivation for doing so.  But I don't doubt for a minute that they could do it.

 

Offline Ducttape

  • Regular Contributor
  • *
  • Posts: 50
  • Country: us
Re: Analog circuits are fast: But how fast?
« Reply #18 on: September 17, 2018, 02:31:14 am »
When I was a kid you could spin the channel knob on the TV as fast as you wanted and still see, ever so briefly, the image/video of each channel for that fraction of a second it was selected. I miss that.  :'(
« Last Edit: September 17, 2018, 02:34:47 am by Ducttape »
 
The following users thanked this post: MK14, BrianHG, james_s

Offline joeqsmith

  • Super Contributor
  • ***
  • Posts: 4736
  • Country: us
Re: Analog circuits are fast: But how fast?
« Reply #19 on: September 17, 2018, 02:33:39 am »
I imagine you will still find analog control systems in use today.   As mentioned, digital is still analog.  Speedwise, seems like the same problems will haunt both.  :-// 

A few of us were playing around with some simple oscillators some time ago. 
https://www.eevblog.com/forum/projects/el-cheapo-oscillator/

I was playing around with some old 7400 logic making ring oscillators.  How times have changed.  :-DD
https://www.eevblog.com/forum/projects/waveforms-in-a-74ls04-ring-oscillator/
How electrically robust is your meter?? https://www.youtube.com/channel/UCsK99WXk9VhcghnAauTBsbg
 
The following users thanked this post: MK14

Offline MK14

  • Super Contributor
  • ***
  • Posts: 1955
  • Country: gb
Re: Analog circuits are fast: But how fast?
« Reply #20 on: September 17, 2018, 02:36:59 am »
But why would you use an all-transistor design these days?  We don't still use the CK-722 Germanium transistor and not many new designs will use a 741 op amp.  Things change...

The analog companies are all still in business making new devices all the time.  The engineers wouldn't design a circuit using all transistors because there is no motivation for doing so.  But I don't doubt for a minute that they could do it.

Actually, the answer to that question, is quite simple.
I don't know the latest world wide figures, as to how many of the following designs, still need to be created.

But there is a need to design new ICs (ASIC's, full custom ICs and similar such things), for all sorts of application areas.

E.g. A new sensing technology is invented, and it is expected to sell and be produced in the tens of millions each year.
Perhaps a new acceleration/motion detector, for the latest games console controller or virtual reality or car sensors etc etc.
That may need one or more, brand new designed, fully custom (or similar) all analogue ICs to be designed.
No op-amps or digital gates on it at all. Just 20 (or 5 or 50,000), cleverly placed analogue transistor designed, circuits.

So in order to design this new, all analogue (transistor) IC. It needs, clever, experienced analogue design engineers to create it.

Even if it goes to an all analogue design, but uses op-amps, via an IC ASIC design. It still needs analogue designers to create it.

Also, there are plenty of mixed, analogue and digital, custom ICs. Sometimes including MCUs.
The analogue bits, still need analogue designers, even if the onboard MCUs needs other types of engineers (software etc).
« Last Edit: September 17, 2018, 02:40:06 am by MK14 »
 

Offline rstofer

  • Super Contributor
  • ***
  • Posts: 4700
  • Country: us
Re: Analog circuits are fast: But how fast?
« Reply #21 on: September 17, 2018, 02:53:33 am »
This kind of design is being done all over the world!  There is no shortage (that I know of) of people capable of designing the chips.  It's a learnable skill and the design rules are pretty well documented.  Particularly in the case of ASICs where the final design is just an assembly of standard cells.

OTOH, I couldn't imagine making a 16 bit full adder from transistors.  And why would I?  I can just plunk down a pre-built IC and call it good.  I probably wouldn't even do that!  I would just use a CPLD or FPGA and write a line of code to instantiate the adder.
 
The following users thanked this post: MK14

Online coppercone2

  • Frequent Contributor
  • **
  • Posts: 571
  • Country: zw
Re: Analog circuits are fast: But how fast?
« Reply #22 on: September 17, 2018, 02:55:24 am »
the difference between digital and analog is the initial comparator, then its just an analog signal but in the frequency/timing domain only. You still get noise but in the time axis and of course duty cycle.

You also get things like noise margins that are different in different parts of the circuit and stuff like that, voltage levels will be different in different parts, but so long it works no one really cares, but you can design the shit out of it just like anything else... even the IC has some kind of specification throughout its signal routing.. its not really magically stable or something, and this all comes down to analog design choices that are made, the digital most people are used to is just idiotproofed very well so it seems invariant compared to digital design. Most people won't care past a specification like fanout.

You can persue specs like noise related effects (stability, margins,thresholds), propagation delay (stability), jitter stability, power consumption,

in anything engineering the only reason you are happy with off the shelf parts is that they meet your expectations. They can always be built differently to conform to a different standard or design choice you choose for whatever reason, its just hardly economical to pursue the benefits of things like custom digital memory, but it irritates me none the less.
« Last Edit: September 17, 2018, 03:01:34 am by coppercone2 »
 
The following users thanked this post: MK14, petert

Online coppercone2

  • Frequent Contributor
  • **
  • Posts: 571
  • Country: zw
Re: Analog circuits are fast: But how fast?
« Reply #23 on: September 17, 2018, 03:08:55 am »


OTOH, I couldn't imagine making a 16 bit full adder from transistors.  And why would I?  I can just plunk down a pre-built IC and call it good.  I probably wouldn't even do that!  I would just use a CPLD or FPGA and write a line of code to instantiate the adder.

because you don't have a specification that requires it yet
 

Offline james_s

  • Super Contributor
  • ***
  • Posts: 6079
  • Country: us
Re: Analog circuits are fast: But how fast?
« Reply #24 on: September 17, 2018, 04:11:41 am »
Sadly, I suspect that these (analogue) design skills are being lost. So if we suddenly need them again (and to an extent we do, as they are still needed), it is not so good to lose them.

I think they are. A while back when I found myself looking for work and toying with going more toward hardware rather than the software work I'd been doing I interviewed at a place that makes automated testing systems. They said they normally hire fresh EE grads but had been disappointed in the lack of analog skills in people coming out of the universities. As a college dropout almost entirely self taught hobbyist engineer I wasn't expecting much and I actually preemptively turned down the position because the commute would have absolutely killed me and I knew going in that it would have meant a pay cut but they told me later I was easily the most qualified person they talked to.

Ultimately I decided to stay in software which tends to pay a lot better, and keep EE as a hobby. It's something I enjoy doing enough that I don't want to turn it into work. Anyway the point of all this is that if my analog abilities which while passable are not expert level are that much better than recent EE grads then analog is in trouble.
 
The following users thanked this post: MK14, petert


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf