And when you say 'transfer data', what do you mean?

Running Linux on both machines, I'd set up a daemon process on the laptop to read the serial data and stick it in a file. That way, you won't have data loss if your ssh session dies for any reason. Then, the daemon process could also use use something like netcat to transfer the data.

Ah, much better once I know to just specify SOT-23 packaging. Sparkfun, e.g., has adapters.

My hovercraft is full of eels.

 

Thanks CrossRoads.

So I'll just have to learn to live with SMD, and buy/build PSOP-DIP adapters. (Or SOIC, SSOP, etc).

[ETA:]

Go with a single transistor like 2N2222 or BC547, any package you're comfortable with.

Well, I was specifically thinking of using your inverter circuit driving a bi-color.

ETA: ... because it seems a lot cleaner, uses fewer components, etc.

I'm doing some window shopping for components, and maybe I'm being too fussy about it, but I noticed that the 74LS04 is a DIP-14 package, with 6 inverter circuits, so I was wondering if there's a smaller package, maybe even a single inverter in something resembling a TO-92 (except with 4 leads). Not that board real estate is something that concerns me now -- but if it did ... ?

And then I noticed the output current rating on the Fairchild 74LS04 (PDF) and thinking I can't drive an LED with that, or it'll be pretty dim. Same thing for one single inverter I found, the ON M74VHC1GT04DFT1G.

Looking for something with a higher current rating, I find things like the Toshiba 7SZ04F. Seems like a lot of these things are SMD. (Not that I mind SMD per se, but it ain't handy for breadboarding.)

Any advice / suggestions?

Yeah. Red + green = yellow, but the exact shade would depend on the color temperatures of the LED in question. In another week, I'll have money to throw at this, so I'll toss appropriate stuff into a Digikey order.

Try this then.

It just occurred to me that I could use a varying duty-cycle PWM signal to get a range of colors from such a device. That could be useful for a gross indicator of over/under some condition.

Just popping in because I want to follow this discussion, especially since it'll probably get to the point of waterproof sensors at some point. A forum search for 'waterproof' comes up with a lot of threads, and a web search for waterproofing yields many suggestions, but none that I've seen that specifically talk about being safe for fish. My own interest is in only water temperature, since I have just a little fresh-water tank.

A couple of thoughts for that:
1) Just buy a DS18B20 in a stainless enclosure
2) Find some stainless tubing and seal the sensor inside using aquarium sealant
3) I've read that the DS18B20 is already waterproof (older versions were notorious for moisture issues)

My somewhat limited experience with silicone aquarium sealant is that it doesn't always bond well, so I don't know for sure that I'd trust it. I'm pondering pure acrylic hot-melt.

[ETA:]

... You can get the LM35 in a TO-46 metal case for good coupling to your heatsink. ...

I don't believe that this is a viable option as the LM35 has a minimum operating temperature of -40C. What happens if the temperature drops below -40? Does it continue to read -40, or does it create an error message somehow?

Well, it'd help to know what the operating temperature range is. My assumption is that if they're reading the heatsink fan RPM, as driven from a motherboard header based on what the BIOS thinks the CPU temp is, then -40C isn't going to be an issue. I also assume attaching the sensor to the heatsink as close to the CPU as possible. But I admit those are assumptions.

ETA: Actually, going by the OP:

... that is currently controlling the fan mounted to the CPU heat sink.

That implies air inside the computer case. Unless he's (she?) using something like a very dry inert gas (which I think would be a poor thermal transfer medium). But let's talk about air. Cooling the interior of the case with LN₂ is going to result in condensation and then frost on the motherboard and components. Maybe that's not an issue. Again, more info needed. (See also Computer Cooling: Liquid Nitrogen.)

I'm not completely grokking using liquid nitrogen, when you're still using a an air-cooled heatsink with a fan. I guess I'm picturing LN₂ being used in a manner similar to water cooling. I suppose you could be just cooling the internal case air so as to then more thoroughly cool the heatsink itself.

But I wonder if using something like an LM35 wouldn't be easier than trying to pick up fan control PWM. Really, you're inferring more/less coolant flow based on the PWM frequency, so I think you can just as well make the same infererence by measuring the heatsink temperature. A bit of experimentation should get you temps to use for when to increase/decrease your flow. You can get the LM35 in a TO-46 metal case for good coupling to your heatsink. No, it won't be your CPU core temp, but I'm guessing it'll follow it closely enough for you to set your flow rate.

reconnecting it with a diode to prevent current flow. Probably there's a good reason to not do that.

This is a hairy area.  The diode is going to drop 0.7v.  So 5v - 0.7v is 4.3v.  And I am assuming a nominal 5v.

Ah, thanks for pointing that out.

So, the upshot is that if you're going to use an Arduino in a context where you're collecting data, and show up once a day or week or something, and plug in a laptop to retrieve the data, and you can't pull external power before doing so, you can't use the USB port for that. At least if you're using the Duemilanove or any model with the same power select logic (and haven't replaced the MOSFET). Am I understanding this correctly?

I have a Belkin powered USB hub, and now I'm wondering whether it has the same problem -- obviously there's no comm involved, but something like a bus reset if I happen to be using it unpowered, and then connect the external power.

This sounds like what's been discussed:
http://www.david-laserscanner.com/forum/viewtopic.php?f=9&t=2261
http://www.arduino.cc/cgi-bin/yabb2/YaBB.pl?num=1284034373

I'm not knowledgeable enough to know whether it'd work, but I was wondering about cutting the USB power pin, and reconnecting it with a diode to prevent current flow. Probably there's a good reason to not do that.

Hey, that Mims name rings a bell. Did he write anything like a 555 project booklet? Anyways, I'll bookmark that.

Heh. I guess part of my brain is still working after all. I have this: 555 Timer IC Circuits

Hi cr0sh.

Hey, that Mims name rings a bell. Did he write anything like a 555 project booklet? Anyways, I'll bookmark that.

As for expensive hobbies, I seem to have a talent for finding those -- astronomy, photography, and (speaking of turning money into smoke) guns.

Sadly, my Beckman DMM died years ago. I now have just an el-cheapo, but it'll tell get me by for a bit. I might look into something like a USB o-scope kit -- some of those are cheap. I haven't looked at logic analyzers at all. If a pile of money landed in my lap, I'd be looking at a Fluke 179 and a Rigol o-scope. I noticed LadyAda had some low-end DMM recommendations, but I prefer to spend my money on something good that will last. I saw a vid at EEVblog comparing cheaper DMMs, and it pretty well put me off any interest in the off brands.

I did look around at various "experimenters" kits, such as those at Sparkfun, but after looking at prices, my bang-for-the-buck works out better buying exactly what I intend to use right now. I think the risk of letting the smoke out on a small 1-wire network is pretty minimal.