Searching for I2C literature

it

The Microchip C18 I2C library routines will wait forever for a flag to be set if something goes wrong. I rewrote the ones I needed with wait-and-check-loop limiter s.

A lot of the decline in I2C may be the continuing licence fees that mfrs have to pay Philips. The other solutions are just as good and royalty free. SMBus is almost like I2C except it _isn't_ specified to DC (a good thing[tm] in hotswap environments) and is what is usually used in computers for systems management. (Well, that is what SMBus stands for ;)

Cheers

PeteS

Hi Joerg,

Same here. I've never even done a multi-master system, although I'm told they can be quite nice.

True, although given the relative ease with which I2C can be bit-banged, I doubt it's going to completely go away anytime particularly soon. I've seen people take "legacy" designs using those little 8 pin I2C EEPROMs and replace them with 8 pin microcontrollers that provided just as much storage and extra functionality to boot. Even if one isn't using any "hardware" I2C devices, as a protocol I see value in I2C in that newer scopes or PCs with a little interface board and software can be used to trace/troubleshoot the bus -- definitely preferable, in many cases, to people inventing their own protocols when I2C will get the job done.

Not the original one... if you read the message (on SED here) titled, "Dishal tuning question" is discusses what I did instead... building a capacitor-coupling resonator filter using the same "style" of build as the original, and finding that it worked pretty much as designed, but questioning if building, e.g., shielded sections and removing ground plane copper would make it even better (something I'm trying today or tomorrow). If those techniques help, I'll go back and try to make an "improved" version of the original.

Based on the results of the cap-coupled filter, the Big Problem seems to be obtaining decent finite frequency zeroes; if I don't get them working I probably will just stick with the coupled resonators.

---Joel

Hello Joel,

Absolutely. And if I2C indeed vanishes into the "just being a standard" category there ought to be a new standard that doesn't require a home-run architecture or complicated selector line structures.

I only saw the last four posts there, my ISP (or one in the fwd loop) sometimes drops a slew of posts. But that's what I suggested in the a.b.s.e. thread, removing the ground plane under the traces around L6 and L6 itself. I even had filters where milling away the FR-4 underneath an air coil helped.

Whenever I prototype filters I grab lots of thin steel, 240-grit sandpaper to take the cutting edges off, and the first-aid kit. The FR-4 board in there merely acts as mechanical support, it isn't really a "circuit board" anymore. Still, a Q of 100 at 200MHz as the mfg claims in your case remains a dream. Unless you are building a cavity resonator and polish it daily ;-)

Best investment here unless you already have it: A dip meter that goes up to a few hundred MHz. That'll tell you smack dab where a resonant circuit sits without having to touch the circuit.

Hi Joerg,

Here's the core of the original:

---

After the not-so-great bandpass filter I built last week, I built another one this week using a capacitor-coupled resonator topology. Happily, it works pretty much as designed! However, I did discover that the "real world" method of Dishal tuning isn't quite so clean as the literature would have you believe.

Articles such as this one -->

(which, ironically enough, is trying to sell you on laser timming rather than trimmer caps, which is what I used :-) ) have, e.g., figures 3 and 4 where the first and second resonator, respectively, are tuned. Well, I can get figure 3 just fine, but when I go to the second resonator, the two peaks in figure 4 are asymmetrical in amplitude, and I have to re-adjust the first resonator to make them symmetrical again, kind of iterating between centering the second resonator's dip and making the first resonators response symmetric.

As you could imagine, this problem tends to compound itself as you move from resonator to resonator. So, I'm curious:

1) Presumably the assymetries in one section arise due to the additional parasitics present when you un-short the next section? Or perhaps they're due to a little through-the-air coupling between sections? I'm making another board that shields each resonator section to find out... 2) I eventually was able to tune up the entire filter reasonably well by deciding that I was only going to tweak the "current" section and the immediately prior sections, getting things as centered and symmetrical as possible (a somewhat iterative process). Trying to go through multiple prior sections became as exercise in futility, and I rationalized that the further away a section was, the less impacts the parasitics of the current section had on it -- hence the idea of only considering the current section and the one immediately prior. How's that sound for a tuning procedure?

---

Yep, that is what I'm trying. I have the bare board here now; I just need to go find a late lunch and start populating it.

I've been making do with foil tape, although having something a little stiffer around would be nice.

You think their measurement techniques has problems, eh? Interesting...

I don't have one; I'll pursue obtaining one.

---Joel

Hello Joel,

To be honest I never had much luck with tuning methods that required shorting of parts. Shorts in the 200MHz world aren't really shorts but small inductors along with some stray capacitance or dielectric displacement introduced by the fixture that effects this short.

After a (long) while you'll get the hang of it. It's like tuning the carburetors on an old but hot Alfa Romeo. Even in Italy the guys at a repair shop told us "Oh, only Giuseppe can do that stuff really well so you'll have to wait until tomorrow". Next day Giuseppe said that after a proper tune you should be able to place a coin sideways on the engine block without it falling over. He was well into his 70's...

Both. You can muffle over-the-air stuff but not the board level coupling.

It sounds good. The usual trick is to tune each segment only a wee bit, then move to the next. Then back again for another notch, and so on. I even did laser trim procedures that way despite the fact that it required a lot of target movement. Excessive target movement is frowned upon because it costs time and laser trims are usually charged by the number of seconds the process takes per board.

That's a really late lunch. Might have to skip dinner then ;-)

I did thin beef ribs on the barbie. That's the beauty of having the office next door.

Copper tape isn't cutting it for this stuff. I use rolls of 1/2mm and less of copper. In a pinch (when at a client) I have cut up butter cookie cans, the ones that aren't colored. The nice thing is you can get them at the next food mart. The other engineers helped in making sure it was empty in a jiffy ;-)

Sometimes I get the impression that such data is often measured in super-ideal conditions. No other materials in the vicinity and somewhere on the international space station.

Absolutely. If you do this LC circuit stuff a lot it is a must. I couldn't imagine working without one.

Now imagine doing this work without a network anaylzer. Before I had one I managed to wear out the bearings on the encoder of the comms receiver. So bad that the knob now scrapes the front panel.