Configuration Spartan 3

Hi
We work with a XCF01S and a Spartan 3 XC3S200.
JTAG works with 2.5V, Spartan Vcco (INIT_B and DIN/D0) has 3.3V. The
connections between XCF and Spartan are according the application note
for 3.3V output (INIT_B) to 2.5V input CE/RES.

Programming the Flash and booting from Flash works fine.
Programming the FPGA works only, when we have the same code in the
Flash.
When we try to load a different code (as is in the Flash) per JTAG to
the FPGA, the FPGA doesn't start up.

Does anybody know what's going wrong?

Luigi

"luigi" <luigi.zadra@bluewin.ch> schrieb im Newsbeitrag 
news:1138031781.352244.275760@g14g2000cwa.googlegroups.com...
> Hi
> We work with a XCF01S and a Spartan 3 XC3S200.
> JTAG works with 2.5V, Spartan Vcco (INIT_B and DIN/D0) has 3.3V. The
> connections between XCF and Spartan are according the application note
> for 3.3V output (INIT_B) to 2.5V input CE/RES.
>
> Programming the Flash and booting from Flash works fine.
> Programming the FPGA works only, when we have the same code in the
> Flash.
> When we try to load a different code (as is in the Flash) per JTAG to
> the FPGA, the FPGA doesn't start up.
>
> Does anybody know what's going wrong?
>
> Luigi
>
yes. I know ==>  I am anybody am I? :)

there is xilinx AR on this, their suggested solution is to change FPGA MODE 
pin setting.

similar or even worse happens with S3e and Parallel flash, the configuration 
from Parallel flash
(if located at close to the start address) prohibits the FPGA from being 
configured over JTAG
and as Flash programmer over FPGA JTAG indirect so flash programming also 
fails.

we have special workaround for this that puts the parallel flash into read 
id mode and clears
FPGA config and then reconfigures FPGA with Flash programming IP and then 
programs flash.

for Platform flash, you must erase platform flash and force FPGA 
reconfiguration then it will
be possible to reconfigure over JTAG again




-- 
Antti Lukats
http://www.xilant.com

luigi wrote:
> Hi
> We work with a XCF01S and a Spartan 3 XC3S200.
> JTAG works with 2.5V, Spartan Vcco (INIT_B and DIN/D0) has 3.3V. The
> connections between XCF and Spartan are according the application note
> for 3.3V output (INIT_B) to 2.5V input CE/RES.
> 
> Programming the Flash and booting from Flash works fine.
> Programming the FPGA works only, when we have the same code in the
> Flash.
> When we try to load a different code (as is in the Flash) per JTAG to
> the FPGA, the FPGA doesn't start up.
> 
> Does anybody know what's going wrong?
> 
> Luigi
> 
Change the mode pins on the FPGA
Aurash

"Aurelian Lazarut" <aurash@xilinx.com> schrieb im Newsbeitrag 
news:dr2v1k$in82@cliff.xsj.xilinx.com...
> luigi wrote:
>> Hi
>> We work with a XCF01S and a Spartan 3 XC3S200.
>> JTAG works with 2.5V, Spartan Vcco (INIT_B and DIN/D0) has 3.3V. The
>> connections between XCF and Spartan are according the application note
>> for 3.3V output (INIT_B) to 2.5V input CE/RES.
>>
>> Programming the Flash and booting from Flash works fine.
>> Programming the FPGA works only, when we have the same code in the
>> Flash.
>> When we try to load a different code (as is in the Flash) per JTAG to
>> the FPGA, the FPGA doesn't start up.
>>
>> Does anybody know what's going wrong?
>>
>> Luigi
>>
> Change the mode pins on the FPGA
> Aurash

typish Xilinx !

for many many PCB boards where mode pins are hard wired under BGA this is 
not an option.

In all Xilinx documents its clearly written that JTAG configuration 
over-rides any other modes.
Still there are several scenarios where this is not entirely true, eg the 
restart of non JTAG
configurtion if valid signature is present at some time slot will make the 
JTAG configuration
to fail.

I have had that fight many many many times.

Antti

Thanks Antti

This was a very helpful hint.
Since I know that this is a Xilinx fault, I can live with. Important is
to know, that we don't have a hardware error.
We need "master serial mode" and want change constantly the resistors,
so we erase the flash.

Luigi

"luigi" <luigi.zadra@bluewin.ch> schrieb im Newsbeitrag 
news:1138050174.499175.245290@f14g2000cwb.googlegroups.com...
> Thanks Antti
>
> This was a very helpful hint.
> Since I know that this is a Xilinx fault, I can live with. Important is
> to know, that we don't have a hardware error.
> We need "master serial mode" and want change constantly the resistors,
> so we erase the flash.
>
> Luigi
>
there is special Answer Record that says the 'change mode' if having 
problems, but I have too few BRAMs overleft to remember the AR numbers.

if in doubt always clean sweep the config memories.

Antti