DS2431 wrong answer (0x55) at Copy Scratchpad Command

[bauergeorg]

Hi folks,
we are using OneWireHub and we are very happy about this cool project. Thanks a lot!

Bug
By using the DS2431 with an own server implementation we found a hysteric behavior by writing memory function.
I found a different behavior between the real chip and the emulation. The real chip answers with 0xAA, the emulation answers with 0x55 (or different) during the “Copy Scratchpad” procedure.

Here is an extract of the DS2431s datasheet.

Below you can see a timing diagram of a logic analyzer (Saleae Logic Analyzer) of the real chip (Answer 0xAA):

Below you can see a timing diagram of a logic analyzer (Saleae Logic Analyzer) of the emulated chip (Answer 0x55):

Environment

• We found the line in the code. But we can’t find any error. It looks perfect.
• We used a Arduino Nano and a Arduino Uno. Same devices showed the same behavior (answering 0x55).
• Changing the timing behavior (changing a little bit the code) we saw in some case an answer 0xFF.
• We checked the same behavior with two one-wire-master applications: an own python application and Maxim's OneWireViewer.

Thank's a lot!

Georg

[bauergeorg]

Comparing last commit bring me to change folling code section:

OneWireHub/src/DS2431.cpp

Lines 92 to 126 in 68e9392

	case 0x55: // COPY SCRATCHPAD COMMAND
	if (hub->recv(&data)) return;
	if (data != reinterpret_cast<uint8_t *>(&reg_TA)[0]) break;
	if (hub->recv(&data)) return;
	if (data != reinterpret_cast<uint8_t *>(&reg_TA)[1]) break;
	if (hub->recv(&data)) return;
	if (data != reg_ES) return; // Auth code must match
	if (getPageProtection(uint8_t(reg_TA)))
	break; // stop if page is protected (WriteMemory also checks this)
	if ((reg_ES & REG_ES_PF_MASK) != 0) break; // stop if error occurred earlier
	reg_ES |= REG_ES_AA_MASK; // compare was successful
	reg_TA &= ~uint16_t(SCRATCHPAD_MASK);
	// Write Scratchpad to memory, writing takes about 10ms
	writeMemory(scratchpad, SCRATCHPAD_SIZE,
	reinterpret_cast<uint8_t *>(&reg_TA)[0]); // checks if copy protected
	noInterrupts();
	do {
	hub->clearError();
	hub->sendBit(true); // send passive 1s
	}
	while (hub->getError() == Error::AWAIT_TIMESLOT_TIMEOUT_HIGH); // wait for timeslots
	interrupts();
	while (!hub->send(&ALTERNATING_10))
	; // alternating 1 & 0 after copy is complete
	break;

My solution:

And I will see 0xAA with my own master and Maxims's one-wire-viewer. (The one wire viewer checks only 0xFF and not 0x55 as an wrong answer to copy scratchpad command).