[Kostya (OP)]

So, there's my first post and i've decided to connect it with question which many people can tell that it's just one of thous stupid questions. But indeed many of you, who are interesting in electronic, make a choice between analog or digital electronic. So there is my questions what do u prefer deal with - analog or digital electronic and why?

As for me, my blood is almost analog,almost - because sometimes life turn me in to a digital world

[CliffordK]

I suppose it is a complex question, but really depends on the application.
Computers are primarily digital, and contain analog to digital converters.
Audio obviously is analog and is imported into the computer with some kind of analog to digital conversion.  Personally I can't tell the difference between CD/MP3 based music and LP (old fashioned record) based music.

I'm not a big TV connoisseur, but the new flat screen digital TV's sure seem to be better than the old analog CRT TVs with lots of SNOW  [ xx(]

Now there are some things that analog are "better" with.  At least in a sense.
I dislike my color laser as I believe it is somewhat like old fashioned French Impressionism, with distinct dots for colors.  I believe that the ink jet printers are somewhat better at mixing colors (more analog-like), and thus give a much better color image quality.  Obviously the computerized images are digital though, but with a reasonable amount of color depth, one can get a good color image with even somewhat grainy resolution.

[Kostya (OP)]

One of my university's professor tell me that future is for analog electronic, especially in medicine. Digital devices have, as you know, two statements - 0 or 1. And for example if u try to create a model of neuron, signals of with is analog, u will find a big trouble with it.

As for me analog electronic is more flexible. If u design a digital stuff u almost know that it will be working or not working. But if you design an analog, all your calculations in one second can go to hell, coz u might not include some physical facts, that must be known by you.

[RD]

... Digital devices have, as you know, two statements - 0 or 1. And for example if u try to create a model of neuron, signals of with is analog, u will find a big trouble with it.

Even analog electronic signals are quantized, e.g. ... http://en.wikipedia.org/wiki/Shot_noise#In_electronic_devices
i.e. they can be lumpy like digital. 

[CliffordK]

One of my university's professor tell me that future is for analog electronic, especially in medicine. Digital devices have, as you know, two statements - 0 or 1. And for example if u try to create a model of neuron, signals of with is analog, u will find a big trouble with it.

I think that is an oversimplification. 
While a bit has two states 0/1, you can assign an arbitrary number of bits to anything.  For colors, 16 bits gives you 65 thousand potential colors.  32 bits gives you 4 billion potential colors.

As far as the human brain, there are about 100 billion distinct neurons.  Each has a single axon, and several dendrites, that will make a distinct number of synapses.  One can look at the threshold as a sum of the discrete activations at the dendrites, and an action potential more or less reduces to a 0/1 binary occurrence.  At a smaller level, you have individual neurotransmitters that are activating specific receptor sites.

I don't see why you would say that it is inherently analog, although an analog system might add a certain amount of randomness that would otherwise require additional specific calculations in a digital system.

Of course, there are the issues of simulating an inherently parallel system on a serial system.

[graham.d]

For the forseeable future is the growth is definitely in the digital arena for almost all electronics. This may seem a bold statement, but as someone who has been in the business of semiconductors for about 40 years, and who's main expertise is (at least, was) in analogue design, I have some direct knowledge of how things have developed. It is not to say that analogue circuits will not have their own growth - the outside world is analogue after all - but that more and more signal processing (and much more sophisticated and very accurate signal processing) is now carried out by digital circuitry. As device sizes have shrunk down in size over the years, more and more circuitry can be put on a chip. A lot of analogue circuitry cannot be shrunk down to the fine geometries for a whole host of reasons which I won't go into now, but digital gates can be. Many very complex analogue processing comprises something like a few amplifiers, an Analogue to Digital Converter (ADC), a Digital Signal Processor (DSP) and a Digital to Analogue Converter (DAC). Todays DSPs are very powerful and fast and there is often not even any possibility to use analogue processing and achieve the required specification. In addition, digital circuitry can be programmable and/or flexibly changed to fit requirements or to fix bugs. Digital circuits are also robust and not subject to variations with temperature or the manufacturing process such that the resulting product's quality may vary (provided they are designed correctly) and are much easier to test.

However, if you are looking at a career in electronics (which is maybe the angle that Kostya's professor is coming from) the I would advise learning analogue design techniques. It is very varied and complex and will always be in demand. The DSPs have to be programmed too and they need a lot of understanding of the mathematics behind the processes involved which are usually trying to emulate some analogue functions. Much of the challenges of digital design today is about managing complexity - and this isn't easy but requires a different sort of skill. The basic design is similar to programming (in Verilog or some HDL variant) and then using VERY expensive and complicated tools to verify and build the design (and verify again). Many aspects are automated but the complexity means that they don't always get everthing 100% right, especially with very fine geometries. It is challenging and the art is in mastering the tools; usually people specialise in some parts of the toolset so that nowadays "backend" design is carried out by a different set of engineers.

Then, of course, there is process technology, which requires a whole different set of skills (mostly not considered as "electronics" though).

[Kostya (OP)]

After this post I'm not sure that something need to be added. Yep, want we it or not, we are happened in a digital world so we must take his rules. But recently I've been read Texas Instruments book about operational amplifiers, and thous brainiacs tell that 25% of all electronic components, produced of TI,  is analog. At least we must somehow connected to real world, by the way, not natural, fictitious world that we create to have deal with things that we can understand.

But sometime people find an interesting symbiosis between two statements digital and unpredictable analogue world : newbielink:http://en.wikipedia.org/wiki/Fuzzy_logic [nonactive]

And noise in analog electronics doesn't have any with digital. It's inherently nature.          

[MikeS]

Seems to me, on a human scale, it's essentially an analogue world.   Analogue is better for some applications, digital for others.

[Geezer]

They are really only different methods of producing particular results, and they are both based on the same, and frequently interchangeable, technology.

I'll probably get clobbered for making this statement, but it all boils down to how you represent and operate on "stuff". Whether it is represented in analog or digital form, sooner or later, whatever it was is quantized. In other words, it either crossed some threshold, or it didn't.

In that respect, ultimately, everything is "digital". It may not be a convenient binary digital representation, but ultimately a decision was made at some threshold somewhere. Today's digital systems impose strict limitations on thresholds that are typically binary, but that does not automatically exclude all kinds of other representations.

As Graham pointed out, the main reasons for the current distinctions between "analog" and "digital" are economic. I don't believe there is a distinction that is based on a fundamental scientific principle.

[graham.d]

Back in the 1960s virtually all electronic components were analogue. When DTL then TTL logic gates became available there was a big move to digital; especially in the numbers of components but this is because there was so little functionality in each chip (just a few logic gates). As you say, now the majority of components from (say) TI are digital - though they will often have some analogue on them and most analog components will have some digital on them.

How you judge any comparisons between "analogue" and "digital" depends on the metrics used to decide on their relative importance. Total market value, Numbers of components sold, Total chip area, Effort to design etc. etc.? It is not a matter of it being a competition - both are essential and required and both have their own difficulties and challenges. Personally, I enjoy analogue and RF design as this was my background, and I have not the time or inclination to try to get involved with the complex software tools needed to design big digital chips. I also have probably lost the ability to absorb such vast amounts of data too with each software package having a 1000+ page instruction manual :-) The software is far from being bug free too as the push for advancements outweighs the need for thoroughness in development; and the tools are very sophisticated.

However, the boundaries between analog and digital design have been pushed back over the years so that digital processing has absorbed more and more of the analog functions. And this is still continuing because the advantages are huge. The vast majority in the functionality increase the world has seen in everything electronic has resulted from the use of digital processing, and this trend is continuing. The electronics industry has seen a massive growth as a result and a corresponding growth in automation of the design process.

Of course analogue design is also a vital part and will also continue to grow, though it is often limited by the inability to scale designs to very fine geometries. Semiconductor processes are widely produced with options to allow higher voltage and larger geometry structures to accommodate some analogue functions that cannot be made to scale down. Pressures to reduce costs (and chip size) demand this. The only alternative is to try to eliminate the analogue part and to absorb the (say) few hundred large transistors and resistors into a more reliable, easier to test and better yielding digital circuit where even a few tens of thousands of transistors would still be smaller and a cheaper solution. And, what is more, next time you need to shrink down in geometry the digital circuit will scale down with little effort.

Nonetheless, I would advise anyone starting out in the industry to get a good understanding of all aspects of analogue design and, more fundamentally, the basics of electronics and the physics of semiconductors. These skills will always be in demand. The pressures in the digital design arena are to de-skill the detailed design process and move it to a higher level of system design. This is a challenging area in itself, but is more akin to programming and systems analysis. It depends on where your interests lie.

[Geezer]

Nonetheless, I would advise anyone starting out in the industry to get a good understanding of all aspects of analogue design and, more fundamentally, the basics of electronics and the physics of semiconductors. These skills will always be in demand. The pressures in the digital design arena are to de-skill the detailed design process and move it to a higher level of system design. This is a challenging area in itself, but is more akin to programming and systems analysis. It depends on where your interests lie.

I believe Graham is correct. There is so much pressure to crank out devices that are faster, and more extensively integrated, that it's quite easy for engineers to become highly specialized in the use of some particular tool while becoming more disconnected from the underlying technology.

 

[Kostya (OP)]

You have so broad knowledges in modern electronic that my mind don't give me placidity.
So I wonder what do you do?

As for me, I'm an electrician student and after a half year I'll get a bachelor degree and soon must to decide my next steps.

[syhprum]

Given a good strong signal to give an adequate s/n ratio and a freedom from ghosts an anologue TV picture will always be superior to the compressed digital one but of course these demanding condition are not always met especialy as people are always keen to spend more on receivers than antennas.
where the signal is weak and ghosty a digital picture may well be prefered.

[graham.d]

Hmm, a lot of caveats needed for that to be true, Syphrum. Of course if you capture analog data perfectly, transmit it without loss, and then reproduce it perfectly, what you say is true (it has to be). However, the practical limitations of bandwidth and SNR mean that this is far from being the case. One advantage of digital transmission is the available Forward Error Correction which allows for a considerable degree of loss without any deteriation in the recovered data.

In any case, if you throw enough digital bits and bandwidth at the problem you can also achieve as near to perfection as you like.

[syhprum]

May I quote a simple case, in the preamble to the FOX Simpsons episodes Marge is is shown checking out the baby in a store when a price appears on the till and the baby ends up in a paper bag.
If I record an anologue broadcst of this on my VHS recorder I Can play it back one frame at a time and see how much the store values the baby at but if I record a digital broadcast and record it on a HDD as is the modern fashion it is quite impossible to do this.
I also record APT satellite pictures from polar orbiting satellites that uses the newspaper photofax system deveoped in the twenties that will often produce usable pictures as the satellite passes behind buildings at low altitudes where a digital system would give up.
As Geezer would say "Vive la résistance!"



 

[graham.d]

You must be a member of the Ohm Guard :-)

[techmind]

To paraphrase what I've written elsewhere...
From the earliest days of gramophone record, telephones, and radio, engineers worked to replicate the form of audio sound pressure waves in some other medium. In a gramophone record, the depth of the groove is varied in exact sympathy with the sound; in telephones an electrical voltage (or current) is generated in proportion to the sound pressure; in amplitude-modulated (AM) radio transmissions the strength of the radio wave is varied in direct accordance with the sound. All these systems are classed as 'analogue' (or American, 'analog'), because the stored or transmitted signal bears a direct analogy to the original information.
Despite their very attractive simplicity, the most obvious problem with analogue systems is that the reproduction is similarly an analogue (like-copy) of the stored or transmitted signal ...and so any slight distortion or interference of the stored or transmitted signal will cause likewise distortions and interference in the recovered sound. For example, dust on a gramophone record results in pops and crackles; on an AM radio receiver, interference (from sources such as strip-lights or electric motors) often causes buzzes and whistles in the received sound.

The word 'digital' is a derivation of 'digit' and refers to systems which process numbers. In the case of digital recording or transmission a numerical description of the original signal is made, and it is this description, rather than a direct analogue, which is stored. In practice, this numerical description is created by sampling (measuring) the original signal to a finite resolution, at a series of discrete time intervals. The result can be thought of as a log-book of numerical measurements. As an example, in the case of compact-disc audio, that log-book has 44100 entries for every second of the recording, and each entry represents the sound pressure to 5 decimal digits (in fact, a resolution of 1 part in 65536), for the left- and right-hand stereo channels.
These digital signals (think: numerical description) are inherently more resilient to corruption during storage or transmission than their analogue counterparts. Visualise a laboratory notebook from school; even if the page (the recording) becomes blotched and stained, for as long as your numbers remain legible, the data you took remain exactly the same, and can be re-written without loss or damage of the information contained.

Furthermore, 'checksums' (a sophisticated sort of built-in tally) can be appended to digital signals so that even if some of the digits do get irrecoverably corrupted or confused (up to some finite limit of course) the receiver or playback device can still deduce the original signal with no impairment at all.

Digital is more complicated in concept, but provides what I call 'assured quality' (the 'quality' is defined and fixed at the recording/encoding stage, and you're pretty-much guaranteed to get that back with any playback or receiving device). Resilient to distortion and degradation, perfect (loss-less) copies can be made. Processing (such as studio mixing or effects) can be performed numerically with negligible loss of fidelity. Possible to increase quality for relatively marginal increases in cost. Almost any sort of specialised analogue processing can be done at least as well in the digital domain, and in fact the scope for digital processing is almost unbounded.

Huge advantage once your data is digital is that all sorts of data can be treated the same, you no-longer need a cassette for audio, VHS for video, prints for photographs, disks for computer-files... you can store the whole lot on a generic hard disk or CD/DVD.

Of course digital, because it's so flexible, allows you to apply "bit rate reduction" (data-compression) technologies, which reduce the storage or transmission requirements... but with this comes a loss in fidelity (which can be negligible if done sensitively) or ghastly if dictated top-down by accountants... (hence the loss of clear still-frames in video sequences). On the other hand, all the loss can be 'managed' by the broadcaster, then every customer gets the same - unlike analog where the privilaged few with line-of-site to the transmistter get a good picture and everyone else gets progressively more snowy/noisey/ghosty excuse for a picture depending on where they live and the state of their aerial.

[techmind]

Hmmm... editing doesn't seem to be working - edits don't stick. Mods - can you help?
I corrected my typos above twice, and added another paragraph, now lost :-(

I said essentially that the losses due to "bit rate reduction" are not inherent to digital (don't blame 'digital tech') but are a conscious choice by the broadcaster or person encoding the file to a lossy format. We could have superb TV pictures, if they chose to only broadcast 1/3rd as many stations...

[CliffordK]

Hmmm... editing doesn't seem to be working - edits don't stick. Mods - can you help?
I corrected my typos above twice, and added another paragraph, now lost :-(

I think I noticed this earlier. 
http://www.thenakedscientists.com/forum/index.php?topic=42525.msg375814#msg375814

There are two modify buttons.  The one at the bottom of the post does not work.  The one at the top works fine.
Perhaps it doesn't hurt to remind those working on the system maintenance that it still needs some improvement.

I said essentially that the losses due to "bit rate reduction" are not inherent to digital (don't blame 'digital tech') but are a conscious choice by the broadcaster or person encoding the file to a lossy format. We could have superb TV pictures, if they chose to only broadcast 1/3rd as many stations...

A friend of mine was trying out Netflix yesterday.
It was fine when we ran it through a little fuzzy 14" CRT TV.

When we connected it to a digital TV, what became obvious was that during slow scenes, the image quality was excellent.  However, during intense action sequences, the lossy format caused unacceptable pixelation.

Of course, it could have also been a hardware issue, so we'll do more testing shortly.

[Geezer]

A friend of mine was trying out Netflix yesterday.
It was fine when we ran it through a little fuzzy 14" CRT TV.

When we connected it to a digital TV, what became obvious was that during slow scenes, the image quality was excellent.  However, during intense action sequences, the lossy format caused unacceptable pixelation.

Of course, it could have also been a hardware issue, so we'll do more testing shortly.

Might be the link that's a bit congested. We run Netflix in HD on Rokus, and it's usually very good, although you can sometimes see it working hard. Mind you, that might also be the TV.