khex <khexeme1@n_o_s_p_a_m.binghamton.edu> wrote:
> Hello all,

> We are trying to decide whether or not to purchase a cheap USB host
> controller to handle a single peripheral device or setup a Spartan 3E 500
> gate FPGA to act as the host. 

> .. we are relatively sure that trying to implement Host code on a spartan
> 3E is a terrible idea... but cannot actually confirm why other than common
> sense.

> Could you guys help us out with some reasons why the dedicated host chip or
> FPGA solution might be better? (code or physical layout might be too
> difficult?)

Milkymist has a USB Host softcore in XC6SLX
-- 
Uwe Bonnes                bon@elektron.ikp.physik.tu-darmstadt.de

Institut fuer Kernphysik  Schlossgartenstrasse 9  64289 Darmstadt
--------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------

2010-11-16 04:14, khex skrev:
> Hello all,
>
> We are trying to decide whether or not to purchase a cheap USB host
> controller to handle a single peripheral device or setup a Spartan 3E 500
> gate FPGA to act as the host.
>
Why not get a Microcontroller with USB host?
You can either use it as the main processor,
or as an auxiliary processor which will handle
the USB protocol fully.
That is what the "cheap hosts" are doing,
but this gives you more flexibility.


> .. we are relatively sure that trying to implement Host code on a spartan
> 3E is a terrible idea... but cannot actually confirm why other than common
> sense.
>
> Could you guys help us out with some reasons why the dedicated host chip or
> FPGA solution might be better? (code or physical layout might be too
> difficult?)
>
> Thanks in advance
>
> khex
>
> 	
> 					
> ---------------------------------------		
> Posted through http://www.EmbeddedRelated.com

Best Regards
Ulf Samuelsson

On 16 Nov., 15:54, Andy <amd...@gmail.com> wrote:
> On Nov 15, 9:14=A0pm, "khex" <khexeme1@n_o_s_p_a_m.binghamton.edu>
> wrote:
>
> > Hello all,
>
> > We are trying to decide whether or not to purchase a cheap USB host
> > controller to handle a single peripheral device or setup a Spartan 3E 5=
00
> > gate FPGA to act as the host.
>
> > .. we are relatively sure that trying to implement Host code on a spart=
an
> > 3E is a terrible idea... but cannot actually confirm why other than com=
mon
> > sense.
>
> > Could you guys help us out with some reasons why the dedicated host chi=
p or
> > FPGA solution might be better? (code or physical layout might be too
> > difficult?)
>
> USB host is mostly about software, the hardware isn't too much
> different, but there is a lot of s/w involved, especially once you
> start pulling in drivers for things like the HID class. =A0You might be
> better served using something like a cpu module that gives you a whole
> cpu-memory-usb complex and has the code done, then implementing your
> special stuff in the FPGA.

I'm sure if you only have to support one specific device you could
make a lot of short cuts with the fpga implementation, with the
external host chip that might not be so easy, but it will most
definitely take more time to get it to work

-Lasse

On Nov 15, 9:14=A0pm, "khex" <khexeme1@n_o_s_p_a_m.binghamton.edu>
wrote:
> Hello all,
>
> We are trying to decide whether or not to purchase a cheap USB host
> controller to handle a single peripheral device or setup a Spartan 3E 500
> gate FPGA to act as the host.
>
> .. we are relatively sure that trying to implement Host code on a spartan
> 3E is a terrible idea... but cannot actually confirm why other than commo=
n
> sense.
>
> Could you guys help us out with some reasons why the dedicated host chip =
or
> FPGA solution might be better? (code or physical layout might be too
> difficult?)
>

USB host is mostly about software, the hardware isn't too much
different, but there is a lot of s/w involved, especially once you
start pulling in drivers for things like the HID class.  You might be
better served using something like a cpu module that gives you a whole
cpu-memory-usb complex and has the code done, then implementing your
special stuff in the FPGA.

I'd say go for an external controller chip, unless you are in for the
educational experience of designing one.

Writing the code (driver + stack) to talk to a device through an
external chip already amounts for quite a challenge.  Also, I'd
consider it a helpful prerequisite before rolling your own fully
custom controller.

On Nov 16, 4:14=A0pm, "khex" <khexeme1@n_o_s_p_a_m.binghamton.edu>
wrote:
> Hello all,
>
> We are trying to decide whether or not to purchase a cheap USB host
> controller to handle a single peripheral device or setup a Spartan 3E 500
> gate FPGA to act as the host.
>
> .. we are relatively sure that trying to implement Host code on a spartan
> 3E is a terrible idea... but cannot actually confirm why other than commo=
n
> sense.
>
> Could you guys help us out with some reasons why the dedicated host chip =
or
> FPGA solution might be better? (code or physical layout might be too
> difficult?)
>
> Thanks in advance
>
> khex
>
> --------------------------------------- =A0 =A0 =A0 =A0
> Posted throughhttp://www.EmbeddedRelated.com

Isn't this answered by the cost of the FPGA resource ?

If half the device is sitting there unused, and the project  worries
more about final
cost than engineering time, then a FPGA USB is a good fit.

Conversely, if the FPGA is ~full, and adding USB bumps you up to be
next device, then it is expensive. Likewise if your project worries
more about completion time, than parts costs.

-jg

Hello all,

We are trying to decide whether or not to purchase a cheap USB host
controller to handle a single peripheral device or setup a Spartan 3E 500
gate FPGA to act as the host. 

.. we are relatively sure that trying to implement Host code on a spartan
3E is a terrible idea... but cannot actually confirm why other than common
sense.

Could you guys help us out with some reasons why the dedicated host chip or
FPGA solution might be better? (code or physical layout might be too
difficult?)

Thanks in advance

khex

	   
					
---------------------------------------		
Posted through http://www.EmbeddedRelated.com