|
Message
From: "kw@nie" <kwanie@p...>
Date: Thu, 21 Feb 2002 10:33:45 +0800
Subject: Re: [usb] crc algorithm
Hi Bo,
The 1st method take for x^5 xor x^2 xor x^0 and feedback into LSB. There is
only 1 feedback point for this method, which is the LSB. That's why it needs
5 additional cycle after the last data bit to work, which is also why it is
never used in real design.
cheers,
kwanie
----- Original Message -----
From: "Bo" <bo_ye@y...>
To: <usb@o...>
Sent: Thursday, February 21, 2002 10:10 AM
Subject: Re: [usb] crc algorithm
> Hi Kwnie,
> Actually, it doesn't mater what value
> you need to preset, as far as encode/decode
> are using the same seed. But people set it to
> 1's for some protocals, so everybody can
> communicate each other, either in transmit side
> or receive side. That's my understanding.
> The result for the first diagram looks to me
> is y = 1 + x^3 + x^5, but the second one is
> y = 1 + x^2 + x^5. If you change the first one,
> feedback from Msb -> Lsb to Msb -> B[2], actually
> from 2-string feed back to single-string feedback,
> it would has the same result as the second one.
> Which is y = 1 + x^2 + x^5.
>
> regards
> bo
>
>
> --- "kw@nie" <kwanie@p...> wrote:
> > Bi Bo,
> >
> > You are right about pre-loading the register with
> > '1's. it's careless of me
> > to not have included it.
> >
> > > The first part, looks like using two-string
> > > feed back instead. It will result in
> > > Y = 1 + X^3 + X^5 instead of Y = 1 + X^2 + X^5 in
> > > USB protocol for token packet checksum.
> >
> > The result is still Y = 1 + X^3 + X^5 => 2^0 + 2^3 +
> > 2^5, with 2^0 being the
> > data bit.
> >
> >
> > ----- Original Message -----
> > From: "Bo" <bo_ye@y...>
> > To: <usb@o...>
> > Sent: Thursday, February 21, 2002 7:33 AM
> > Subject: RE: [usb] crc algorithm
> >
> >
> > > Hi Kwanie,
> > > Thanks for the diagram.
> > > I agree on the second part. Xor the X0 with
> > > the MSB will yield the correct when the last bit
> > > is shifted in. But need to preset all the register
> > > to 1's. (correct me if I am wrong)
> > > The first part, looks like using two-string
> > > feed back instead. It will result in
> > > Y = 1 + X^3 + X^5 instead of Y = 1 + X^2 + X^5 in
> > > USB protocol for token packet checksum.
> > >
> > > regards
> > > bo
> > >
> > >
> > > --- "kw@nie" <kwanie@p...> wrote:
> > > > I think I know why the binary representations,
> > in
> > > > the USB specification, are
> > > > different from the polynomials. Attached is a
> > brief
> > > > explanation based on the
> > > > 5-bit CRC but the principle applies for both CRC
> > > > generators.
> > > >
> > > > Cheers,
> > > > kwanie
> > > >
> > > >
> > > >
> > > >
> > > > -----Original Message-----
> > > > From: owner-usb@o...
> > > > [mailto:owner-usb@o...] On Behalf
> > > > Of kebloo@m...
> > > > Sent: 21 February 2002 08:34
> > > > To: usb@o...
> > > > Subject: [usb] crc algorithm
> > > >
> > > > Hello,
> > > >
> > > > Sorry for this beginner question, but I wish
> > some
> > > > one could
> > > > answer my curiosity...
> > > >
> > > > USB 1.1 specification defines generator
> > polynomial
> > > > of 5-bit crc
> > > > as G(x) = x^5+x^2+1 and binary pattern that
> > > > represent it as
> > > > 00101B and 16-bit crc as G(x) = x^16+x^15+x^2+1
> > and
> > > > binary pattern that represent it as
> > > > 1000000000000101B.
> > > > My question is why the binary pattern stated
> > like
> > > > that and
> > > > not 100101B for 5-bit crc and 11000000000000101B
> > ?
> > > > Thanks....
> > > >
> > > >
> > > >
> > > > Best Regard,
> > > >
> > > >
> > > > RE Tambunan
> > > > --
> > > > To unsubscribe from usb mailing list please
> > visit
> > > > http://www.opencores.org/mailinglists.shtml
> > > >
> > >
> > > > ATTACHMENT part 2 image/jpeg name=crc5.JPG
> > >
> > >
> >
> >
>
>
>
|
 |
Browse
