Low power quadrature decoder.

I need a low power quadrature decoder with following features

  1. Average current less than 10uA
  2. Supply voltage 2.0V to 3.6V
  3. Max pulse frequency 4kHz. Since there are 2 inputs 90 degree apart that means the pulses could be at 16khz actually
  4. Optional I2C or SPI interface
  5. 32bit pulse count with direction

The closest I can find is LS7366 chip but it does not work with low voltages and is a hard to find part.
So, I am wondering whether it can be implemented with an RTC DS1307/PCF8583 chip (actually 2 chips) which can do pulse counting. These chips meet all 5 criteria above.
Is it technically feasible to implement rotation count with direction sensing with some additional hardware logic?

Thanks,
WonderfulIOT

ATtiny and a little programming may work. The 441 has a working voltage 1.7V up; it can maintain a clock speed of 4 MHz at 2V which is enough for counting the pulses and handle I2C at 100 kHz. Only problem is power use, but you can bring this down by letting it sleep most of the time (1.3 µA) using the pulse interrupt and I2C interrupts as wake up signal.

Counting a single pulse should take something like 20-30 clock pulses, so 5-7.5 µs at 4 MHz, and back to sleep. At 4 kHz that’d be 20-30 ms per second that the processor is active. So as long as the active power use is <290 µA you’re within 10 µA. For lower speeds (or no pulses) the average power consumption goes down.

Thanks for the idea.

However, I am not sure whether the mcu is fast due to the following overheads

  1. wake up
  2. enter ISR
  3. Leave ISR

As per the article https://gammon.com.au/interrupts, and am quoting it here

"I count 82 cycles there (5.125 µS in total at 16 MHz) as overhead plus whatever is actually done in the supplied interrupt routine. That is, 2.9375 µS before entering your interrupt handler, and another 2.1875 µS after it returns."

It translates to an overhead of 20.5 µS at 4Mhz.
That article is for atmega328p, would it be similar for Attiny441?

Didn’t count the wake up time indeed. I suppose it’s very similar, if not exactly the same in both chips.