Our interest in the M289xx chipset

As mentioned on a few other pages of this site, we are interested in doing some experiments with the M289xx kitchen sink chipset for G.shdsl, SDSL/2B1Q and IDSL. To reiterate, the reasons for our interest are:

• We need some connectivity solution for the new G.shdsl standard.
• It appears that some of the competing SHDSL chipsets don't provide a UTOPIA port. From our perspective of achieving connectivity to existing S(H)DSL lines over which we don't have any control at the CO end, we need to support the dominant ATM flavors, and thus UTOPIA support is very important for us.
• Since the kitchen sink appears to have turnkey support for IDSL, perhaps we can avoid building Yet Another hardware platform for IDSL and instead kill two birds with one stone if we choose M289xx for our SHDSL solution.
• If we ever decide to build a low cost, end user-oriented SDSL CPE product, having one piece of hardware support all 3 of G.shdsl, SDSL/2B1Q and IDSL is very attractive. (Please read though about the paradox of modularity and openness.)
• We could perform some interesting experiments if we had an M289xx under our control running against an OSDCU.

LNX8945 platform

The ideal platform for us to evaluate and play with the M28945 chip would be one in which this chip is connected to a Freescale (former Motorola) MPC8xx processor running Linux. It would need to be an MPC8xx model that has UTOPIA so we can exercise the ATM block of the M28945, and there would need to be one full-featured SCC available to which we would connect the PCM interface. MPC850SR fits the bill perfectly. The host port (mailbox) would be connected to the peripheral bus.

We have codenamed this platform LNX8945. Unfortunately however, this hardware platform does not currently exist except as an idea in the mind of the hacker writing these web pages. There are several different ways in which the LNX8945 platform could be brought into physical existence, each with its own problems:

• We can design and build an LNX8945 board ourselves from scratch. Problem: an investment of blood, sweat and tears with somewhat uncertain return.

• Perhaps we can reverse-engineer and hack an existing board which seems to fit the bill such as Siemens 5890. Problem: reverse engineering is also major pain.

• There is Mindspeed's own evaluation platform for M289xx, and it's supposed to come with schematics and PCB layout files. (The latter are probably the most valuable part of the whole package.) However, obtaining the article may be very problematic.

  First off, we would need to go through a secret proxy agent in order to get anything at all out of Mindspeed as they hate our guts. But then unless someone can social-engineer them to provide just the schematics and layout files without any physical hardware, buying their complete evaluation platform package described in their document 28945-EVMD-001-B would probably be absolutely cost-prohibitive, especially considering that only a very small part of all that monstrosity would be actually useful to us, and the rest would be money thrown away. I don't know how much they would want for the whole thing, but read their document and think for yourself how much that beastie would have to cost — it hurts to think.

Project priority

LNX8945 is not currently a high priority project for our group. Instead our primary current emphasis is on a 100% open source solution consisting of an RS8973-based OSDCU, a piggy-backed router module for it and perhaps an MC145572-based IDSL DSU.

However, we are aware of some people in other countries who are watching our project from the sidelines and who are more interested in SHDSL from the perspective of new deployment rather than in connecting to existing weird flavors of SDSL/2B1Q like we are. If you, the reader of this web page, are in that category, are interested in the M289xx chipset and/or in our LNX8945 evaluation platform for it, and would like to see it come to fruition a little sooner, we have the following offer for you:

If you are interested, you can strike a deal with us as follows: you will need to do the grunt work of drawing up the design of the LNX8945 board, i.e., draw the complete schematics and do the PCB layout (the latter is the absolutely critical part), and if you design the board in accord with our guidelines as far as the choice of processor and the basic block-level architectural features (contact us for the details), we can in return agree to cover the cost of actually manufacturing the first few prototypes (we can have the PCBs fabbed from your gerbers, we'll buy the parts (within reason), and we'll do the assembly), and then we'll do the board bringup to make it work, write the boot ROM monitor, port Linux to the board and get it to run, and write the software enabling one to play with the M28945 chip from the on-board Linux processor. The processor would need to be Freescale MPC850SR or some other agreed-upon member of the MPC8xx family.