KiCad, which I mostly really like, has some 3-pin library parts with non- standard pin locations. So when you package a schematic, you have to make sure you're choosing the _right_ SOT23.
And, the pinout of a SOT89 (I think I have that number right) is a constant trap, if I'm not paying attention to choosing the correct schematic symbol in the first place.
I've been bitten by the way they spec comparators with a huge 'overdrive'.
Every possible SOT23 pin numbering scheme has been used by at least one chip vendor. I believe that every possible 4-pin SOT143 has been used too.
We use the "Motorola" pin number convention for all SOT23s, which means that the pin numbers on our schematics may not match the pin numbers on their data sheet. We got into troube once when someone entered a new part with the Zetex convention, and that messed up the decal, and that broke all the other parts in the library.
We buy one SOT143 transistor that is the mirror image of the most common 1-2-3-4 convention. That part has the pin1 dot on *the bottom side*
Maxim has a comparator with back-to-back diodes across the inputs!
I remember you mentioning that one... it's filed under both 'lessons that don't need repeating' and 'reasons to avoid M*x*m'.
LOL, it was almost 9 years ago:-
May as well repeat the good stories now and then, for the newbies.
That's a mistake, IMO. The pin numbers should *always* match the datasheet. The librarian at teh PPoE changed all LEDs to have the Anode Pin-1 (IPC standard, or so he said) but Avago didn't agree. We
*always* had trouble with the parts. My CPoE does somethign strange with connectors (have no idea what their "stanard" is) and they're constantly causing trouble. If they'd just use the datasheet convention they'd save a lot of board spins.That's useful!
Disagree. We'd need a zillion different SOT23 decals in our library. And if I looked at a schematic and at a board, I wouldn't know how the pins are numbered, so I'd have to fetch the data sheet and figure it out for each part. It's convenient to know that the lower-left pin is always pin 1.
So what's the "Motorola" pinout? The one I see used most often numbers the two bottom pins "1" & "2", and the top pin "3" -- just as if it were an IC package with only three pins.
That's Motorola.
Strange, the datasheet of the MAX9691 now lists a maximum differential input voltage of 5 V. Compared to rail-to-rail operation (which would be
10.2 V in this case), 5V looks pretty low, but hardly low enough for diode limiting.Did they change the part and/or the datasheet? Or do they just expect "finite" driver impedances to be high enough ...
Dimitrij
Your SOT23s aren't any different from anyone else's. I'd prefer to have to look up pin-2 (pin-1 is usually marked ;) than spin a board. By your logic schematic symbols should look like the part, too. They do make debugging easier [spit!].
People used consistant pin numbers on octal tubes, and DIP14s, and SO8s, so why not SOT23s?
In our shop, they do.
But pin1 marked on a SOT23? Rare if ever.
Q: Why is the sky blue? A:Deal with it.
Thinking about how I use them, I just don't probe the SOT-23s. There are usually resistors around them. Even 0402s are easier to probe, anyway.
It's probably a differential NPN stage, and it's bad news to zener a transistor b-e junction. The discontinued and unreliable MAX9690 (which the 9691 was claimed to replace) was rated +-3.5 volts max diff.
I don't think they changed it. The graph of input current vs diff voltage peaks at 4 MILLIamperes, so they probably have a 500 ohm resistor in series with each input, before the base clamp diodes. So 5 volts won't damage it, but it will leak mAs of current between the things being compared. That's the real problem, a crazy thing for a comparator to do.
Never buy Maxim.
The fast ADI ECL comparators are good for +-7 diff, except the insanely fast (180 ps) ADCMP580 is +-2; I think that one is SiGe.
Some opamps have input diodes, but that's usually an asset in an opamp; free integrator anti-windup. I sometimes use opamps as comparators, so have to think about that.
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
^^^^^^^
Cripes! Never noticed THAT graph! Yep. One can read the resistor values right off the bias current slopes...
To do something like that it probably takes both a committee and the CxO's nephew. Who in his right mind could ever design that atrocity?
Well, thanks for the warning. One more type of graph to watch out for.
Dimitrij
I was using the 9690 to compare a linear ramp (current source driving a cap) to a DAC voltage, to make a programmable delay. They quit delivering the 9690 for a couple months, then announced that it was discontinued. The 9691 was offered as its replacement, but it broke my circuit. We had to make a couple thousand adapter boards to replace the failing 9690s. Eventually all of them in the field would have failed. Same nephew must have designed that chip.
Really, you have to force yourself to read the entire data sheet, including the specs, the graphs, the test circuits, the application hints, and the footnotes, to find the hidden gotchas. And then be lucky.
Unless it's Maxim; all you have to do is see the logo.
I managed to do that on a BAT54S -- t'was a schematic pinout error in that case.
For anyone who happens to swap pins 1 and 2 (which reverses the "supply" pins on a BAT54S, if you're using it as a clamp), there happens to be a BAV99 variant in SOT-323 that's easy enough to tack down, and has the 'reversed' pinout.
Tim
Have something to add? Share your thoughts — no account required.
Ask the community — no account required