Best utility for formatting SD Memory Card ?

[soonc]

I tried using Windows to format an SD memory card and it left a bunch of boot messages in sector 0.

I also tried "SD Card Formatter" by www.SDCard.org but it also put a bunch of garbage in sector 0.

Does anyone have a link to a formatting utility aimed at formatting SD card for embedded use ?

Or suggest another way to get the boot sector back to normal.

Excuse if this question has been asked before but I could not find anything.

[Ttelmah]

First, how big is the card?.

Then neither routine should put garbage in sector 0. However a lot of things may be left in here, which don't matter.

What are the last two bytes of the sector?. They should be 55 AA. If not then I'd say you have a faulty card.

[soonc]

[Ttelmah]

If you are not using FAT, there is no point in formatting the card....

All formatting does, is partition and put the FAT file-system on the card. There is no low level formatting done on an SD card. No sector boundaries to be initialised, these are in the hardware. If you want to talk to the card as a simple memory device, you don't need FAT drivers, just the basic hardware drivers. 32K per cluster is quite normal, but again you don't need this if you are not using FAT.

Most of the first sector won't actually be initialised at all, since it contains the space designed for the boot sector, and unless the card is bootable this will be left unchanged. This is the first 446 bytes of the first sector.

[soonc]

[temtronic]

Unless you really, really HAVE to use an SD card, you might consider the Vinculum series of flashdrive interfaces.
EASY to use, simple serial interface commands and EASY to transfer data to/from a PC.
I used one for a remote data logging product, having the PIC store the data in CSV (Comma separated values) format. This allowed Excel to import data and display without any 'translations'.

Perhaps this can be an option for you?

There are other 'data logging' devices, though the flash drive system is EASY as every PC has USB these days not all have SD interfaces.

Jay

[Ttelmah]

[soonc]

[Ttelmah]

SLC is still by far the most reliable form of SD.
pSLC, is an attempt to 'appear' to be SLC, using cheaper technology.
SLC is still the preferred technology for 'industrial' rated storage devices.

However both should work the same for simple read/writes.

One thing often missed, is that these later cards take a lot longer to initialise than their simpler brethren. The internal error correction controller takes a significant time to wake up. The SD card specifications talk about units of mSec, but leave the actual time undefined. However some (especially latter cards) can take up to about 400mSec before they have finished their own initialisation....
Second comment, though you talk about 3.3v (good), it is vital to have a high capacity low ESR capacitor close to the card. At nSec timings, there are large current spikes on the supply _particularly when writing_. Your write problems could simply be that the supply is does not have sufficient reservoir capacitance close to the card.
The initialisation done by the CCS driver, is actually correct. The 80 pulse requirement, was removed in the 2007 edition SD specifications, and was replaced with using the IO_SEND_OP_COND command to initialise. This is what CCS do. However the older command is still often more reliable...
The resistors are _mandatory_ (not optional).

I'd be suspicious that either you do not have an adequate reservoir on the supply close to the card, or that you are starting a little too quickly.

[asmallri]

[Ttelmah]

I think the key thing is that whichever driver he is using or intends to use, he needs to get his hardware working first.
The Brush drivers work fine, and I found them easy to modify to use DMA, with plenty of explanation. However all the drivers require that the hardware is able to read and write a sector properly. His erratic behaviour with different chips, says to me that he has a hardware issue. Now SD normally works fine (except with the increasing number of fake cards around), provided you are using 3.3v (which he is), have the correct resistors (he says he has), and have a really clean supply both for the PIC and the SD. A failure to be able to write 'smells' of inadequate reservoir capacitance on the supply, since this is when the demand is highest, and people often don't realise the frequencies involved. The other possibility is ringing in the signal traces. Again good HF layout is vital.

[soonc]

[soonc]

[Ttelmah]

The delay needs to be _before_ you init the card.
Unless your scope has GHz responses, it will not see the micro ripples the SD can generate. This is why there must be a traps for these really close to the card itself.

[temtronic]

This post is a classic 'should have left well enough alone' nightmare. While my 'business' hat sees the save $5 per unit is good, the 'tech' hat sees a LOT of $$ wasted on precious R&D time. That $5,000 'saved' by the new device is 50 hrs to me,so 4 days, more or less, for ONE person.
You should consider buying the more expensive part. It would allow you time to R&D while those 1,000 units get shipped out. I don't know if 1,000 is a weeks sales, month or year BUT using a known good device does get the product out the door, into the client's hands and his cash in yours.
The FIRST thing I do when a client asks for 'cost reduction' is to spend a day doing the 'numbers'. Reading a lot(like part availability((2nd source?)), deciding if a new PCB needs to be made($$), parts pricing($/1,000 vs $/5,000) and then the R&D required( double that time!).
I find it scary that there ARE fake chips out there, buy those and ALL profits can be wiped out.
What I'm saying is , it's better to go with what you KNOW works. When(if !) you have time, then play with new devices to see IF you can save.

Have you kept a log of all the time you have spent on this? Research of the chips, buying,bench testing, posting here ? It ALL should be accounted for and the time spent may suprise you.



Jay