prop and multiple I2C devices best practice?

mctrivia · 2009-10-09 14:44

I need to connect 7 I2C devices to a prop. What is better PCB design practice. Long chain with no branches or a short bus with branches to the i2c devices?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
propmod_us and propmod_1x1 are in stock. Only $30. PCB available for $5

Want to make projects and have Gadget Gangster sell them for you? propmod-us_ps_sd and propmod-1x1 are now available for use in your Gadget Gangster Projects.

Need to upload large images or movies for use in the forum. you can do so at uploader.propmodule.com for free.

Mike Green · 2009-10-09 15:42

There's no particular reason to use a long chain. I'd use whatever signal distribution pattern fits in with the rest of the design. I2C is not that fast (usually max. 400KHz), so you'd need to have a pretty big PCB for the signal path lengths to make much of a difference.

mctrivia · 2009-10-09 15:45

thanks didn't think the 400Khz would be much of a problem but I would thought I would ask first since both layout methods are possible just branching being easier then a chain.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
propmod_us and propmod_1x1 are in stock. Only $30. PCB available for $5

Want to make projects and have Gadget Gangster sell them for you? propmod-us_ps_sd and propmod-1x1 are now available for use in your Gadget Gangster Projects.

Need to upload large images or movies for use in the forum. you can do so at uploader.propmodule.com for free.

BigFoot · 2009-10-09 15:50

I like to use branches but if you go with the long chain it is a good idea to put the pull up resistors at the end of the chain. Also if you plan on running your bus at the faster speeds use 2.2 K pull up resistors to keep the signals nice and square.

Russ

jazzed · 2009-10-09 16:48

I'm sure Beau has input on this ....

Still, here is my experience: I worked on a project years ago where there were 5 end points on a daisy-chain topology about 10 inches long (I was not the hardware engineer). Like the Propeller outputs, the transition time of the signal was very fast (which is more important AFAIK than the aggregate frequency of the transitions). We had big problems with reflections and tried to save the boards with end of line terminations. Eventually we had to re-spin the board to use a series terminated star topology.

Does the Phillips I2C spec offer any recommendations?

mctrivia · 2009-10-09 16:53

that is strange. daisy chain topology usualy reduces reflections but increases capacitance

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
propmod_us and propmod_1x1 are in stock. Only $30. PCB available for $5

Want to make projects and have Gadget Gangster sell them for you? propmod-us_ps_sd and propmod-1x1 are now available for use in your Gadget Gangster Projects.

Need to upload large images or movies for use in the forum. you can do so at uploader.propmodule.com for free.

jazzed · 2009-10-09 16:59

mctrivia said...
that is strange. daisy chain topology usualy reduces reflections but increases capacitance

Do what you know Matt [noparse]:)[/noparse]

In the case I described each point on the chain had different reflection characteristics.
The "S" shape of the transitions was the killer ... it was on the main synchronous interface clock.

mctrivia · 2009-10-09 17:03

ah well that could cause problems.

I have chosen a bus topology for this board. makes routing easier and i am not planning to push the limits of the I2C bus so I don't think it matters much anyways.

thanks for all the information everyone.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
propmod_us and propmod_1x1 are in stock. Only $30. PCB available for $5

Want to make projects and have Gadget Gangster sell them for you? propmod-us_ps_sd and propmod-1x1 are now available for use in your Gadget Gangster Projects.

Need to upload large images or movies for use in the forum. you can do so at uploader.propmodule.com for free.

prop and multiple I2C devices best practice?

Comments