This is not really an LPC issue, even though I am using a LPC2148 to
interface with SD cards.

I'm finally back on tackling the SD card issue with different
manufacturers and making sure all work as advertised as far as
initialization and determining size and type of card for formatting
purposes. I have about 20 SD cards just about all from different
vendors. However, some are OEM I expect, because the some MIDs are
the same.

The only good code initialization code that I have found is Joel W.'s
code which handles all types of cards, and I have based my code on
that. The other code on the site and the Application Note by Philips
only seems to handle Version 1 cards and MMC.

I used the SD Card Spec's (9/2006) flow chart, the 'How to Use
MMC/SDC' description on elm-chan.org, Joel's code, and a SPI Beagle
Analyzer from Total Phase to look at all scenarios of initialization.
I'm trying to put together a better flow chart with exactly what
status to expect and how to handle each status. Hopefully, that will
be of some service to someone - all in one place.

However, I'm getting some unexplained status on some cards upon init,
but not on some other cards of the same size. So, I was hoping
someone might have an answer.

I am getting 3 basic scenarios, as expected, when initializing a
random SD card.

- CMD8 legal, READ_OCR gives valid determination of HCSD card or
SD-Version

- CMD8 illegal, READ_OCR gives valid determination of SD-Version 1.

- CMD8 illegal, READ_OCR give invalid readings (but are Version 1 cards).

On, this last scenario, if you get an illegal CMD8, the READ_OCR gives
all FFs on read. If you read it again, it gives proper voltage
information. If you read it a 3rd time, it gives you all FFs again.
So, every odd read, you get all FFs and every even read you get proper
info. I have 4 Version 1 cards (PNY-1GB, PQ1-1GB, SanDisk-256MB,
Sandisk-MicroSD) that have this anomoly. I have 7 other Version 1
cards (4-1GB, 1-2GB, 1-512MB, 1-128MB) that exhibit the correct
READ_OCR every time. I thought that maybe if you got an illegal CMD8,
that you had to return to idle mode to get the card back to a known
state, but since 7 other cards work fine, I don't believe so. Anybody
had any issues like this?

Also, does anyone know - Can a particular size card be a Version 1 or
Version 2 depending on how the vendor wants to format the card? I
have 2-1GB cards that are Version 2 and 9-1GB that are Version 1. I
have a 2GB that is a Version 1. I was wondering if a 512MB or lower
can be a Version 2?

Sutton

Start a New Thread

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Sutton Mehaffey wrote:
| This is not really an LPC issue, even though I am using a LPC2148 to
| interface with SD cards.
|
| I'm finally back on tackling the SD card issue with different
| manufacturers and making sure all work as advertised as far as
| initialization and determining size and type of card for formatting
| purposes. I have about 20 SD cards just about all from different
| vendors. However, some are OEM I expect, because the some MIDs are
| the same.
|
| The only good code initialization code that I have found is Joel W.'s
| code which handles all types of cards, and I have based my code on
| that. The other code on the site and the Application Note by Philips
| only seems to handle Version 1 cards and MMC.
|
| I used the SD Card Spec's (9/2006) flow chart, the 'How to Use
| MMC/SDC' description on elm-chan.org, Joel's code, and a SPI Beagle
| Analyzer from Total Phase to look at all scenarios of initialization.
| I'm trying to put together a better flow chart with exactly what
| status to expect and how to handle each status. Hopefully, that will
| be of some service to someone - all in one place.
|
| However, I'm getting some unexplained status on some cards upon init,
| but not on some other cards of the same size. So, I was hoping
| someone might have an answer.
|
| I am getting 3 basic scenarios, as expected, when initializing a
| random SD card.
|
| - CMD8 legal, READ_OCR gives valid determination of HCSD card or
| SD-Version
|
| - CMD8 illegal, READ_OCR gives valid determination of SD-Version 1.
|
| - CMD8 illegal, READ_OCR give invalid readings (but are Version 1 cards).
|
| On, this last scenario, if you get an illegal CMD8, the READ_OCR gives
| all FFs on read. If you read it again, it gives proper voltage
| information. If you read it a 3rd time, it gives you all FFs again.
| So, every odd read, you get all FFs and every even read you get proper
| info. I have 4 Version 1 cards (PNY-1GB, PQ1-1GB, SanDisk-256MB,
| Sandisk-MicroSD) that have this anomoly. I have 7 other Version 1
| cards (4-1GB, 1-2GB, 1-512MB, 1-128MB) that exhibit the correct
| READ_OCR every time. I thought that maybe if you got an illegal CMD8,
| that you had to return to idle mode to get the card back to a known
| state, but since 7 other cards work fine, I don't believe so. Anybody
| had any issues like this?
|
| Also, does anyone know - Can a particular size card be a Version 1 or
| Version 2 depending on how the vendor wants to format the card? I
| have 2-1GB cards that are Version 2 and 9-1GB that are Version 1. I
| have a 2GB that is a Version 1. I was wondering if a 512MB or lower
| can be a Version 2?

Hi Sutton,

I haven't tested as many cards as you have (only 5 or or 6), but I
have no problem accessing them so far. I just tried to implement the
flow chart that's in the simplified sd specification.

One thing I got wrong at first, was always trying to read the extended
response to CMD8 regardless whether CMD8 succeeded or not.

To get the op cond, I send the "applicative CMD41", which is cmd55
followed by cmd41. I use a CRC only for two commands:
CMD_GO_IDLE_STATE (CMD0) and CMD_SEND_IF_COND (CMD8)

Things I'm not really sure about are timing and how often CMD0 and
CMD41 have to be repeated.

Kind regards,
Bertrik
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFIQRk4ETD6mlrWxPURArCmAJsEY3YdTvp0jjnd0/aQruNr6qY2uQCdF9KQ
2gcXFdvH66v1icJPvilUDbw=0oMR
-----END PGP SIGNATURE-----

Start a New Thread

Bertrik,

I tried that too. When CMD8 was rejected, I tried reading 4 more
bytes (they were all FFs) to see if that corrected the problem. It
didn't. The wierd thing is that 7 other cards work fine with a
rejected CMD8. And 2 out of the 3 cards that have that READ_OCR odd-
even anomoly are PNY and SanDisk cards, which are supposedly good
cards.

And, like you, I haven't found the answer on when to issue CMD0 (or
if you even need to) upon illegal or bad command responses. The only
thing I did confirm is that CMD55-ACMD41 is the init sequence and you
have to continually issue this sequence and wait for the status to be
0x00 before continuing. 0x01 means 'still initializing'. You can't
just send dummy bytes and read status.

Sutton

--- In l..., Bertrik Sikken wrote:
>
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Sutton Mehaffey wrote:
> | This is not really an LPC issue, even though I am using a LPC2148
to
> | interface with SD cards.
> |
> | I'm finally back on tackling the SD card issue with different
> | manufacturers and making sure all work as advertised as far as
> | initialization and determining size and type of card for
formatting
> | purposes. I have about 20 SD cards just about all from different
> | vendors. However, some are OEM I expect, because the some MIDs
are
> | the same.
> |
> | The only good code initialization code that I have found is Joel
W.'s
> | code which handles all types of cards, and I have based my code on
> | that. The other code on the site and the Application Note by
Philips
> | only seems to handle Version 1 cards and MMC.
> |
> | I used the SD Card Spec's (9/2006) flow chart, the 'How to Use
> | MMC/SDC' description on elm-chan.org, Joel's code, and a SPI
Beagle
> | Analyzer from Total Phase to look at all scenarios of
initialization.
> | I'm trying to put together a better flow chart with exactly what
> | status to expect and how to handle each status. Hopefully, that
will
> | be of some service to someone - all in one place.
> |
> | However, I'm getting some unexplained status on some cards upon
init,
> | but not on some other cards of the same size. So, I was hoping
> | someone might have an answer.
> |
> | I am getting 3 basic scenarios, as expected, when initializing a
> | random SD card.
> |
> | - CMD8 legal, READ_OCR gives valid determination of HCSD card or
> | SD-Version
> |
> | - CMD8 illegal, READ_OCR gives valid determination of SD-Version
1.
> |
> | - CMD8 illegal, READ_OCR give invalid readings (but are Version 1
cards).
> |
> | On, this last scenario, if you get an illegal CMD8, the READ_OCR
gives
> | all FFs on read. If you read it again, it gives proper voltage
> | information. If you read it a 3rd time, it gives you all FFs
again.
> | So, every odd read, you get all FFs and every even read you get
proper
> | info. I have 4 Version 1 cards (PNY-1GB, PQ1-1GB, SanDisk-256MB,
> | Sandisk-MicroSD) that have this anomoly. I have 7 other Version 1
> | cards (4-1GB, 1-2GB, 1-512MB, 1-128MB) that exhibit the correct
> | READ_OCR every time. I thought that maybe if you got an illegal
CMD8,
> | that you had to return to idle mode to get the card back to a
known
> | state, but since 7 other cards work fine, I don't believe so.
Anybody
> | had any issues like this?
> |
> | Also, does anyone know - Can a particular size card be a Version
1 or
> | Version 2 depending on how the vendor wants to format the card? I
> | have 2-1GB cards that are Version 2 and 9-1GB that are Version
1. I
> | have a 2GB that is a Version 1. I was wondering if a 512MB or
lower
> | can be a Version 2?
>
> Hi Sutton,
>
> I haven't tested as many cards as you have (only 5 or or 6), but I
> have no problem accessing them so far. I just tried to implement the
> flow chart that's in the simplified sd specification.
>
> One thing I got wrong at first, was always trying to read the
extended
> response to CMD8 regardless whether CMD8 succeeded or not.
>
> To get the op cond, I send the "applicative CMD41", which is cmd55
> followed by cmd41. I use a CRC only for two commands:
> CMD_GO_IDLE_STATE (CMD0) and CMD_SEND_IF_COND (CMD8)
>
> Things I'm not really sure about are timing and how often CMD0 and
> CMD41 have to be repeated.
>
> Kind regards,
> Bertrik
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.7 (MingW32)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
>
> iD8DBQFIQRk4ETD6mlrWxPURArCmAJsEY3YdTvp0jjnd0/aQruNr6qY2uQCdF9KQ
> 2gcXFdvH66v1icJPvilUDbw> =0oMR
> -----END PGP SIGNATURE-----
>

Start a New Thread

--- In l..., "Sutton Mehaffey" wrote:
>
> This is not really an LPC issue, even though I am using a LPC2148 to
> interface with SD cards.
>
> I'm finally back on tackling the SD card issue with different
> manufacturers and making sure all work as advertised as far as
> initialization and determining size and type of card for formatting
> purposes. I have about 20 SD cards just about all from different
> vendors. However, some are OEM I expect, because the some MIDs are
> the same.
>
> The only good code initialization code that I have found is Joel W.'s
> code which handles all types of cards, and I have based my code on
> that. The other code on the site and the Application Note by Philips
> only seems to handle Version 1 cards and MMC.
>
> I used the SD Card Spec's (9/2006) flow chart, the 'How to Use
> MMC/SDC' description on elm-chan.org, Joel's code, and a SPI Beagle
> Analyzer from Total Phase to look at all scenarios of initialization.
> I'm trying to put together a better flow chart with exactly what
> status to expect and how to handle each status. Hopefully, that will
> be of some service to someone - all in one place.
>
> However, I'm getting some unexplained status on some cards upon init,
> but not on some other cards of the same size. So, I was hoping
> someone might have an answer.
>
> I am getting 3 basic scenarios, as expected, when initializing a
> random SD card.
>
> - CMD8 legal, READ_OCR gives valid determination of HCSD card or
> SD-Version
>
> - CMD8 illegal, READ_OCR gives valid determination of SD-Version 1.
>
> - CMD8 illegal, READ_OCR give invalid readings (but are Version 1
cards).
>
> On, this last scenario, if you get an illegal CMD8, the READ_OCR gives
> all FFs on read. If you read it again, it gives proper voltage
> information. If you read it a 3rd time, it gives you all FFs again.
> So, every odd read, you get all FFs and every even read you get proper
> info. I have 4 Version 1 cards (PNY-1GB, PQ1-1GB, SanDisk-256MB,
> Sandisk-MicroSD) that have this anomoly. I have 7 other Version 1
> cards (4-1GB, 1-2GB, 1-512MB, 1-128MB) that exhibit the correct
> READ_OCR every time. I thought that maybe if you got an illegal CMD8,
> that you had to return to idle mode to get the card back to a known
> state, but since 7 other cards work fine, I don't believe so. Anybody
> had any issues like this?
>
> Also, does anyone know - Can a particular size card be a Version 1 or
> Version 2 depending on how the vendor wants to format the card? I
> have 2-1GB cards that are Version 2 and 9-1GB that are Version 1. I
> have a 2GB that is a Version 1. I was wondering if a 512MB or lower
> can be a Version 2?
>
> Sutton
>

Start a New Thread

One of my cards (Edge 512MB) causes a hugh momentary voltage drop
(3.3V down to about 2.5V) and causes a brownout on my LPC2148, which
causes a reset. No other card does this. Does anyone think this is a
bad card or should I expect this occasionally and put a 1uF cap on the
power bus instead of my .1uf?

I have a Lexar 128MB card that gives me an ILLEGAL COMMAND for
READ_CSD (Status 04), but reads other registers OK (CID). I don't
have another 128MB card to test. Are there conditions whereby a 128MB
card doesn't have a CSD?

Sutton

--- In l..., "Sutton Mehaffey" wrote:
>
> This is not really an LPC issue, even though I am using a LPC2148 to
> interface with SD cards.
>
> I'm finally back on tackling the SD card issue with different
> manufacturers and making sure all work as advertised as far as
> initialization and determining size and type of card for formatting
> purposes. I have about 20 SD cards just about all from different
> vendors. However, some are OEM I expect, because the some MIDs are
> the same.
>
> The only good code initialization code that I have found is Joel W.'s
> code which handles all types of cards, and I have based my code on
> that. The other code on the site and the Application Note by Philips
> only seems to handle Version 1 cards and MMC.
>
> I used the SD Card Spec's (9/2006) flow chart, the 'How to Use
> MMC/SDC' description on elm-chan.org, Joel's code, and a SPI Beagle
> Analyzer from Total Phase to look at all scenarios of initialization.
> I'm trying to put together a better flow chart with exactly what
> status to expect and how to handle each status. Hopefully, that will
> be of some service to someone - all in one place.
>
> However, I'm getting some unexplained status on some cards upon init,
> but not on some other cards of the same size. So, I was hoping
> someone might have an answer.
>
> I am getting 3 basic scenarios, as expected, when initializing a
> random SD card.
>
> - CMD8 legal, READ_OCR gives valid determination of HCSD card or
> SD-Version
>
> - CMD8 illegal, READ_OCR gives valid determination of SD-Version 1.
>
> - CMD8 illegal, READ_OCR give invalid readings (but are Version 1
cards).
>
> On, this last scenario, if you get an illegal CMD8, the READ_OCR gives
> all FFs on read. If you read it again, it gives proper voltage
> information. If you read it a 3rd time, it gives you all FFs again.
> So, every odd read, you get all FFs and every even read you get proper
> info. I have 4 Version 1 cards (PNY-1GB, PQ1-1GB, SanDisk-256MB,
> Sandisk-MicroSD) that have this anomoly. I have 7 other Version 1
> cards (4-1GB, 1-2GB, 1-512MB, 1-128MB) that exhibit the correct
> READ_OCR every time. I thought that maybe if you got an illegal CMD8,
> that you had to return to idle mode to get the card back to a known
> state, but since 7 other cards work fine, I don't believe so. Anybody
> had any issues like this?
>
> Also, does anyone know - Can a particular size card be a Version 1 or
> Version 2 depending on how the vendor wants to format the card? I
> have 2-1GB cards that are Version 2 and 9-1GB that are Version 1. I
> have a 2GB that is a Version 1. I was wondering if a 512MB or lower
> can be a Version 2?
>
> Sutton
>

Start a New Thread

At 15:36 02.06.2008 +0000, you wrote:
>One of my cards (Edge 512MB) causes a hugh momentary voltage drop
>(3.3V down to about 2.5V) and causes a brownout on my LPC2148, which
>causes a reset. No other card does this. Does anyone think this is a
>bad card or should I expect this occasionally and put a 1uF cap on the
>power bus instead of my .1uf?

Ooops, you should have at least 4u7 *and* a 22uH in series, this is
what I've seen recommended to exactly avoid what you mentioned above.
I'm just designing a board with a (micro-)SD card and originally did
not understand what sense the choke has (and found the explanation on
the web later then), but thanks to your post now I *know* that it
does make sense ;-) - see at http://elm-chan.org/docs/mmc/mmc_e.html
"Consideration to Bus Floating and Hot Insertion"

Herbert
>I have a Lexar 128MB card that gives me an ILLEGAL COMMAND for
>READ_CSD (Status 04), but reads other registers OK (CID). I don't
>have another 128MB card to test. Are there conditions whereby a 128MB
>card doesn't have a CSD?
>
>Sutton
>
>--- In l..., "Sutton Mehaffey" wrote:
> >
> > This is not really an LPC issue, even though I am using a LPC2148 to
> > interface with SD cards.
> >
> > I'm finally back on tackling the SD card issue with different
> > manufacturers and making sure all work as advertised as far as
> > initialization and determining size and type of card for formatting
> > purposes. I have about 20 SD cards just about all from different
> > vendors. However, some are OEM I expect, because the some MIDs are
> > the same.
> >
> > The only good code initialization code that I have found is Joel W.'s
> > code which handles all types of cards, and I have based my code on
> > that. The other code on the site and the Application Note by Philips
> > only seems to handle Version 1 cards and MMC.
> >
> > I used the SD Card Spec's (9/2006) flow chart, the 'How to Use
> > MMC/SDC' description on elm-chan.org, Joel's code, and a SPI Beagle
> > Analyzer from Total Phase to look at all scenarios of initialization.
> > I'm trying to put together a better flow chart with exactly what
> > status to expect and how to handle each status. Hopefully, that will
> > be of some service to someone - all in one place.
> >
> > However, I'm getting some unexplained status on some cards upon init,
> > but not on some other cards of the same size. So, I was hoping
> > someone might have an answer.
> >
> > I am getting 3 basic scenarios, as expected, when initializing a
> > random SD card.
> >
> > - CMD8 legal, READ_OCR gives valid determination of HCSD card or
> > SD-Version
> >
> > - CMD8 illegal, READ_OCR gives valid determination of SD-Version 1.
> >
> > - CMD8 illegal, READ_OCR give invalid readings (but are Version 1
>cards).
> >
> > On, this last scenario, if you get an illegal CMD8, the READ_OCR gives
> > all FFs on read. If you read it again, it gives proper voltage
> > information. If you read it a 3rd time, it gives you all FFs again.
> > So, every odd read, you get all FFs and every even read you get proper
> > info. I have 4 Version 1 cards (PNY-1GB, PQ1-1GB, SanDisk-256MB,
> > Sandisk-MicroSD) that have this anomoly. I have 7 other Version 1
> > cards (4-1GB, 1-2GB, 1-512MB, 1-128MB) that exhibit the correct
> > READ_OCR every time. I thought that maybe if you got an illegal CMD8,
> > that you had to return to idle mode to get the card back to a known
> > state, but since 7 other cards work fine, I don't believe so. Anybody
> > had any issues like this?
> >
> > Also, does anyone know - Can a particular size card be a Version 1 or
> > Version 2 depending on how the vendor wants to format the card? I
> > have 2-1GB cards that are Version 2 and 9-1GB that are Version 1. I
> > have a 2GB that is a Version 1. I was wondering if a 512MB or lower
> > can be a Version 2?
> >
> > Sutton
> > Start a New Thread

Thanks. That explains it.

--- In l..., Herbert Demmel wrote:
>
> At 15:36 02.06.2008 +0000, you wrote:
> >One of my cards (Edge 512MB) causes a hugh momentary voltage drop
> >(3.3V down to about 2.5V) and causes a brownout on my LPC2148, which
> >causes a reset. No other card does this. Does anyone think this is a
> >bad card or should I expect this occasionally and put a 1uF cap on the
> >power bus instead of my .1uf?
>
> Ooops, you should have at least 4u7 *and* a 22uH in series, this is
> what I've seen recommended to exactly avoid what you mentioned above.
> I'm just designing a board with a (micro-)SD card and originally did
> not understand what sense the choke has (and found the explanation on
> the web later then), but thanks to your post now I *know* that it
> does make sense ;-) - see at http://elm-chan.org/docs/mmc/mmc_e.html
> "Consideration to Bus Floating and Hot Insertion"
>
> Herbert
> >I have a Lexar 128MB card that gives me an ILLEGAL COMMAND for
> >READ_CSD (Status 04), but reads other registers OK (CID). I don't
> >have another 128MB card to test. Are there conditions whereby a 128MB
> >card doesn't have a CSD?
> >
> >Sutton
> >
> >
> >
> >
> >
> >--- In l..., "Sutton Mehaffey" wrote:
> > >
> > > This is not really an LPC issue, even though I am using a LPC2148 to
> > > interface with SD cards.
> > >
> > > I'm finally back on tackling the SD card issue with different
> > > manufacturers and making sure all work as advertised as far as
> > > initialization and determining size and type of card for formatting
> > > purposes. I have about 20 SD cards just about all from different
> > > vendors. However, some are OEM I expect, because the some MIDs are
> > > the same.
> > >
> > > The only good code initialization code that I have found is Joel
W.'s
> > > code which handles all types of cards, and I have based my code on
> > > that. The other code on the site and the Application Note by
Philips
> > > only seems to handle Version 1 cards and MMC.
> > >
> > > I used the SD Card Spec's (9/2006) flow chart, the 'How to Use
> > > MMC/SDC' description on elm-chan.org, Joel's code, and a SPI Beagle
> > > Analyzer from Total Phase to look at all scenarios of
initialization.
> > > I'm trying to put together a better flow chart with exactly what
> > > status to expect and how to handle each status. Hopefully, that
will
> > > be of some service to someone - all in one place.
> > >
> > > However, I'm getting some unexplained status on some cards upon
init,
> > > but not on some other cards of the same size. So, I was hoping
> > > someone might have an answer.
> > >
> > > I am getting 3 basic scenarios, as expected, when initializing a
> > > random SD card.
> > >
> > > - CMD8 legal, READ_OCR gives valid determination of HCSD card or
> > > SD-Version
> > >
> > > - CMD8 illegal, READ_OCR gives valid determination of SD-Version 1.
> > >
> > > - CMD8 illegal, READ_OCR give invalid readings (but are Version 1
> >cards).
> > >
> > > On, this last scenario, if you get an illegal CMD8, the READ_OCR
gives
> > > all FFs on read. If you read it again, it gives proper voltage
> > > information. If you read it a 3rd time, it gives you all FFs again.
> > > So, every odd read, you get all FFs and every even read you get
proper
> > > info. I have 4 Version 1 cards (PNY-1GB, PQ1-1GB, SanDisk-256MB,
> > > Sandisk-MicroSD) that have this anomoly. I have 7 other Version 1
> > > cards (4-1GB, 1-2GB, 1-512MB, 1-128MB) that exhibit the correct
> > > READ_OCR every time. I thought that maybe if you got an illegal
CMD8,
> > > that you had to return to idle mode to get the card back to a known
> > > state, but since 7 other cards work fine, I don't believe so.
Anybody
> > > had any issues like this?
> > >
> > > Also, does anyone know - Can a particular size card be a Version
1 or
> > > Version 2 depending on how the vendor wants to format the card? I
> > > have 2-1GB cards that are Version 2 and 9-1GB that are Version 1. I
> > > have a 2GB that is a Version 1. I was wondering if a 512MB or lower
> > > can be a Version 2?
> > >
> > > Sutton
> > >
> >
> >
> >
> >
> >
> > Start a New Thread

OK. I think I figured something out. Apparently, the flow chart on
page 95 of the SD Spec. is not correct for all SD cards.

It appears there can be two VERSION 1 setup possibilities when a CMD8
is rejected (reject means ILLEGAL COMMAND).

Some SD cards (and most in my case) can execute CMD58 and initialize
on ACMD41 successfully as per the spec.

Other cards cannot. After rejecting CMD8, some cards also reject
CMD58 and CMD55 (so you can't use ACMD41 as the init command). In
this case, you must first issue CMD1 as the init command until status
is 00. THEN, you can READ_OCR. This seems to solve the problem.

In my case, I have 4 1GB, 1 2GB, and 1 512MB cards that reject CMD8,
but can immediately READ_OCR and then be initialized by CMD55-ACMD41.
I have 3 other 1GB cards that require the 2nd method of
initialization described above.

I hope to write up a better flow chart once I feel 100% confident that
all cards are working properly.

Any other input to this is appreciated.

Sutton

--- In l..., Herbert Demmel wrote:
>
> At 15:36 02.06.2008 +0000, you wrote:
> >One of my cards (Edge 512MB) causes a hugh momentary voltage drop
> >(3.3V down to about 2.5V) and causes a brownout on my LPC2148, which
> >causes a reset. No other card does this. Does anyone think this is a
> >bad card or should I expect this occasionally and put a 1uF cap on the
> >power bus instead of my .1uf?
>
> Ooops, you should have at least 4u7 *and* a 22uH in series, this is
> what I've seen recommended to exactly avoid what you mentioned above.
> I'm just designing a board with a (micro-)SD card and originally did
> not understand what sense the choke has (and found the explanation on
> the web later then), but thanks to your post now I *know* that it
> does make sense ;-) - see at http://elm-chan.org/docs/mmc/mmc_e.html
> "Consideration to Bus Floating and Hot Insertion"
>
> Herbert
> >I have a Lexar 128MB card that gives me an ILLEGAL COMMAND for
> >READ_CSD (Status 04), but reads other registers OK (CID). I don't
> >have another 128MB card to test. Are there conditions whereby a 128MB
> >card doesn't have a CSD?
> >
> >Sutton
> >
> >
> >
> >
> >
> >--- In l..., "Sutton Mehaffey" wrote:
> > >
> > > This is not really an LPC issue, even though I am using a LPC2148 to
> > > interface with SD cards.
> > >
> > > I'm finally back on tackling the SD card issue with different
> > > manufacturers and making sure all work as advertised as far as
> > > initialization and determining size and type of card for formatting
> > > purposes. I have about 20 SD cards just about all from different
> > > vendors. However, some are OEM I expect, because the some MIDs are
> > > the same.
> > >
> > > The only good code initialization code that I have found is Joel
W.'s
> > > code which handles all types of cards, and I have based my code on
> > > that. The other code on the site and the Application Note by
Philips
> > > only seems to handle Version 1 cards and MMC.
> > >
> > > I used the SD Card Spec's (9/2006) flow chart, the 'How to Use
> > > MMC/SDC' description on elm-chan.org, Joel's code, and a SPI Beagle
> > > Analyzer from Total Phase to look at all scenarios of
initialization.
> > > I'm trying to put together a better flow chart with exactly what
> > > status to expect and how to handle each status. Hopefully, that
will
> > > be of some service to someone - all in one place.
> > >
> > > However, I'm getting some unexplained status on some cards upon
init,
> > > but not on some other cards of the same size. So, I was hoping
> > > someone might have an answer.
> > >
> > > I am getting 3 basic scenarios, as expected, when initializing a
> > > random SD card.
> > >
> > > - CMD8 legal, READ_OCR gives valid determination of HCSD card or
> > > SD-Version
> > >
> > > - CMD8 illegal, READ_OCR gives valid determination of SD-Version 1.
> > >
> > > - CMD8 illegal, READ_OCR give invalid readings (but are Version 1
> >cards).
> > >
> > > On, this last scenario, if you get an illegal CMD8, the READ_OCR
gives
> > > all FFs on read. If you read it again, it gives proper voltage
> > > information. If you read it a 3rd time, it gives you all FFs again.
> > > So, every odd read, you get all FFs and every even read you get
proper
> > > info. I have 4 Version 1 cards (PNY-1GB, PQ1-1GB, SanDisk-256MB,
> > > Sandisk-MicroSD) that have this anomoly. I have 7 other Version 1
> > > cards (4-1GB, 1-2GB, 1-512MB, 1-128MB) that exhibit the correct
> > > READ_OCR every time. I thought that maybe if you got an illegal
CMD8,
> > > that you had to return to idle mode to get the card back to a known
> > > state, but since 7 other cards work fine, I don't believe so.
Anybody
> > > had any issues like this?
> > >
> > > Also, does anyone know - Can a particular size card be a Version
1 or
> > > Version 2 depending on how the vendor wants to format the card? I
> > > have 2-1GB cards that are Version 2 and 9-1GB that are Version 1. I
> > > have a 2GB that is a Version 1. I was wondering if a 512MB or lower
> > > can be a Version 2?
> > >
> > > Sutton
> > >
> >
> >
> >
> >
> >
> > Start a New Thread

Sutton,

you definitely seem to make a good job when finding out all those
weird issues. As mentioned in my former email, I'm just now designing
the hardware (the SD-specific firmware will follow in some months or
so), so I currently have absolutely no knowledge about SD cards yet,
but I would be interested in getting some source code from you, if
this is possible (once you have finished your work).

Please let me know if this is possible.

Best regards
Herbert

At 18:28 02.06.2008 +0000, you wrote:
>OK. I think I figured something out. Apparently, the flow chart on
>page 95 of the SD Spec. is not correct for all SD cards.
>
>It appears there can be two VERSION 1 setup possibilities when a CMD8
>is rejected (reject means ILLEGAL COMMAND).
>
>Some SD cards (and most in my case) can execute CMD58 and initialize
>on ACMD41 successfully as per the spec.
>
>Other cards cannot. After rejecting CMD8, some cards also reject
>CMD58 and CMD55 (so you can't use ACMD41 as the init command). In
>this case, you must first issue CMD1 as the init command until status
>is 00. THEN, you can READ_OCR. This seems to solve the problem.
>
>In my case, I have 4 1GB, 1 2GB, and 1 512MB cards that reject CMD8,
>but can immediately READ_OCR and then be initialized by CMD55-ACMD41.
> I have 3 other 1GB cards that require the 2nd method of
>initialization described above.
>
>I hope to write up a better flow chart once I feel 100% confident that
>all cards are working properly.
>
>Any other input to this is appreciated.
>
>Sutton
>--- In l..., Herbert Demmel wrote:
> >
> > At 15:36 02.06.2008 +0000, you wrote:
> > >One of my cards (Edge 512MB) causes a hugh momentary voltage drop
> > >(3.3V down to about 2.5V) and causes a brownout on my LPC2148, which
> > >causes a reset. No other card does this. Does anyone think this is a
> > >bad card or should I expect this occasionally and put a 1uF cap on the
> > >power bus instead of my .1uf?
> >
> > Ooops, you should have at least 4u7 *and* a 22uH in series, this is
> > what I've seen recommended to exactly avoid what you mentioned above.
> > I'm just designing a board with a (micro-)SD card and originally did
> > not understand what sense the choke has (and found the explanation on
> > the web later then), but thanks to your post now I *know* that it
> > does make sense ;-) - see at http://elm-chan.org/docs/mmc/mmc_e.html
> > "Consideration to Bus Floating and Hot Insertion"
> >
> > Herbert
> >
> >
> > >I have a Lexar 128MB card that gives me an ILLEGAL COMMAND for
> > >READ_CSD (Status 04), but reads other registers OK (CID). I don't
> > >have another 128MB card to test. Are there conditions whereby a 128MB
> > >card doesn't have a CSD?
> > >
> > >Sutton
> > >
> > >
> > >
> > >
> > >
> > >--- In l..., "Sutton Mehaffey" wrote:
> > > >
> > > > This is not really an LPC issue, even though I am using a LPC2148 to
> > > > interface with SD cards.
> > > >
> > > > I'm finally back on tackling the SD card issue with different
> > > > manufacturers and making sure all work as advertised as far as
> > > > initialization and determining size and type of card for formatting
> > > > purposes. I have about 20 SD cards just about all from different
> > > > vendors. However, some are OEM I expect, because the some MIDs are
> > > > the same.
> > > >
> > > > The only good code initialization code that I have found is Joel
>W.'s
> > > > code which handles all types of cards, and I have based my code on
> > > > that. The other code on the site and the Application Note by
>Philips
> > > > only seems to handle Version 1 cards and MMC.
> > > >
> > > > I used the SD Card Spec's (9/2006) flow chart, the 'How to Use
> > > > MMC/SDC' description on elm-chan.org, Joel's code, and a SPI Beagle
> > > > Analyzer from Total Phase to look at all scenarios of
>initialization.
> > > > I'm trying to put together a better flow chart with exactly what
> > > > status to expect and how to handle each status. Hopefully, that
>will
> > > > be of some service to someone - all in one place.
> > > >
> > > > However, I'm getting some unexplained status on some cards upon
>init,
> > > > but not on some other cards of the same size. So, I was hoping
> > > > someone might have an answer.
> > > >
> > > > I am getting 3 basic scenarios, as expected, when initializing a
> > > > random SD card.
> > > >
> > > > - CMD8 legal, READ_OCR gives valid determination of HCSD card or
> > > > SD-Version
> > > >
> > > > - CMD8 illegal, READ_OCR gives valid determination of SD-Version 1.
> > > >
> > > > - CMD8 illegal, READ_OCR give invalid readings (but are Version 1
> > >cards).
> > > >
> > > > On, this last scenario, if you get an illegal CMD8, the READ_OCR
>gives
> > > > all FFs on read. If you read it again, it gives proper voltage
> > > > information. If you read it a 3rd time, it gives you all FFs again.
> > > > So, every odd read, you get all FFs and every even read you get
>proper
> > > > info. I have 4 Version 1 cards (PNY-1GB, PQ1-1GB, SanDisk-256MB,
> > > > Sandisk-MicroSD) that have this anomoly. I have 7 other Version 1
> > > > cards (4-1GB, 1-2GB, 1-512MB, 1-128MB) that exhibit the correct
> > > > READ_OCR every time. I thought that maybe if you got an illegal
>CMD8,
> > > > that you had to return to idle mode to get the card back to a known
> > > > state, but since 7 other cards work fine, I don't believe so.
>Anybody
> > > > had any issues like this?
> > > >
> > > > Also, does anyone know - Can a particular size card be a Version
>1 or
> > > > Version 2 depending on how the vendor wants to format the card? I
> > > > have 2-1GB cards that are Version 2 and 9-1GB that are Version 1. I
> > > > have a 2GB that is a Version 1. I was wondering if a 512MB or lower
> > > > can be a Version 2?
> > > >
> > > > Sutton
> > > >
> > >
> > >
> > >
> > >
> > >
> > > Start a New Thread

I will post code and a flow chart when finished. If anyone else comes
up with anything for me to try out, I will do my best to accommodate.

I also noticed that the second initialize case for Version 1 cards
seems to be the older MMC/SD method (where init was done before
CMD58). However, I don't see any mention of that method in the Spec
anymore. I did confirm that the older init method works for either
version of the Version 1 card. But, I'm assuming they (SD) wanted you
to issue CMD58 to check the operation voltage before initializing for
some reason. This, however, is impossible because of the rejected
CMD58 on some Version 1 cards.

Sutton

--- In l..., Herbert Demmel wrote:
>
> Sutton,
>
> you definitely seem to make a good job when finding out all those
> weird issues. As mentioned in my former email, I'm just now designing
> the hardware (the SD-specific firmware will follow in some months or
> so), so I currently have absolutely no knowledge about SD cards yet,
> but I would be interested in getting some source code from you, if
> this is possible (once you have finished your work).
>
> Please let me know if this is possible.
>
> Best regards
> Herbert
>
> At 18:28 02.06.2008 +0000, you wrote:
> >OK. I think I figured something out. Apparently, the flow chart on
> >page 95 of the SD Spec. is not correct for all SD cards.
> >
> >It appears there can be two VERSION 1 setup possibilities when a CMD8
> >is rejected (reject means ILLEGAL COMMAND).
> >
> >Some SD cards (and most in my case) can execute CMD58 and initialize
> >on ACMD41 successfully as per the spec.
> >
> >Other cards cannot. After rejecting CMD8, some cards also reject
> >CMD58 and CMD55 (so you can't use ACMD41 as the init command). In
> >this case, you must first issue CMD1 as the init command until status
> >is 00. THEN, you can READ_OCR. This seems to solve the problem.
> >
> >In my case, I have 4 1GB, 1 2GB, and 1 512MB cards that reject CMD8,
> >but can immediately READ_OCR and then be initialized by CMD55-ACMD41.
> > I have 3 other 1GB cards that require the 2nd method of
> >initialization described above.
> >
> >I hope to write up a better flow chart once I feel 100% confident that
> >all cards are working properly.
> >
> >Any other input to this is appreciated.
> >
> >Sutton
> >
> >
> >
> >
> >--- In l..., Herbert Demmel wrote:
> > >
> > > At 15:36 02.06.2008 +0000, you wrote:
> > > >One of my cards (Edge 512MB) causes a hugh momentary voltage drop
> > > >(3.3V down to about 2.5V) and causes a brownout on my LPC2148,
which
> > > >causes a reset. No other card does this. Does anyone think
this is a
> > > >bad card or should I expect this occasionally and put a 1uF cap
on the
> > > >power bus instead of my .1uf?
> > >
> > > Ooops, you should have at least 4u7 *and* a 22uH in series, this is
> > > what I've seen recommended to exactly avoid what you mentioned
above.
> > > I'm just designing a board with a (micro-)SD card and originally did
> > > not understand what sense the choke has (and found the
explanation on
> > > the web later then), but thanks to your post now I *know* that it
> > > does make sense ;-) - see at http://elm-chan.org/docs/mmc/mmc_e.html
> > > "Consideration to Bus Floating and Hot Insertion"
> > >
> > > Herbert
> > >
> > >
> > > >I have a Lexar 128MB card that gives me an ILLEGAL COMMAND for
> > > >READ_CSD (Status 04), but reads other registers OK (CID). I don't
> > > >have another 128MB card to test. Are there conditions whereby
a 128MB
> > > >card doesn't have a CSD?
> > > >
> > > >Sutton
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >--- In l..., "Sutton Mehaffey" wrote:
> > > > >
> > > > > This is not really an LPC issue, even though I am using a
LPC2148 to
> > > > > interface with SD cards.
> > > > >
> > > > > I'm finally back on tackling the SD card issue with different
> > > > > manufacturers and making sure all work as advertised as far as
> > > > > initialization and determining size and type of card for
formatting
> > > > > purposes. I have about 20 SD cards just about all from
different
> > > > > vendors. However, some are OEM I expect, because the some
MIDs are
> > > > > the same.
> > > > >
> > > > > The only good code initialization code that I have found is Joel
> >W.'s
> > > > > code which handles all types of cards, and I have based my
code on
> > > > > that. The other code on the site and the Application Note by
> >Philips
> > > > > only seems to handle Version 1 cards and MMC.
> > > > >
> > > > > I used the SD Card Spec's (9/2006) flow chart, the 'How to Use
> > > > > MMC/SDC' description on elm-chan.org, Joel's code, and a SPI
Beagle
> > > > > Analyzer from Total Phase to look at all scenarios of
> >initialization.
> > > > > I'm trying to put together a better flow chart with exactly
what
> > > > > status to expect and how to handle each status. Hopefully, that
> >will
> > > > > be of some service to someone - all in one place.
> > > > >
> > > > > However, I'm getting some unexplained status on some cards upon
> >init,
> > > > > but not on some other cards of the same size. So, I was hoping
> > > > > someone might have an answer.
> > > > >
> > > > > I am getting 3 basic scenarios, as expected, when initializing a
> > > > > random SD card.
> > > > >
> > > > > - CMD8 legal, READ_OCR gives valid determination of HCSD card or
> > > > > SD-Version
> > > > >
> > > > > - CMD8 illegal, READ_OCR gives valid determination of
SD-Version 1.
> > > > >
> > > > > - CMD8 illegal, READ_OCR give invalid readings (but are
Version 1
> > > >cards).
> > > > >
> > > > > On, this last scenario, if you get an illegal CMD8, the READ_OCR
> >gives
> > > > > all FFs on read. If you read it again, it gives proper voltage
> > > > > information. If you read it a 3rd time, it gives you all
FFs again.
> > > > > So, every odd read, you get all FFs and every even read you get
> >proper
> > > > > info. I have 4 Version 1 cards (PNY-1GB, PQ1-1GB,
SanDisk-256MB,
> > > > > Sandisk-MicroSD) that have this anomoly. I have 7 other
Version 1
> > > > > cards (4-1GB, 1-2GB, 1-512MB, 1-128MB) that exhibit the correct
> > > > > READ_OCR every time. I thought that maybe if you got an illegal
> >CMD8,
> > > > > that you had to return to idle mode to get the card back to
a known
> > > > > state, but since 7 other cards work fine, I don't believe so.
> >Anybody
> > > > > had any issues like this?
> > > > >
> > > > > Also, does anyone know - Can a particular size card be a Version
> >1 or
> > > > > Version 2 depending on how the vendor wants to format the
card? I
> > > > > have 2-1GB cards that are Version 2 and 9-1GB that are
Version 1. I
> > > > > have a 2GB that is a Version 1. I was wondering if a 512MB
or lower
> > > > > can be a Version 2?
> > > > >
> > > > > Sutton
> > > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > > Start a New Thread