SoftSPI works with cheap card, not with expensive card about sdfat

[augustoborn]

Follow-up:

So I went and added longer delays beween clock edges in SoftSPI::sendBit and receiveBit (originally two nops), and writing to my faster SD card seems to work consistently after I add a (rather ridiculous) 100+ microsecond delay.

Chalk this one up to my setup? I'm on a breadboard, using a 74HC125 to level-shift, maybe the settling times on that are not quite industry standard.

[greiman]

The best way to tune software SPI is with a scope. software SPI can never equal the timing of the SPI controller.

Interrupts for other devices can cause errors. At one point I disabled interrupts while a byte was transferred. Most SD cards are very tolerant of SPI timing so I removed it.

I test software SPI with a variety of cards using the bench example and have not had a problem.

I am working on a major rewrite of SdFat so I will look at software SPI timing again on a scope.