Category moved from livejournal

Working on the HMLiberator

Oh, and as mentioned in the previous post, it works. Well, mostly. I had the Honda Music Link passing audio from the iPod and communicating with the HMLiberator, but then I ran into a bit of a snag. In the byte I was keying off of for a response (the checksum of the packets) I’ve run into a collision, so I now need to come up with a better way to selectively respond to commands from the HML.

I think I’ll try matching off of the actual command itself. This might not be too difficult, but either tomorrow or maybe over the weekend I’ll try and flesh out a good way to do that.

For now, though, I’m just happy that it’s proven to be possible to do what I want. :D

The EasyPIC4, iPod, Honda Music Link iPod Cable, and serial breakout cable,
all connected and ready for testing.
(Click for more photos of HMLiberator development.)

Well, it pretty much works. By “pretty much” I mean that the Honda Music Link talks to it, the HMLiberator replies, and things seem okay. I’m stuck at one point, where the Honda Music Link (HML) asks the iPod (really, the HMLiberator) for information about the two playlists it reports containing, and the HML doesn’t seem to like the answer. I think I screwed up at least one of the packets, so I’ll look into that tomorrow and do a bit more testing. The good news is that it all generally works and right now the problem is software related.

I’ll try and fix it tomorrow.

I uploaded a few more photos, which I’ll just mention here:

· Inside of one of the MiniDIN 8 connectors, with fingerprints for scale.
· Completed straight-through cable, to be split open for serial data testing.
· Cut open data cable, ready to tap into the serial TX and RX lines.
· Completed serial RX/TX tap, except I forgot to connect the ground line. Whoops! It works when grounded through the sniffer, though.
· My Honda Music Link sniffer. A level shifter, strip board, and some DB9 connectors.
· Detail of the top of the HML sniffer.
· Showing where the HML sniffer taps into the serial TX and RX lines which the HML uses to talk to an iPod.

This part is probably even less interesting to most people, so I’ll hide it behind a cut:

Click to see…

HMLiberator Schematic First Draft
(Click for 150dpi version.)

Well, there’s that. I think that’s the appropriate first draft of the schematic for the HMLiberator. All parts are SMT, including the SOT-8 voltage regulator and crystal. I’m still not sure how the cables running to the Honda Music Link and iPod will be terminated, but for now I’m thinking that the wires will simply be soldered into the board, then the body of the cable strapped down to the board for strain relief.

The whole board is intended to be considerably smaller than the Honda Music Link itself, capable of being strapped to the HML and sort-of piggyback on it. It will likely be cased in simple shrink-tubing, as being all SMT and low profile this shouldn’t be difficult to do at all.

The ICSP connector will be a right-angle .100″ connector on the dev board, and just holes on the production ones (if I do any production runs). I figure the right angle connector can be set in the holes and held in place during programming, and any purchaser who has the means to flash it themselves likely has the means to add the ICSP connector itself.

Now, to fix the FIFO problem and lay out the board. (Hopefully I’ll get both done by the end of the week.)

Oh, and if any of you can see any problems with that design, please let me know.

EasyPIC4 w/ PIC16F688 at 12MHz with
rigged connection from hardware UART to on-board level shifter.

After some jiggery to get the PIC16F688 sample I received today talking to the built-in level shifter on the EasyPIC4 I had no problem getting the basic software for the HMLiberator running and responding to commands. It’s currently running at 12MHz, but as soon as I get some 20MHz crystals it’ll run there. This will give a baud rate just about 0.16% skewed from the specified 9600 baud rate, and I think that’s close enough.

I still have one problem to sort out before I try plugging it into the car, and that’s with the receive FIFO. With it set at the default of 40 bytes I can get five commands sent and responded to, and then the sixth just fails. As I was testing with a seven-byte command (and 7×5 = 35, just below 40), I think I’m hitting point where the FIFO wraps around (or something — I don’t understand it well enough yet). The incoming commands are of varying size, so I’m not really sure how to work around this yet.

At least I know where the problem lies. Once that is sorted (or whatever) I’ll plug the thing into the car and see what it does. If that goes well, which I expect it will after some basic software changes, I’ll finish up the v1.0 circuit / board design and order a run of boards and make up a first batch of five or so.

Hmm, the software for the HMLiberator^† is (almost) working. It seems to respond as I hoped five out of six times. The next time it just seems to not do what I want. I’m not sure how to debug it, though, as I can’t really trigger an interrupt with the device in hardware debug mode.

This might take some thinking.

That said, it’s time to go home and build the appropriate cable so I can actually sling it in between an iPod and the HML itself for once.

^† The HMLiberator is the PIC-based device I’ve been cobbling together to sit between the iPod and the Honda Music Link, translating the Mode 4 remote commands into Mode 2 commands. This will allow the user to keep using the controls on the face of the iPod, while the HML thinks it has one locked down and under its control.