I got the MaxSerial kit today, soldered it together, and then figured out how to get the RS-232 data the 150 feet from my "laboratory" (where my computer is) to the basement (where the Arduino figures out when to circulate water through the solar panel). I'd considered using the 150 feet of shielded USB cable but opted instead to use a few conductors inside a mostly-unused CAT3 cable (I have a lot of CAT3 that I run to sensors and what not).  The data to and from the MaxSerial is very reliable, and it's great having the Diecimila-vintage ability to upload software any time without a reboot - I just change the software, upload it, and sit back and watch the effects.  It's like communicating with a probe on Mars, but without the speed-of-light latency issues.

Oh, for some reason I was using the term "MaxSerial" wrong - I meant a serial chip such as the Max232, and your use of it primed my brain with the right idea but the wrong word.  I didn't know about MaxSerial - it looks like exactly what I need - a serial version of the Arduino that plugs in where the NG was.  Thanks!  I'll look further at your stuff to see what's what. You should get yourself listed on this page:
http://www.arduino.cc/en/Main/Buy
or is it another Iduino-type situation?  Sometimes the distinguishing characteristics between different open source licenses and their commercial derivatives gives me a headache.

Does anyone know where I could buy an old Serial Arduino?  I could hook up a MaxSerial or retrofit one of my Arduino-capable & serial-equipped Olimex boards, but a Serial Arduino would plug right in and require a minimum of work.  If anyone here has one, are you interested in selling it? It doesn't even need to come with a working Atmega processor.

In this case I'm powering it with USB and "reboot" means to completely disconnect the USB cable, which kills the power to the board.  When I'm programming it remotely, I use a relay to jumper the reset button (in these old Arduino NGs you must reboot before reprogramming, though there's too much noise in a line that long to run the reset wire 150 feet without getting constant reboots - thus the relay).  This setup allows me to update software on the Arduino from my computer at the other end of the house as I add more routines and nuances to the solar panel control algorithm.

When you say "CAT6 with USB ends" - you mean just the connectors, no repeater hardware?  And that actually worked (albeit imperfectly) across 150 feet?  And without shielding?  I've actually thought about doing a simple serial link, since I have no need for high speed communications. I could just remove the repeater hardware and the USB connectors and use DB9 instead.  The cable is shielded and the individual conductors are twisted - even with primitive RS232 tech it should be able to work reliably for 150 feet.

I'm pretty sure from my tests that all my reliability issues come from the cable. I guess this is why it was discontinued.

Restarting the serial port from the Arduino would actually be an ideal way to handle the problem. I could do it every hour even, since in this application constant communication isn't essential, but eventual communication is.  Is there an easy way to do this?  I know how to do all the usual things with the serial port, but I've never seen an example of someone restarting it.

I should add, by the way, that the serial port lockups I'm talking about only affect the Arduino's serial port - the program it's looping through is unaffected.  But when this happens, my communication with the Arduino is cut off until I reboot it.

i've hooked up a laptop near the Arduino to see if the problems are cable related, but...

The 150 foot USB cable has hardware repeaters built into it and is exactly what it is -- a prebuilt USB cable, but it is cheap and Chinese, so perhaps it is the problem.  But why should the serial port require an Arduino reboot when there is a problem?  Shouldn't it be able to handle issues more gracefully?  I don't mind occasional occurrences of bad data -- I just can't handle the lockups.

I've had an Arduino-based circuit controlling a solar hydronic installation for one and a half years now (it's  an Arduino NG with an ATMEGA168 chip). This system is completely reliable, making sure water gets heated when the sun is out and turning off everything when the water coming from the panel is too cool to be useful.  I can monitor this Arduino remotely using a 150 foot USB cable (they used to be available for cheap from PacificGeeks.com until that company started charging a minimum of $35 for shipping and handling).  My question concerns the reliability of the Arduino serial port.  It seems to work for awhile (a few hours) and then it inevitably locks up, requiring the board to be rebooted in order to resume communications. I've seen this behavior with several NG boards, connected to both Windows and Debian machines.  At least on the Windows machine, I'm running the latest version of the FTDI USB Drivers. Does anyone know how to improve the reliability of the serial port?

i've only used the avr-pg1 with the Olimex avr-p28 board, which is a great cheap $16 substitute for an arduino, though it trades USB support for open breadboard space. you can examine the schematic of that board and compare it to the arduino to deduce the mapping.


avr-p28:
http://www.olimex.com/dev/images/avr-p28-sch.gif

arduino:
http://www.potemkin.org/uploads/Pid/arduino_usb.jpg

50 meters is a long distance.  If you want to go with a USB-only solution, you can try using cheap Chinese usb 1.1 extension cables:
http://www.geeks.com/details.asp?InvtId=100E&cm_mmc=Froogle-_-Cables-_-USBCables-_-100E
they're 30 meters, only $19 each, and if you put a hub halfway between them you could probably keep them going.

I'm using one of these lengths to successfully control an Arduino solar heat sufficiency controller down in the house's boiler room.

Does anyone know to what extent the Atmega168 and Atmega8 have different PDIP pinouts?  I'm noticing a strange behavior with my Arduino board when I use Atmega168 vs Atmega8.  When I set Arduino's Digital Pin 13 to input, I can use it that way successfully if the processor is an Atmega8.  But if it's an Atmega168 it always reads as 0.  Pin 12 works fine with both processors.  Now it's possible I have a defective Atmega168 - but I've seen this behavior with two Atmega168s.  


I just looked at the Arduino USB schematic
http://www.potemkin.org/uploads/Pid/arduino_usb.jpg
and see that there is a 1K resistor only on Digital Pin 13 - perhaps the Atmega168 is more sensitive to its effects than the Atmega8.

unless, of course, you're making a generic USB-to-one-wire interface with an Arduino board.  that would be a useful gizmo.  i imagine the protocol is some sort of serial thing with device addressing, sort of like usb but half-duplex.

i'd save those for a non-Arduino project, since the A/D would be wasted on it.  it's too easy to just use a cheap thermistor.  

hmm, one wire?  they'd have to have two to actually do anything, and then they'd be thermistors, in which case they'd be variable resistors.

1. measure their resistance at room temperature.
2.  find a fixed resistor with that measured value (or something near it)

3. connect one between VCC and an analog-in on the Arduino
4.  connect the fixed resistor between analog-in and ground.

This makes a voltage divider based on the value of your thermistor, and the numbers from reading the analog in will have a linear relationship to temperature.  you just have to find out what that relationship is and write a function to translate.  I keep everything integer in my thermistor readings to keep things simple.  Thus 989 means 98.9 degrees Fahrenheit, etc.

I don't know what was going on with my parallel programmer - I resoldered it and it seems to be perfect now. Shielding and such were unnecessary. I guess I'd done a crappy solder job before.  All the those resistors are buried in epoxy now and it is perfectly reliable.