| ??? 12/27/05 16:22 Read: times |
#106018 - Hot-swap I2C Responding to: ???'s previous message |
I was involved in a project with hot-swap I2C some time ago. There were 9 removable cartridges that had I2C and SCSI on them. If memory serves, the I2C address for the carts was set by the location they were inserted in (like Apple II slots I think). That way there were only 9 addresses to poll. What made the problem worse was that the 9 carts had Atmel 2051's so the I2C slave was bit-banged. This made for a lot of fun because the carts would sometimes miss a poll if they were physically present! I didn't write the I2C slave code, but the guy that did did a pretty good job -- it'a hard to write a bit-bang I2C slave (especially since the hardware designers gave him no hardware assist). Anyway, since the addresses were known ahead of time, it was pretty easy to do. PLUS there was a process for the user to attach and detach. They didn't always do it, but it was there.
Today you have the luxury of hardware that allows for the hot-plug part (tha hardware part anyway): http://www.semiconductors.philips.com/pip/PCA95...12ADP.html "PCA9512A; Level shifting hot swappable I²C-bus and SMBus bus buffer" |
| Topic | Author | Date |
| Plug & play by I2C ,is it possibe? | 01/01/70 00:00 | |
| You have two problems? | 01/01/70 00:00 | |
| I know | 01/01/70 00:00 | |
| with a "master" with IIC hardware it sho | 01/01/70 00:00 | |
| lot of time not needed ! | 01/01/70 00:00 | |
| limited feasibility of I2C PnP | 01/01/70 00:00 | |
| the fact is | 01/01/70 00:00 | |
| Re: the facts is | 01/01/70 00:00 | |
| CAN | 01/01/70 00:00 | |
| unkowm state | 01/01/70 00:00 | |
| I must try | 01/01/70 00:00 | |
| Inappropriate Design Choice? | 01/01/70 00:00 | |
| Hot-swap I2C | 01/01/70 00:00 | |
| a fvew points | 01/01/70 00:00 | |
| ACCESS.bus is PnP I2C | 01/01/70 00:00 | |
just recalled | 01/01/70 00:00 |