Your favorite 10 analog IC's

You do get some charge injection, but it cancels reasonably well if the two inputs are near the same voltage, and if you have a well-defined quiescent state, it can be taken out afterwards. I used it in a project about 10 years ago to build a 5-dimensional user interface, where by moving your head you could look around things on the screen, zoom by leaning forward and back, that sort of thing. I had an MRI tomograph of somebody's head, and you could navigate anywhere, at any angle, just by leaning slightly. Keyboard and mouse control let you freeze it when you got the view you wanted, so you didn't have to sit still all the time. It was headed towards being a real product before my management changed and the new guy axed it.

It used three small solar cells behind a shadow mask cut in the lid of a plastic box, with an array of IR LEDs to illuminate your forehead as you sat at the monitor. Sums and differences of the signals from the photodiodes gave X, Y, and Z to pretty good accuracy, for very very cheap. The LEDs were chopped at 100 kHz by a 555 driving a small MOSFET, and each of the three (cascoded) TIAs drove a one-channel lock-in amplifier made out of a 1496, by putting a 10nF capacitor between the collectors at the output. The bad news about a 1496 as a lockin is that its outputs are differential, and some distance from ground.

The 4052 had a 1000-pF capacitor across the two poles. One pair of the throws was connected between ground and the buffer amp input, and the other three pairs went to the 1496 outputs. The same 8051-family MCU controlled the 4052, did the digitizing, and sent the data out an RS232 port.

Cute gizmo, and really really cheap.

Cheers,

Phil Hobbs

They are not. Current accuracy (matching) is much poorer, and they are not Wilson mirrors (see AoE page 89) like the TI parts, which means they suffer dramatically from Early-effect errors.

I liked the TL014, which with its 4x current step-up was more practical for use in battery-operated circuits. I still have a few in my inventory, but am "saving" them...

Though I've heard he always said/says that it should be the "Jones/Gilbert" Cell multiplier.

And with a quick look you can find the same disclaimer at:

"... low-noise commutating mixer. This is a very important mode of use, and the form has become known as the "Gilbert Mixer," in both its BJT and more recent MOS embodiments. It should be noted that, while processing these early multiplier patents, this more basic form of the core was found embedded in a patent concerning a synchronous demodulator by H.E. Jones, issued in 1963. In spite of strenuous efforts to correct this incorrect assignment at conferences, in Short Courses, and at informal gatherings, the name persists. "

Robert

Never misses an chance for an insult, does he? He must troll the Usenet looking for opportunities to offend; truly a thankless job.

John

The CoolRunner product was actually developed by Philips. Xilinx bought it away from them and promptly killed off the 5V version of the product line.

I used the 5V version in quite a few places. It could drive the gate of a small power MOSFET directly, intefaced nicely to other 5V logic, had fairly good speed and a nice set of development tools from Philips. It also had "industry standard" pinout (same as Altera) so you were not locked into their product if you had the current.

If you wanted to be a bit Mickey Mouse about things, the device could be made to work as the inverter in a crystal oscillator.

I used the PZ5032 when it was Philips. It is in a PLCC package and is programmed in a standard programmer.

Today, Lattice has the lowest current CPLDs of any size. If there developnment tools didn't suck so bad, I may have considered them. My current project uses a Cypress part.

I expect Xilinx to cancel the entire CoolRunner product line at some point.

Is it possible that every electronic device containing more than 3 chips contains something you have designed originally (and others copied / improved over time)?

TL07x NE5532 NJM4560 NJM4565 NJM4580 LM339/393 LM317 uA78xx uA79xx TL431

All the above have been around for ages. But then I'm not doing anything esoteric that needs those 'bleeding edge' parts.

Graham

Jim,

It's Barrie's name at the bottom of that link, referencing the words above. Or so it seems to me. Did you even look at the link?

Robert

You've led a sheltered life, then. There are not many *production* sockets for SMT parts (with the exception of those for PLCC packages), but plenty of test/burn-in/programming sockets. I don't think there is an IC package made that does not have a mating socket of some kind.

Here's one for a part I'm working with right now-- a TQFP-100 SMT part with 0.5mm lead pitch and a 14mm square body:

Best regards, Spehro Pefhany

I thought the same thing (not Moto), and have done reasonably well (whew!). I don't collect residuals either. ;-)

And I've mostly just heard it referred to as a "Gilbert Cell", no matter what it's intended use was. Perhaps that's what Barrie was trying to correct.

You would have heard more references to it than I.

Robert

Hello Tim,

Ouch. Now I am going to get even more goose pimples when I see "mature". I don't know why the 3046 went but maybe many young engineers just aren't familiar enough with discrete designs. Or maybe it is because you can buy a CD4007UB for less money and they came out in TSSOP (now that was really good news). Who knows.

Regards, Joerg

Hello Bill

The 4046 is ok and the price is certainly right. But at the end of the day I have more often designed one out than in. Linearity was the issue most of the time. You can straighten that out nicely with external references and stuff but then the price and size advantages kind of fizzle away. Might as well build a PLL from scratch, at least for some higher frequeny applications.

But you are right. It is a pretty nifty chip.

Regards, Joerg

"Ken Smith" a écrit dans le message de news:cv089d$u9f$ snipped-for-privacy@blue.rahul.net...

I don't remember the figures for the Lattice parts, but have a look at Atmel ATF15xxASL parts. These are still 5V and pretty low power.

"Analog

of

the guy

I tried to use your embodiments when they first became available, and found them too restricted to be useable. The Analog Devices embodiments, when they finally hit the market, were useful in my kind of work. In this case I couldn't give a rats arse about whether you cited his famous paper or not - the problem was the number of extra components required in my particular circuits.

------- Bill Sloman, Nijmegen

Forget the tomograph, that baby screams "Game Controller!" If you build it people will buy it, and they'll pay a lot for it.

Yes the Atmels are "pretty low power" when compared to the Altera parts but the last time I looked at them, the current was higher than the CoolRunner which is higher than the Lattice.

I've used the Atmel micros but never actually designed in the CPLDs.

Atmel has a tool that crosses the part from the Altera version automatically. This is kind of a good idea but if I'm using a free tool from Altera, it will be an Altera part I use.

OB analog: None of these are intended by their maker as analog chips but you can use them to create analog waveforms by PWMing. 5 resistors and a 5 bit Johnson counter will get you pretty close to a sine wave.

In article , Joerg wrote: [.. 4046 ..]

I'd like it a lot better if the HC versions didn't tend to have that sudden hook on the end of the curve. The CD4046 just isn't fast enough on

I vaguely (25-30 years ago) remember using a TI 74LS624 VCO, but substituted discrete current mirrors for the R's to ground, and got quite good linearity.

...Jim Thompson