Re: [flashrom] [coreboot] SPI in-system programming

[Carl-Daniel Hailfinger]

Hi Corey,

[adding [e-mail] to CC]

On 18.07.2010 11:54, Corey Osgood wrote:
> I need a way that I can test images without actually being in the same
> location as the board I'll be testing them on. What would you guys
> recommend for programming an SPI chip with it still on the
> motherboard?

flashrom.
> There is no ISP header, and the flash chip is socketed,
> the board will be a Zotac Atom/NM10 ITX board.

You could stack two SPI chips easily, and switch the CS# pin.
> I've seen programmers
> that explain how to isolate the signals so their programmer can be
> used to do it, but does it require a special programmer, or is it
> something that could be done with a $20 ebay programmer and a little
> wire? And can the board be operated with the programmer still
> attached? I'm getting a lot of conflicting info, hoping someone can
> clarify this.

If you want to reflash with an external programmer, it gets complicated
really fast. Reflashing a chip which is connected to a running system is
extremely risky, and you may fry your board. One way around that is to
hook some buffer in between, but you'd need one which is fast enough for
33 MHz operation (and you want a safety margin to not disturb
communication, so shooting for one with a delay around 10 ns is a good
idea), and then electrically disconnect the flash chip from the board
while flashing.
Maybe flashing a chip in a powered down system with an external
programmer works, but you can't be sure if the external programmer has
enough power to feed 3.3V to the whole board (if the flash chip is
connected there) and you may see undesirable side effects as well.
> If this doesn't work out, how can I flip back and forth between a pair
> of SPI chips?

Switching the CS# pin should do the trick.

Regards,
Carl-Daniel

[coreboot] SPI in-system programming

[Michael Karcher]

Am Sonntag, den 18.07.2010, 18:20 +0200 schrieb Carl-Daniel Hailfinger:
> > There is no ISP header, and the flash chip is socketed,
> > the board will be a Zotac Atom/NM10 ITX board.
>
> You could stack two SPI chips easily, and switch the CS# pin.

Don't forget to switch the inactive chip to a pull-up.

Regards,
  Michael Karcher

[coreboot] SPI in-system programming

[Corey Osgood]

On Sun, Jul 18, 2010 at 12:30 PM, Michael Karcher
 wrote:
> Am Sonntag, den 18.07.2010, 18:20 +0200 schrieb Carl-Daniel Hailfinger:
> > > There is no ISP header, and the flash chip is socketed,
> > > the board will be a Zotac Atom/NM10 ITX board.
> >
> > You could stack two SPI chips easily, and switch the CS# pin.
>
> Don't forget to switch the inactive chip to a pull-up.

Thanks for the info, didn't realize it was quite that simple, looks
like that's the route I'll be going.

-Corey

[coreboot] SPI in-system programming

[Rudolf Marek]

Hi,

I guess you need to have some driver to disconnect the bus from chipset.
And maybe some diode not to power on whole MB while doing ISP when MB is off
Check this:

www.dediprog.com/chipset/via8237s.pdf

Maybe you will need tri state buffer?

Thanks,
Rudolf

[coreboot] SPI in-system programming

[Corey Osgood]

On Sun, Jul 18, 2010 at 1:11 PM, Rudolf Marek  wrote:
> Hi,
>
> I guess you need to have some driver to disconnect the bus from chipset.
> And maybe some diode not to power on whole MB while doing ISP when MB is off
> Check this:
>
> www.dediprog.com/chipset/via8237s.pdf
>
> Maybe you will need tri state buffer?

I've seen that, along with the intel version:

http://www.dediprog.com/chipset/In%20System%20Programming%20Solution%20for%2...

But they rely on an SPI header and the reference design, neither of
which I'll have. I was hoping someone had some solution for hooking a
programmer directly to a board, but with switching chips so easy I
can't imagine why they'd bother.

-Corey

[coreboot] SPI in-system programming

[baiyin cai]

SF100 can fit your requirement with testclip SO8, you can see the details
from:
http://www.dediprog.com/SPI-flash-in-circuit-programming/ISP-Testclip-SO8

2010/7/19 Corey Osgood
> On Sun, Jul 18, 2010 at 1:11 PM, Rudolf Marek 
> wrote:
> > Hi,
> >
> > I guess you need to have some driver to disconnect the bus from chipset.
> > And maybe some diode not to power on whole MB while doing ISP when MB is
>
> off
> > Check this:
> >
> > www.dediprog.com/chipset/via8237s.pdf
> >
> > Maybe you will need tri state buffer?
>
> I've seen that, along with the intel version:
>
> http://www.dediprog.com/chipset/In%20System%20Programming%20Solution%20for%2...
>
> But they rely on an SPI header and the reference design, neither of
> which I'll have. I was hoping someone had some solution for hooking a
> programmer directly to a board, but with switching chips so easy I
> can't imagine why they'd bother.
>
> -Corey