Mark's 20 Percent Time

Yes, yes. I realize I've got several projects in progress and that I really don't need to start another one. But this one has a specific time-line associated with it, so I kinda need to bump it up in the queue.

The Wildflower Triathlon is coming up May 1st, 2010. Like every year, I'll be bringing my motorcycle to be a chase vehicle on the long bike course. Like most years, I'll have my ham radio and APRS gear mounted on the bike so I can provide the communications that are ostensibly my purpose for being there. Being a ham, I'd love to have a working radio on the bike year round too.

...but I also want to be able to listen to podcasts or audio books or music on my iPhone, maybe even make and receive a call while riding.

The iPhone 3GS supports voice commands for things like making calls and playing music. As long as I have a button I can press without taking my hands off the bars, and a way of feeding mic audio to it, I can control many functions vocally. And, of course, a USB charging port for the iPhone or any other device that charges via USB.

But I need to mix the audio out from the iPhone with the audio out from the ham radio, amplify it (so I can hear it over road noise and through ear plugs) and send it to the speakers in my helmet. I'd love to be able to adjust the volume of each input individually, and the master volume too, with big nobs that are easy to manipulate with gloves on.

And the mic audio from the helmet needs to be amplified (it's a noise canceling dynamic, both the radio and iPhone expect an electret) and sent to both the radio and iPhone. A PTT on the handle bars is key, and a channel Up/Down controller would be groovy too.

....

Holy crap, that's a lot. But the reality is, it's not very complex, just a lot of connections and a bit of audio work. I've got a circuit mostly designed and am currently selecting components and working on a board layout.

There are some complicating factors, though. The project box I'm putting this in is a plastic two piece top/bottom thing. Both halves can mount a board. The "top" part is mounted permanently to the motorcycle. The "bottom" part (the part into which you put the screws to connect to the "top" part) is removable. There is 5/16" between the board and case on the "bottom" section, which is just about right for the pots and Mini-DIN connector for helmet audio, and possibly for other "front panel"-ish bits. There is 13/16" between the board and the case on the "top" section, perfect for things like RJ45 and other tall-ish connectors. (I think those are actually 7.5mm and 20mm (respectively) rather than 5/16" and 13/16".) There is 9/16" between the two boards when they're put together. Selecting the connectors to mate these two is proving to be the hard part.

And ExpressPCB is pretty cheap to add the second board as part of the first board if you're willing to cut them apart yourself.

TODO:
- I've got the circuit sent to a few friends looking for comments and suggestions. Already found a few bugs, which is good.
- Need to find a better 5vDC regulation system. USB spec is for 500mA at 5vDC. Regulating that down linearly from 14vDC is 3.5W of dissipation. Ouch. Would very much like to switch this instead; just need to select the right component.
- Find the right connectors to mate the two boards back-to-back.
- Figure out how to convince an iPhone to accept a charge over USB.
- Finish board layout and component selection and build the sucker.

So, yeah. About that audio mixer project I was working on. Turns out, a friend of mine with whom I'm working on Vesuvius was selling his MaxNC 15OL CNC mill. So I bought it.

That took about $1200 of my $800 budget. No, that's not a type-o.

So I have precisely zero money left with which to buy parts and circuit boards for the audio mixer project. The good news is, the radio bits I need for Wildflower already work; they just don't mix in iPhone audio. So I'm in good shape for the hard requirements of the event.

BUT! As far as Vesuvius is concerned, I have a CNC Mill! So that's cool. I'll post more about the mill on the Vesuvius blog later tonight.

Tonight, I had another idea. I want to make a clock (big shock, 'eh?) that is just a couple stepper motors moving a whiteboard pen and eraser around drawing an analog clock. Though, strictly speaking, it could be made to draw a digital clock too. A servo would push the pen down, or lift it up and push the eraser down, or ride in the middle lifting them both up.

I have the micros, steppers, and servos. Just need the frame, threaded rods, and a few bearings. I think this could be done pretty easily and cheaply (read: with little to no budget.)

The real problem is, I've got SEVERAL other projects I should be working on, like a damn steam engine! *sigh*

This whole week has been vacation for me, a bit of a sabbatical, really. Went to Disneyland for three days in the middle of it, so I didn't get as much done as I would have liked, but I have made some good progress on somethings.

Today's progress has been on getting email migrated from Chiba (old server) to Puffer (new server.) I've successfully gotten ClamAV plugged into email received on SMTP, and am working on SpamAssassin, but my eyes are starting to cross so I'm going to bed. Yes, it's before midnight. Yes, I'm getting old.

But before I go to bed, I have a question for anyone reading this who understands such things: What is the benefit of using Milters with Postfix as opposed to just doing standard SMTP style before-queue content filtering?

If I'm reading it correctly, Postfix supports Milter so they can talk to off-the-shelf plug-ins that support the Sendmail Milter protocol. However, the plugins I'm wanting to install (SpamAssassin and ClamAV for starters, possibly others) seem to only support Milters using a additional glue-layer package that talks Milters on one side and the normal SMTP mechanism on the other.

If this is correct, then it seems that using Milters with these particular plugins is just MORE overhead. The question is, do I get any additional benefit by using Milters?

If not, I'm thinking of simplifying things and just ripping the Milter layer out and either doing pre-queue filtering, or just going lazy and doing post-queue filtering.

Anyone have any thoughts? Feel free to hit me up on GTalk, AIM, Yahoo and/or Twitter as @SmittyHalibut, or throw a comment here. Thanks for your help.