[usb-midi-fw-devel] Re: More tascam

[Martin L. <mar...@gm...>]

On Sun, Aug 24, 2003 at 08:39:56PM +0200, Pedro Lopez-Cabanillas wrote:
> El Domingo, 24 de Agosto de 2003 15:18, Martin Langer escribió:
> > On Sun, Aug 24, 2003 at 01:24:47PM +0200, Karsten Wiese wrote:
> > > it's an XCS20XL.
> >
> > According to http://www.xilinx.com/bvdocs/publications/ds060.pdf the PROM
> > size of the XCS20XL is 179160 Bits (page 33). That's exactly the size of
> > our file (without pre padding bits).
> >
> > But our header information looks different to the described Spartan (XL)
> > header in the documents (xapp098.pdf, page 9). That's bad.
> 
> Because for Spartan-II, i'm using LSB-first translation into ascii bits. So, 
> the hex value 0xaa995566 is translated to ascii bits:
> 0101 0101 1001 1001 1010 1010 0110 0110 
> I've done that because the examples for spartan/virtex rbt files all have this 
> format.
> 
> But if you look to the hex representation in us428fw8001.h
> char Daud0006[ 4096] =
> { 0xFF, 0x20, 0x2B, 0xBD, 0x1F, 0x5B, 0xFF, 0xFE, 0xFF, 0xF7, 0xB5, 0xFB, 
> 0xFE, 0x5F, 0xD7, 0xFB,
> It fits on the layout for Spartan/XL data stream format in serial mode, 
> ds060.pdf, page 32, table 16.
> 
> Data type	Serial Mode
> Fill byte	11111111b	
> Preamble	0010b		
> Length count	0x02bbd1	
> Fill bits	1111b		
> 
> The length count (179153 dec) does not match. But if you look to the example 
> in xapp098.pdf, page 8, it says Bits: 247968, and the length count field is 
> 0x3C899 (247961 dec), again 7 bits shorter.
> 
> We need to investigate deeper on this.
> 

ds060.pdf, page 32, at the end of the first paragraph:
[...]
The 11-bit CRC check of the last frame of an FPGA includes the last seven
data bits.
[...]

Let's assume that the last seven data bits there have no CRC checking
And if the CRC checking depends on the length value in the header it's
necessary to make it 7 bits shorter. That could be a solution on
Spartan/Spartan-XL, or?


bye,
martin