Re: USB as product-internal interconnect system?
From: Roger Hamlett (rogerspamignored_at_ttelmah.demon.co.uk)
Date: 01/20/05
- Next message: Keith Williams: "Re: Cambridge (UK) engineer seeking job and home in Cambridge"
- Previous message: Dirk Bruere at Neopax: "Re: Magnetic Field Q"
- In reply to: Chris Graham: "USB as product-internal interconnect system?"
- Next in thread: Joel Kolstad: "Re: USB as product-internal interconnect system?"
- Reply: Joel Kolstad: "Re: USB as product-internal interconnect system?"
- Messages sorted by: [ date ] [ thread ]
Date: Thu, 20 Jan 2005 13:50:04 GMT
"Chris Graham" <chrisgr@shaw.ca> wrote in message
news:FKGHd.128245$Xk.65070@pd7tw3no...
> I'm designing a product that includes a number of interchangable boards
> that give various capabilities, each with its own small microcontroller
> and sensors/actuators. Any modules in a system would be connected to a
> more powerful main board (basically the reference design of a PDA). It
> must be easy for the user to open the box and plug in new modules, or
> plug them into sites on the outside, and also for third parties to
> provide modules that also plug in.
>
> I was designing my own interconnect standard that included power and
> bidirectional data lines (ideally at least1mbps), maybe based on RS485,
> and a way for devices to announce their type and capabilities. As I
> worked on the design I started to realize I was reinventing USB. I've
> been looking at the details of USB, and it certainly does all I need and
> more.
>
> Also, fortunately, the central processor board is based on a PDA that
> has a USB host capability (and HID driver, which my modules could use),
> and I could use small USB enabled processors in the modules. I guess
> I'd also have to provide a hub based on one of the available hub chips.
>
> I was hoping this group could give me a sanity check before I proceed to
> design a system based on USB.
One big caveat, could be that USB, does not have a timeout in the drivers.
If a command is sent to a device, that the system 'knows' is there, and a
reply is expected, the system can hang waiting for a reply. This becomes
more likely, as noise levels and cable lengths increase. You would either
have to write your own master handler, and add such protection, or add a
hardware watchdog to all the devices (including the master). Personally,
if the system is mant to be reliable, I'd consider Ethernet. Easy to
implement, has built in error checking, better signal protection on the
lines etc. etc..
> - Is anyone aware of other products using USB in this way?
> - Any other approaches to plug-and-play or mix-and-match smart modules
> in a system?
> - To keep the costs down I would not want to use standard USB connectors
> internally - just headers (but still shielded cable). It this likely to
> cause EMI problems? I would have to use the 12mbps version of USB to
> keep latency low.
>
> Thanks for reading all the way to the end :-)
>
> - Chris Graham
Best Wishes
- Next message: Keith Williams: "Re: Cambridge (UK) engineer seeking job and home in Cambridge"
- Previous message: Dirk Bruere at Neopax: "Re: Magnetic Field Q"
- In reply to: Chris Graham: "USB as product-internal interconnect system?"
- Next in thread: Joel Kolstad: "Re: USB as product-internal interconnect system?"
- Reply: Joel Kolstad: "Re: USB as product-internal interconnect system?"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
