Isolated variable resistor function?

What is wrong with the I2C guys? You still have to use two wires for commands but now you can set to a specific value. Or, is there some other reason for using the up/down protocol?

Charlie

My experience with I2C is quite mixed. Sometimes there were undocumented hang-up modes. I'd rather use SPI, it is more robust.

If you're sneaky, you can push SPI through two optos, or even one.

The dredges (second distill) are where the flavor is... I also drink Jack Daniels ;-)

But you have to be a Scot to cope. One night, a friend and I designed a whole chip overnight in his hotel room, lubricated by two fifths of Drambuie... and it came out of fab working perfectly :-) ...Jim Thompson

Two is easy but most of the time when I proposed a bi- or tri-level scheme my clients said it's too much of a white-knuckle thing. So I often used carriers instead.

I was thinking of pulse widths, not levels. I came up with a scheme that use an HC123 and an RC, or something like that. I could try to reproduce it if anybody was interested.

with

you

can

thod

for

me

ted

Isn't that Manchester 2? The telecom companies went in for more elaborate schemes with even lower low frequency content - nowadays they seem to be back to telephone modem style phase and amplitude modulation, but there was a time when ADSL used fairly simple encoding.

-- Bill Sloman, Nijmegen

My idea used long and short pulses for 1s and 0s. The pulse was both data and clock. A simple RC on the receive end separated them back out. One one-shot made CS* too. Something like that.

But then I'd need yet another IC, a HC123. Maybe I should plop in a 555 in Hans Camenzind's memory. Now I almost feel guilty for not ever having designed in one of those.

I would too, but right now I'm using (several of) both because of the number of devices and some only come in one flavor. The problem with SPI standard is that it isn't one. We almost always end up bit-banging the interface because of some defugalty. Even so, I've had fewer problems with it.

There is a TinyLOgic HC123.

Dual optocouplers aren't much more expensive than singles. I think you can use one side for clock and data, and the other for CS*. Just one RC on the receive side can separate zeroes (short pulses) from 1s (longer ones.)

I wouldn't even need CS because there is only one resistance to be controlled. I think the digital potmeter works best. Two-channel optocoupler, the digital potmeter, done. Ok, plus a resistor and zener to supply it.

^^^^^^^^^^

Hah! I learned a new word today. Thanks.

What I found to be of cast-iron performance is RS485 and CAN but very few devices useful for my stuff come with that.

with

What, use the tick-tock up/down control thing?

Too easy, not fair.

with

can

method

Yeah, I know, it's too pedestrian :-)

for

English is well related to German and there is a dative case there as well. English and German also have genitive and reflexive cases.

Enjoy.

?-)

n for

English and German do have a common ancestor around 1100 years ago, but English has eliminated a few complexities that German retains.

-- Bill Sloman, Nijmegen

uck with

so you

, I can

method

es for

some

mented

And that's basically Manchester 2 encoding.

The original publication - IIRR somebody's thesis - listed a whole series of variations on this basic idea. Manchester 2 was the most popular.

-- Bill Sloman, Nijmegen

I don't see how they can be equidistant and not at 256Mhz