I2C as slave reading multiple addresses

russk2txb · Joined: 30 Mar 2008 Posts: 109 Location: New Jersey

Hi. I have a project where we want to replace several Intersil X0241A digital potentiometers with a PIC. There is an existing controller that sends I2C commands to these ePots. We want to read the output values from this controller, with the PIC, and then do our own processing and control. The controller expects the ePots to be in slave mode. I thought at first that I could simply read the I2C address and then the associated data byte, and have the pic perform the desired function. However I see that you have to assign an address to the PIC I2C in slave mode. This would seem to preclude reading more than one output address from the existing controller.

Is there any way to handle this? Is it possible to read the I2C signals from multiple addresses, along with the actual address? (I guess this could be done by bit banging, but I'd rather use the hardware I2C if possible.) I also want to have an interrupt driven input routine so I don't have to tie up my main program loop with handling the input details.

Can anyone give me some suggestions or point me to an article or example?

Thanks, Russ

rnielsen · Joined: 23 Sep 2003 Posts: 852 Location: Utah

I'm not sure if I understand what you're trying to accomplish here. Maybe it's because it's Monday or that I had to get up this morning or that I'm at work or just BECAUSE. (Just trying to be funny).

Are you trying to have one PIC simulate all of the pots and have the master talk to it? Or do you want to replace the existing master with one of your own? Would you be able to expand your description on what you're trying to do? The better the description, the better we can understand and give suggestions.

Ronald
PROCRASINATORS UNITE!........... tomorrow

PCM programmer · Joined: 06 Sep 2003 Posts: 21708

russk2txb · Joined: 30 Mar 2008 Posts: 109 Location: New Jersey

russk2txb · Joined: 30 Mar 2008 Posts: 109 Location: New Jersey

Well, by the silence I'm getting on this topic today, I'm guessing there is no solution for my problem as stated. So I am going to try another tack. I determined that I really only need to emulate two ePots. That means I only need to act as an I2C slave for two addresses. Since the 18F6723 has two SSP channels, I can do it. However I was already using one channel for SPI control of several devices and so I have to switch this to software SPI using different pins.

My question is about baudrate in software SPI. How can I determine what maximum rate I can use? The help for CCS compiler does not seem to address that. Also I was reading that the USE_SPI command does not work very well when it comes to switching parameters. I need to use different modes for some devices, but I am concerned as to whether I can switch modes 'on the fly' when using software SPI. I am controlling a couple of digital synthesizers, some DACs, and an external EEPROM on the SPI bus.

Is this software SPI a reasonable solution, or should I look at putting a second PIC on the board?

Thanks, Russ

PCM programmer · Joined: 06 Sep 2003 Posts: 21708

I assume you mean SPI Master mode, done in software.

It's done by bit-banging. Therefore, you control the entire process
and can configure any mode. I wouldn't trust the #use spi() library
until you have tested each mode and looked at the signals.
There were issues with the default settings. Maybe CCS has corrected
these by now, but I don't know.

To find the maximum baudrate, make a test program with a loop
that sends out one byte, repeatedly. Then look at the signals on your
scope or logic analyzer.

russk2txb · Joined: 30 Mar 2008 Posts: 109 Location: New Jersey

Thanks PCM Programmer. I was hoping I could get some indication of speed. The only thing I have found by browsing this forum is that software SPI is "much slower" than hardware SPI. Since it will be about 2 weeks before I have hardware in my hands, I was trying to determine if this is a viable way to go, or if I should go with a second PIC on the board.

There will be a number of tables stored in the external EEPROM and values need to be read from them during operation. I really do not expect the read speed to be so slow that it dramatically affects performance, but I am also concerned about getting hung in the read loop long enough to affect other things - like simultaenously reading 4 channels of A/D and updating DACs according to changes.

So, I wish I had some qualitative information to make decisions with.

Thanks again, Russ