Alpine bus protocols

C4M · 07-13-2004, 02:28 AM

Hopefully this is of use to someone...

http://kiora.ath.cx/alpine/

http://www.cus.org.uk/~cleggy/discus...rd-topics.html

http://www.mictronics.de/?page=cdc_proto

C!

Hexerei · 07-13-2004, 03:22 AM

This means that I can actually send button presses on my alpine head unit to my PC right? (IE: changes tracks and such like I do now)... and send information back to the display? If so.... you kick much ***

I've been wondering about this stuff since I will be removing my Alpine MP3 CD changer when I put the car PC in.

Just gotta get the video working on my MII 10000 board (in gentoo linux) and then wait on xcar or headunit for a good front end.... atleast it gives me time to save $$ for the other parts, and time for prices to drop

... I can't wait to get this installed!

jics · 07-15-2004, 11:27 AM

Hi,

As i'm currently working on an ai-net interface, it would be interesting what infos were available on this dead site (http://kiora.ath.cx/alpine/).
Is there any mirror available ?

The funny thing about ai-net is, that there's a lot of wrong and misleading info circulating on the net. From my point of view, only very few people have really checked it with an DSO (either that, or there are at least 3 completely different versions of ai-net in use).
If anybody knows a relieable source of info, it would be greatly appreciated.

greets,
jics

Gruneun · 07-15-2004, 12:07 PM

Quote:

Is there any mirror available ?

http://www.google.com/search?q=cache...a+alpine&hl=en

jics · 07-15-2004, 01:24 PM

Hi,

Well, in the meantime i had the same idea, but unfortunatly the protocol described is the old m-bus system, not ai-net.
The only way to get info's on this system seems to be the old way: find out on your own...

snoop · 07-19-2004, 05:52 PM

I am a half decent programmer but im not sure how to sauder all the chips for a (AI-NET to Serial Cable) connecter. I am not sure where to start, all these links above are outdated and dont talk about CD-Text and whatnot. Anybody have any info that can help me.

jics · 07-20-2004, 02:06 AM

A simple cable doesn't work. You need a intelligent converter, based on a microcontroller, fpga or something simillar. The electrical and logical workings oft the ai-net have nothing to do with a RS232.
Maybe it is possible to use a parallel port for reading messages from the bus, but i don't think that you could generate the correct timing for writing to the bus.

MartyM · 01-26-2006, 12:59 AM

I have been peeking at the Ai-Net bus and got all the jack pinout information, but the protocol appears to have a (bus) bitspeed of about 100kHz.

Alpine uses custom ASICs (custom chips) inside the devices where main processing is concerned. The Ai-Net dirty work may be handled by a microcontroller, for example.

I hope to find a means soon to capture at that odd speed. If so, a "sniffer" or comm. monitor program can be used to capture bytes on the bus and decipher the functions.

It's more of a matter of figuring out the protocol. The hardware side of it is easy. I believe (until I figure out otherwise) it is based on RS-422 standard. A microcontroller and cheap RS-422 [or similar] chip will be able to handle the details once the protocol is uncovered.

You can't get access to the program code Alpine uses. It is secured inside the ICs. However, the good thing is that it doesn't matter as long as you can monitor and record bytes traveling on the bus.

I want to use Ai-Net too, but not for the same reasons (I don't have a car PC).

Anyone wish to "compare notes" ?

jics · 01-26-2006, 02:29 PM

Hi,

Well is't been some time since i've loooked at this thread....
I haven been working very much on this topic in the last time, but her comes what i have found out until now:

- The pinout is documented (check ainet.png)
- The bit-level protocol runs at the bitrate of the AD-converter (44.1 khz -> 44.1 Kbaud)
- The waveforms for a bit and a byte are attached to this post.
- The protocol seems to run in a singemaster-poll mode. The master (headunit) polls the slave (a changer in my case). The slave does not send on it's own. This may be wrong, but a multimaster-bus seems very unlikely to me.
- Unfortunatly I was not able to go any furter, as my DSO died almost a year ago and I had to spend my money on other things since then...

Hope this helps...

vines

MartyM · 01-26-2006, 03:21 PM

Hi, there is only one problem...I noticed you have a single line bus while I noticed a two-conductor (differential) bus, similar to CAN. These were on, I believe, what would be pins 3 & 2 of your diagram.

I mave have been wrong about the RS-422 idea. CAN seems much more likely, due to the voltage levels and behavior. Also, I *Thought* I measured a bit width of 10 uS but it was my first time using the capture feature of this o.scop.

Reference system: CDA-9805 and KCA-410C

The KCA-410C has a Fujitsu device which I previously thought was a microcontroller; however, Fujitsu is far too protective and will not release a data sheet for this device (MB88385A) even for use by my company! A rep. there did state:

"...regret to tell you that MB88385 is simply the automotive network driver IC, and no CPU is integrated. Thus, if you like to use this device, you need to have MCU/CPU to control this peripheral chip."...and it was implied that it is controller area networking (CAN).

Alpine's custom IC/processor did use a different speed going into/out of the Fujitsu device.

So, assuming I can check on the bit rate again, I should be able to monitor the differential lines with a CAN-ready microcontroller/transceiver and send to a PC for sniffing (assuming I'm right about the bus protocol).

Ok, well I'm up for more ideas! I need to try out that idea or try a CAN-RS232 converter once I'm sure about the speed.

jics · 01-26-2006, 04:36 PM

Hi,

Of course it's ia a differntial signal. What you see in the jpgs is the Data+ line.
The coding is RZ, not NRZ, so its no 232,423,422, 485 or CAN. That would be too simple...

The voltage levels (i dont remember the absolute values) are lower than an RS* or CAN line.

Because or the RZ coding it could indeed be possible that the datarate is variable, because the clock could be regenerated by the receiver from the rising edges of the data signal.
What speaks FOR for this theory: You observerd another datarate than I.
What speaks AGAINST this theory: More complicated receiver/transmitter design.

BTW.: Have you found a source for the AI-Net cable connectors (specially the female ones) ?

greetings,

jics

MartyM · 01-26-2006, 05:02 PM

Hi, I realized after my last response that you in fact intended to note a diff. bus, but your label threw me off at first.

Note: I only have a few hours worth of experimenting with this so I know my measurements were not finished and I'm probably missing a lot at this point. I can't swear on the data rate until I take some more time with it and measure again (when I figure out why the data capture isn't working everytime).

Here are my ideas based on what you said:
1. Probably not variable data rate
2. I too expect a polling network
3. I wonder if the PIE (or whatever other manufacturer) has a device that can be inspected for more info-didn't PIE or someone else make a Ai-Net compatible adapter?

I had not looked for the connectors yet; that's a good idea. I would like a "break out" cable and much more. I will look into those.

I think I'm going to examine the bus more and come up with another plan.

It kind of ticks me off about the data sheet.

jics · 01-27-2006, 01:53 PM

Hi,

Quote:

3. I wonder if the PIE (or whatever other manufacturer) has a device that can be inspected for more info-didn't PIE or someone else make a Ai-Net compatible adapter?

Something like this would make a (relatively) cheap victim for dissection:
http://www.alpine-usa.com/products/l...h_kca-420i.htm

Who or what is PIE ? I haven't found anything related on google...

regards,

jics

jbors · 01-27-2006, 02:03 PM

Interesting topic. I would say that most complicated part in AiNet is not a physical layer, but rather check sum algorithm itself. That's where all other attempts are stopped

jics · 01-27-2006, 02:18 PM

HI,

Given the fact that most of the slave devices (e.g.: CD-Changer) have a very limited computing power, it won't be anything more complicated then a CRC-16 with an unusual initialisation vector.
If I have to make a guess: It is a cross-sum modulo X.

...If there is a checksum at all.

regards,
jics.

User Name		Remember Me?
Password