HP Forums
Can analog make a comeback? - Printable Version

+- HP Forums (https://www.hpmuseum.org/forum)
+-- Forum: Not HP Calculators (/forum-7.html)
+--- Forum: Not remotely HP Calculators (/forum-9.html)
+--- Thread: Can analog make a comeback? (/thread-10287.html)

Can analog make a comeback? - mfleming - 03-07-2018 01:08 AM

I ran across this interesting article recently from IEEE Spectrum. It discusses some interesting research in the area of analog computing and hybrid analog/digital computing. The author talks about IC's his research group has fabricated to put analog computing to work.


The article was interesting because analog computing can be much faster and consume much less power for certain applications. Picking the application area is crucial because of the limited accuracy of analog computing. Despite the fascination with ever more digits of accuracy (34 digits!), it's not always necessary for real world problems. After all, we went from Kitty Hawk to the Moon with slide rules that give only two or three places of accuracy.

Analog circuit elements, such as summation, multiplication and integration, were arranged in a switching fabric somewhat like cells in an FPGA. Interconnections between elements could be established before computation begins. One interesting circuit element for simulating arbitrary functions was an A/D converter driving the address lines of a RAM whose outputs would then drive a D/A converter.

Any thoughts on where a dead-and-buried technology like analog computing might be applied today?


RE: Can analog make a comeback? - Garth Wilson - 03-07-2018 02:21 AM

From the title, I thought it was going to be about analog recording. Cassette sales are actually increasing again today, and National Audio Company just set up to begin producing cassette tapes again, the first produced in many years, with a new formulation that is supposed to be much superior to all previous type-1 formulations. (Types II and IV manufacture pollutes too much to be allowed today, at least in the US.)

This is interesting, as a friend just sent me something about analog computers recently. Op amps have improved dramatically in speed, accuracy, sensitivity, noise, etc. since the heyday of analog computers, so it stands to reason that an electronic analog computer today could do far better than they could back then.

I bookmarked this interesting video about mechanical analog shipboard fire-control computers, apparently from WWII: https://www.youtube.com/watch?v=s1i-dnAH9Y4

(03-07-2018 01:08 AM)mfleming Wrote:  After all, we went from Kitty Hawk to the Moon with slide rules that give only two or three places of accuracy.

Uh... make that usually three to four, as long as you have good vision and know how to veneer-adjust by rolling your fingers (yes, including on the slide, which many people in online video slide-rule demonstrations don't know how to do). I can quite often get four even with my pocket slide rule. Just for the sport of it, and because of the doubters, I sometimes check against a calculator, and the fourth digit may be off by 1 or 2, definitely closer than not trying for a 4th. (Yes, I do sometimes still use a slide rule, to sharpen my mind.)

RE: Can analog make a comeback? - Maximilian Hohmann - 03-13-2022 12:31 PM


I am resurrecting this old thread because I just saw Derek Muller's latest Video on YouTube on exactly this topic ("Veritasium" channel: https://www.youtube.com/watch?v=GVsUOuSjvcg): Modern analog and hybrid analog/digital computers. I found it very interesting and begin to see real applications in the future.

The "toy" analog computer that is featured in the video is this one here: "The Analog Thing" (https://the-analog-thing.org/ )

I am very much tempted to buy one of these, I already have an oscilloscope so it would be a matter of unpack-and-play. Does by any chance anyone already have one of these and can say something about it?


RE: Can analog make a comeback? - mfleming - 03-13-2022 02:24 PM

Fascinating video, and a really great YouTube channel. Neural nets are an obvious application for analog circuitry. One thing I remember from a course in IC design is that passive component values can't be manufactured very precisely, but component ratios can. Hence with proper design large scale integration can overcome the inaccuracies of an individual component assembly. If I had a standalone oscilloscope I'd also be tempted to get a THAT. Post your experience if you do so.

Somewhat off topic this reminds me of an "analog computer" I bought as a kid from a mail order supplier, Edmund Scientific. It consisted of three logarithmic potentiometers wired to a battery and an analog meter. Each pot had a pointer to a circular log scale beneath. You would input two numbers on two pots then adjust the third pot to null the meter to zero. The product of the two inputs could be read from the third pot position. An analog slide rule, if you will.