Debugging Spartan3 slave serial configuration

Do you have a question? Post it now! No Registration Necessary

Translate This Thread From English to

Threaded View
I am not yet done with my debugging yet, but thought would find out if
anyone out has experimented with slave serial configuraiton of
Spartan3.

Here is the setup on my PCB board.

The HSWAP_EN is tied to ground. The mode pins are floating.
An Atmega128 running at 5V has mapped to it the configuration signal
I/O of the FPGA. The PROG_B, INIT, CCLK, DIN. A series resistor(300ohm)
is connected between the DIN (FPGA 3.3V)and Atmega128(5V).

I can configure the FPGA from a JTAG port without changing the mode
pins.

I configure(via JTAG) the FPGA to run a continuous loop to blink LEDs.
Then my Atmega drives the PROG_B signal low, which resets the FPGA. The
INIT line is then checked and for some reason(using a Oscope) I have
not been able to see the INIT line go low at all.

After checking the INIT line, the atmega drives the CCLK low.
The atmega128 reads one byte from the flash and then unpacks each byte
and toggles the DIN signal according to the byte.

The sequence for the 8 bits  is as follows:

CCLK_LO()
toggle DIN(LSB)
CCLK_HI()
toggleDIN(LSB)

CCLK_LO()
toggle DIN(LSB+1)
CCLK_HI()
toggle_DIN(LSB+1)

and so on.

I have counter which keeps track of the bytes,which is 212393 bytes for
a Spartan3XC3S400. After counter expires, I toggle the CCLK a few more
time (as said in earlier discussions on this fourum).


I am not sure why the DONE bit is not going high?????

Any ideas?


Re: Debugging Spartan3 slave serial configuration

Quoted text here. Click to load it

1) you can use impact to query the configuraton status after your config
attempt, as example you can see if there was CRC error, etc. but my bet it
only say init ok, done = 0

2) make sure you have startupclock set CCLK

3) if in doubt reverse bit order in bytes

4) send some more dummy bits after end

I am using mega8 to load Spartan3 and virtex4 over JTAG from SD-Card and I
have no problems. slave serial is even simpler to implement, just send the
.BIT file into DIN and that should do it

note that INIT will not flag error if the proper signature is never seen

Antti



Re: Debugging Spartan3 slave serial configuration
Antii,

Thanks for you reply.

I tried all your suggested options. I am having even more bizzarre
issues. I modified my process as

Verified that the startup clock is set to CCLK (as a part of
Implementation option of the XST synthesis tool).

1. assert FPGA PROG# pin low.
2. De-assert PROG# high.
3. wait for INIT# to go high.
4. Unpack the byte (into bits) from configuration file (BIN file) and
place first bit (LSB) of bitstream on DIN.
5. pulse CCLK high then low.
6. repeat 4, 5 for remaining bits of bitstream.  While doing this
monitor the INIT# line.  If it goes low, the FPGA has detected
a bitstream error.
7. place 1 on DIN.
8. pulse CCLK high then low. (for like 10 dummy clock cycles).


The DONE goes high after like 30 seconds (after configuration has
ended) or something, but my actual code (LED blinking) never happens.

And then after another 30seconds later the DONE goes low!!!!!

So I connected the JTAG pins and used IMPACT to read the status
register. These pins were different from succesful JTAG configuration.
status of GTS_CFG_B                               :         0
status of GWE                                     :         0
status of GHIGH                                   :         0

And the INIT_B was high and DONE was low.


Any ideas!!!!!!!!

-Yaju


Re: Debugging Spartan3 slave serial configuration
Never mind. I fixed it.

I was sending the bits LSB first. That is incorrect.

The spartan-3 requires the bits MSB first. I  also had a few problems
with my data transfer, since I have not implemented any sort of flow
control. My system basically implements the crudest of flow control,
i.e. write and read back the byte. This is insanely slow and I will be
implementing other faster flow contol mechanism, now that my serial
configuration system works.

Thanks Antti.

Yaju Nagaonkar
---
Electrical and Computer Engineering
Brigham Young University
We've slightly trimmed the long signature. Click to see the full one.
Re: Debugging Spartan3 slave serial configuration
Quoted text here. Click to load it
:)

you did not follow my instructions full, number [3] of the advices was to
use different bit order in bytes 'just in case'

its a generic trouble solver if one setting that can influence the result
can have two values, then check out both - even if you thing your current
setting is correct.

Antti



Re: Debugging Spartan3 slave serial configuration
I actually did follow your instructions.  I was having problems with
the flow control which I had to fix before I could get back to your
instructions.
Thanks eitherways.


Site Timeline