18F2550 / 18F4550 USB

[CDESC]

Hello

Anybody ran experiments with the MC 18F4550 or 2550 USB capable controllers
and managed to acchieve good speed? I'm getting about 1000 messages per
second in any mode (bulk, control or interrupt -- HID or not) and wondering
how to get it to speak a little faster.

Thanks

 

[Paul E. Schoen]

"CDESC" <feedback@realitymedias.com> wrote in message
news:CoMIj.21122$ny4.28025@weber.videotron.net...

Hello

Anybody ran experiments with the MC 18F4550 or 2550 USB capable
controllers and managed to acchieve good speed? I'm getting about 1000
messages per second in any mode (bulk, control or interrupt -- HID or
not) and wondering how to get it to speak a little faster.

Thanks

I made a USB-Serial converter using the PICDEM FS USB board (and actually
using the older PICDEM USB modified for full speed), and I am able to run
it at 57.6 kB, or 4800 characters per second. But I'm not sure of the
actual messages per second.

There are good discussions on this in the Microchip User's Forum

In the thread: http://forum.microchip.com/tm.aspx?m=321195 there is a
discussion of isochronous data rate, and it appears that all FS USB modes
work with 1 mSec time slices, each of which can transfer a packet of 1023
bytes, for a maximum throughput of about 1,000,000 bytes per second. But
apparently the 18F devices do not have enough RAM to support this.

Here is a useful table that was referenced on the forum. But read the
thread for more info.

http://forum.microchip.com/upfiles.aspx/29174/B786CC91A5D747178A7BD31C6BF5B134.JPG

Paul

 

[CDESC]

Hi Paul. Thanks for the re. and for pointing out the thread.

It looks from the posts I saw there that the best this little beast can do
is limited to a packet per frame (1 micro) period, even though
the protocol specifications states that control/bulk modes aren't tied to
the endpoing poll (re)cycle delay.

Though I confirm I ran tests that showed about 4500 Pk/S (of 3 bytes each)
using a control input (pc>pic) endpoint and a bulk out (pic>pc) endpoint. So
pick your flavor;
- I am violating the protocol
- My test results are wrong
- There is light, and a way to do it... in the end

Whoever ran tests using full-speed & bulk and/or control transfer modes,
your test results would be more than welcome.

Thanks.



"Paul E. Schoen" <pstech@smart.net> wrote in message
news:47f3e54f$0$17413$ecde5a14@news.coretel.net...

"CDESC" <feedback@realitymedias.com> wrote in message
news:CoMIj.21122$ny4.28025@weber.videotron.net...
Hello

Anybody ran experiments with the MC 18F4550 or 2550 USB capable
controllers and managed to acchieve good speed? I'm getting about 1000
messages per second in any mode (bulk, control or interrupt -- HID or
not) and wondering how to get it to speak a little faster.

Thanks

I made a USB-Serial converter using the PICDEM FS USB board (and actually
using the older PICDEM USB modified for full speed), and I am able to run
it at 57.6 kB, or 4800 characters per second. But I'm not sure of the
actual messages per second.

There are good discussions on this in the Microchip User's Forum

In the thread: http://forum.microchip.com/tm.aspx?m=321195 there is a
discussion of isochronous data rate, and it appears that all FS USB modes
work with 1 mSec time slices, each of which can transfer a packet of 1023
bytes, for a maximum throughput of about 1,000,000 bytes per second. But
apparently the 18F devices do not have enough RAM to support this.

Here is a useful table that was referenced on the forum. But read the
thread for more info.

http://forum.microchip.com/upfiles.aspx/29174/B786CC91A5D747178A7BD31C6BF5B134.JPG

Paul