SD read/write error

njepsen · Joined: 13 Aug 2007 Posts: 392

I am saving backup data to an SD card, but from time to time - like once a month- I get a situation where I cannot read the card.
Here is an excerpt of the code. Full code is 8000 lines.
Err 241 is "cpErrDriveRead1" and I have no idea what that means, or more importantly, why it is happening or how to recover from it.
I don't have the hardware to cycle the SD card 3.3 power supply, which has its own regulator and 2 bypass caps a 0.1 and a 1000uF, close to the sd socket. I'm using Toshiba micro SD.

AdrianJ · Joined: 16 Jan 2006 Posts: 2460 Location: Queensland

Reaching back into dark history...

I too had rare problems getting SD cards to respond. Maybe not the same failure, but effectively preventing access. My solution at the time was to use a card supply regulator which could be enabled and disabled by a processor port pin.

Never had any problems after that, but I have no idea whether the recovery cycle was ever tripped in the field. What one could have done was to log a failure to the card, after getting access back, but I never thought of that, and in any case I only very rarely had access to the cards once installed.

njepsen · Joined: 13 Aug 2007 Posts: 392

Thanks Adrian,
Unfortunately in the latest population of boards, i cannot cycle the SD card power. On older boards I had a FET that allowed me to do this but (foolishly) decided it was not necessary. Because this issue is so intermittent, its hard to fault find, and I was hoping someone would spot an issue with my code. SD cards for me are a bit of a black art.

AdrianJ · Joined: 16 Jan 2006 Posts: 2460 Location: Queensland

Hi Neil,
My memory of all the steps needed to get an SD card going are long gone. And in any case I used them in simple SPI mode, so all those advanced access settings are different.

The repeated error 241 at initfilesystem is a bit odd, you should not get that, but I dont know if its significant.

Somehow I doubt really rare errors like you get are due to code, except possibly a marginal timing issue.

I would be more suspicious of power supply glitches. You do know that the cards themselves draw high pulse currents during the write cycles, up in the several 100 mA region ? That can cause problems with supplies. Even 1000 uF bypass may not be enough, when you work out the voltage drop over the current pulse time. You need very "stiff" regulators. FETS to switch the output can cause the same problems, its safer to switch the input side, if you go back to that idea.