why analog electronics

Don't forget Max Plank. I think he nailed the bottom turtle.

There is such a thing as "mixed signal" design...

Also, if the amplifier has a DSP in it, that kind of screws things up.

But, why make a silly binary distinction when a continuous division is possible? ;-)

Tim

Grin, hadn't heard that. My ~10 yro copy of Eagle (4.15) is still running fine (on Win 7), I don't ask it to do much. I've got other software, that now is only kept alive on a virtual XP machine. (I'm kinda a ludite with software though.)

Oh and as I've read, storage is best done digital, whether DNA or bits. Analog makes the measurement, digital records it and passes it on.

George H.

Spoilsport.

So the whole analog vs. digital argument is really analog? ;-)

Yeah ! Like that !

Absolutely ! fascinating stuff !

e = h f

Yes. that sentence says it all ! I like it !

So, add in non-linear circuits too...

Kind of in-between. Log amps and stuff.

Of course, nothing is perfectly nonlinear.

boB

Not that Eagle has ever "worked" all that well... :^)

Had an idea for a semi-analog storage (e.g. sample-and-hold), namely one without drift. Though it would be slow. Suppose you use back-to-back F-to-V and V-to-F converters, and lock it with a PLL. Interrupt the loop with a regular S&H when you want to update the value.

I suddenly forget why that would work... you'd have to add another constraint that the frequency is divided evenly, or a ratio is conserved, or something. Otherwise, drift is drift. And yes, once you're locked to a rational number, you're as good as digital, it's just arising from a circuit different from the canonical ADC-latch-DAC.

Tim

I'll cosign. I know there have been other articles on the theme of "there will always be analog" most of which make good points. I couldn't point you to any links, though.

I do think that in general the most common circuits now are ADC -> digital -> DAC, and that the dividing line between the analog and digital keeps getting pushed closer and closer to the edges of the board. But I also firmly believe that there will _always_ be applications for which a pure, or mostly, analog solution is the sensible way to go.

(I'm currently immersed in Shannon's 1948 tutorial paper where he spells out his channel capacity theorem. He mentions PCM modulation, AKA digital replacement of an analog signal, as an example of sending redundant information on a signal. Doing so comes complete with the thresholding effects you see in any other forward error correction scheme, where things are Really Good up to some threshold, and then things get Really Bad.)

I remember that from working on bluetooth nearly 20 years ago, when measuring sensitivity there was a razor edge between between zero bit errors loss and 100% packet loss

I can buy a 32 Gbyte USB memory stick for $9. That's roughly 2e-12 dollars per transistor. Or I can buy a fast opamp for about the same price, with maybe 100 transistors inside, around 10 cents per transistor. The memory stick comes with free shipping, and the opamp doesn't.

I'm designing two boxes just now that are both 100% analog. Not a flipflop or gate in sight. That's refreshing. We have much higher markups on all-analog stuff than on mixed-signal products.

That is seriously good stuff, and pricey. Now that I make my own abbey ale it's not so pricey but even there a small 12oz bottle easily comes to $1. Not $5 like in the store though.

It's more like flatulence delivering stink. A neuron "fires" when the permeability properties of the cell change to allow sodium to diffuse through the cell membrane.

Thank you,

Roland Moreno, the inventor of the smart card gave an interview to Electronics Times and came out with this pithy observation: "Digital is easy. Analog.... that's professional."

Cool. I'll include that in my essay.

It's true, though. 'Digital engineers' that want their signals to switch faster or propagate faster are going to run into the unavoidable dominance of analogue effects pretty quickly. They'll very soon learn which is boss when that happens.

Let me know how easy "digital" is after you've worked on a modern microprocessor design.

Lots of digital design, these days, is typing VHDL or Verilog for a compiler to do mysterious things inside an FPGA. We generally have only a fuzzy idea of what actually happens inside that FPGA, and rarely control any of the details of logic implementation or place+route.

A few of us design logic on PCBs where the speed of light matters.

"Typing VHDL" is rather a large oversimplificaiton of the job. You don't know what a LUT is?

The speed of light matters on some chip desings, too.