I2C (not SMBus) SW interface for M10k?

rando · 09-13-2004, 12:07 AM

Can anyone tell me if there is a true I2C host adapter on board the M10k and, more importantly, point me to some reference information and/or code that describes it?

As part of Seth's AM/FM RDS radio project, I've been working on the basic I2C interface driver. Some of you are aware, we were planning to hook the boards up to the M10k I2C header for development/testing, before we waste any time developing a USB to I2C bridge.

Based on my weekend of research, the only public control interface for the M10K I2C port implements the SMBus protocols only. Basically all the code or discussion threads I've found point back to the Linux i2c-viapro.c driver. That driver and the underlying control interface is essentially a clone of the Intel PIIX4 SMBus. I have a port of the Linux driver in Windows running in user space. It communicates with the underlying hardware through the Numega portio.sys sample driver from the MS DDK. This code works fine and correctly issues SMBus commands on the bus without any problems.

SMBus is unfortunately only a subset of the full I2C bus. While it's possible to implement SMBus on top of the general I2C interface, the reverse is not possible. In the case of our tuner and signal processor ICs, the commands required cannot be contructed from the SMBus protocol primitives. We need native I2C (Start, Stop, Ack, Nack, Set/Clear Data, Set/Clear Clock), or at least something lower level than what SMBus offers.

antimatter · 09-13-2004, 12:26 AM

Crapola.... I knew there was some funky business with the via's implimentation of the I2C.

Good luck in finding what you need Rando

rando · 09-13-2004, 12:36 AM

Well don't lose sleep over it or anything dude.

We can always bitbang over the parallel port for initial testing. It doesn't look impossible to implement though performance will surely suck -- not that performance will be an issue in this case. The I/O interface sits idle except when you issue a command to change something. Even then, it's only a handful of bytes.

frodobaggins · 09-13-2004, 01:37 AM

http://www.devasys.com/usbi2cio.htm

You can pick one of these up to test with.

frodobaggins · 09-13-2004, 01:46 AM

Or rs232 to i2c
http://www.emicros.com/i2c232.htm

frodobaggins · 09-13-2004, 01:49 AM

http://www.ftdichip.com/FTProject.htm

GruvThang · 09-13-2004, 01:51 AM

Quote: Originally Posted by frodobaggins

http://www.devasys.com/usbi2cio.htm

You can pick one of these up to test with.

That looks nice, kinda pricey though!

frodobaggins · 09-13-2004, 01:53 AM

Quote: Originally Posted by GruvThang

That looks nice, kinda pricey though!

I'm not suggesting things for production, just so he can have something
to test his chips out with.

GruvThang · 09-13-2004, 02:16 AM

I am not thinking that it would be for production, it is just pricey for what the unit actually does, the rs232 is not any cheaper, they do offer "ease" of testing though!

rando · 09-13-2004, 07:51 AM

OK, maybe I wasn't clear in the original post. I'm not looking for an alternate interface -- I'm aware that there are many options. The question is the M10K. Does it or doesn't it support native I2C over its (supposed) I2C header?

BeamRider · 09-13-2004, 08:18 AM

I was interested in playing with EPIA I2C sometime ago. Looking for windows support (MS I2C APIs are used for video DDC only) I've found this package but I haven't tested it. Hope this helps.

Seth · 09-13-2004, 08:43 AM

Thanks BeamRider but rando has already looked into that. It's not exactly what we are looking for. We tend to belive that VIA is misleading when using the name "I2C" for that port and that the underlying controller (VT1211) supports only the SMBus command subset.
We're investigating though and we could use all the help we can get. That means that if someone has an "inside" VIA connection we could use his/hers help.

rando · 09-13-2004, 10:09 AM

@Beamrider: That code is essentially a port of the i2c-viapro driver from the Linux LM-Sensors project (now integrated with most Linux Kernel distros). It implements an SMBus client library on top of the SMBus host controller inside the M10k. It's basically the code I was referring to in the original post #1.

What we're looking for is some code or documents that describe the I2C host controller -- if it exists at all.

@Seth: If I understand things correctly, the VT1211 is not the underlying I2C host adapter. Rather it's an I2C sensor client. At least in the SMBus case, the Southbridge implements the host adapter. I guess it's possible the VT1211 also implements a I2C host adapter but perusing the Linux drivers for that chip doesn't suggest that's the case.

frodobaggins · 09-13-2004, 10:23 AM

Quote: Originally Posted by rando

OK, maybe I wasn't clear in the original post. I'm not looking for an alternate interface -- I'm aware that there are many options. The question is the M10K. Does it or doesn't it support native I2C over its (supposed) I2C header?

I realize what you are doing. But I believe from the literature that I've read
that you will not be able to use the i2c port on the M10K. Thus to test your
chips you will need an alternate interface to work with during initial test.
I was trying to help you find the most efficient route for doing so.

Frodo

rando · 09-13-2004, 10:56 AM

Quote: Originally Posted by frodobaggins

.. I believe from the literature that I've read
that you will not be able to use the i2c port on the M10K.

From what I've seen so far, I agree. But I cannot be certain without additional information. Can you provide a reference to this literature? Thanks for the help ... post whore.