Old Palm as serial terminal

As I started setting up to use ATTiny85s with Arduino IDE and ArduinoISP as programmer, I quickly came across the “How do I get debug output?” question.  While the chip doesn’t have a UART, there is a Tiny Debug Serial package for it, so Serial.print() can bit-bang out PB3.  But what do I display it on?

I could use an Arduino to take serial in and send it out to the serial console.  But I’m not sure I can do that and keep the ArduinoISP sketch in place – which I’ll need frequently to download new code versions.

I considered getting a serial LCD as a general purpose display, but they’re a little expensive, so I thought about what old stuff I might repurpose for this.  My old Palms!  I have a Treo and a Centro that got outdated before they got broken.  They’re real bored these days, and have serial I/O, so I started digging.

The two devices both use the Palm “MultiConnector” and the same cables.  Here’s a doc with pinouts.  The serial is 3.3V TTL.  Of the various cables I have for the two, only one has physical pins in positions 10,11 where the serial is.  I took it apart, and found those pins were not connected.  (It only had USB wires Gnd, +5,  Data+, and Data-.)  But there are enough wires, so I should be able to move some.

Done!  The range of small pins (L to R from the back) is 17-5.  White and green are now on 11 and 10.  The original black – via one of the yellows – goes to 8, so I’ve got ground.  After it’s all done it looks a lot like it did before I started.  But now white and green ring out to the RX, TX pins .

The Tiny Debug Serial package included with current tiny 45/85 Arduino cores is almost a drop-in for the normal Serial stuff, automagically figuring out if the target chip has no UART and slipping itself in.  The package is send-only, but that’s really all I need.  Just added a Serial.begin(9600) and some Serial.print()s, just like normal.  I added a startup message and print the value read from the ADC once/sec.  The analog in pin is currently hooked to a pot for testing.

I found an old Palm Hotsync manager and reminded myself how to install stuff on Palms.  Found, downloaded and installed (free) ptelnet-0.61.prc to the Centro.  Got it running on the serial port at 9600 and hooked Gnd/RX of the new cable to PB3 (pin 2) on the Tiny.  Since the Centro is 3V logic, I used a 3.3K/10K voltage divider from the Tiny pin to protect it from the 5V logic levels.

And it works!  I’m just using the built-in 8MHz oscillator divided by 8 for a 1MHz clock.  Seems to be accurate and stable enough to run at 9600.  Woo hoo!

Now that it actually works, I terminated the cable for convenient use with a breadboard.  It’s even labeled for TX and 3 and 5 volt RX.  Hmm, seems to need a pullup to work.  I don’t understand that, but it works.

Update 8/30/11:  I have another Palm, so I’d like another cable.  The hot sync cables lying around have the wrong pins populated.  They needed USB, not serial, and so populated positions 5-8 and 12.  Let’s see if I can move some.

I opened up the connector and found a little PCB to hold the switch – no surprise there.  But the pins seem to be soldered to both sides of the board.  How’d they do that?  And how did they get spring-loaded pins in there in the first place?

After meeting the challenge of unsoldering the board without doing too much damage to the pins (though the plastic suffered a little) I gingerly pulled a couple of pins out.  That answered both questions:  Clever manufacturing provided the spring.  And the pins are reversible, with the solderable finger offset from center by about half a board thickness.  The good news is that I can just pull them out and relocate them to the positions I needed.  The other good news is that I can take advantage of the pin offset to give myself a little more working room on the two adjacent positions (10,11) I need pins for.

Of course the existing wires (molded into the strain relief) were too short, so I had to extend them before I could connect to the moved pins.  And let’s hear it for heat-shrink!

I only needed 3 pins (Gnd, TX, RX; 8,10,11) but I put the extra pins back in so I’d have them if I needed to disassemble it again.

A quick test and it works!  Haven’t terminated the other end yet.

Update 1/19/12:  The goal of this project was a cheap, simple way to get serial debug info at TTL levels from a micro on a breadboard.  I’m now using another approach that mostly works better for me.

While the Palm is cool and portable, it has a couple of drawbacks:
* The connector is flaky.  I have to keep pushing it in to make sure it works.  I think what I did is OK, it just doesn’t snap very securely into the Palm.
* The terminal emulation I’m using responds to BREAK (line not held high) with an error popup and stops.  I have to clear the error and then click the tiny On button to go online.  Power cycling the micro produces a BREAK condition – and it’s a hassle.  Another terminal emulation would probably work more gracefully, but I never got around to finding one.
* Since it’s battery powered, I feel like I have to turn it off all the time except when I’m actually looking at it.

My PC has several real RS-232 ports, so if I had a TTL to RS-232 converter, I could just run a terminal emulation in its own window and overcome all the problems with the Palm.  Not very portable, but I’m in front of the PC most of the time anyway.  Yeah, I could make something with 1488s/1489s or a MAX232, but there’s still the issues of power and 3.3 and 5V devices.

Taking advantage of the fact that most RS-232 implementations don’t actually require negative voltages on input, I used this cheap and cheerful approach:

It wasn’t even worth making a PCB.  Each side provides its own appropriate positive voltage.  The RS-232 end (PC) must assert DTR true to get a positive voltage source, and by connecting the TTL side to local power on the breadboard – 5 or 3.3V, it works happily with either.  And no external power required!

I’m still happy to have my Palm and cable as a backup, but this is now my main serial TTL tool.