52810 QDEC wake-up

[bevis]

Hi:

I am using a shake sensor as an analog input wake-up system. The system is not system off. The analog input signal only generates an event to wake up turn on mode. Before I used 52832 to wake up with LPCOMP, now I use 52810, only COMP and QDEC. Bluetooth beacon of a competing company also uses the same shake sensor. When the sensor is connected to GND in static state, GPIO leakage is almost zero, and the power consumption of the whole machine is only 3ua. But 52810 also uses QDEC. When the sensor is connected to GND in static state, the leakage current reaches 400ua，

Moreover, QDEC triggers are far less sensitive than competitors. I tried to adjust different internal reference voltage, but the result is not very satisfactory. According to the data, QDEC is a double input signal comparison, but the competition can also achieve high sensitivity input detection only by single input. I think 52810 should also be able to do so. Thank you very much.

The following are the various possible input characteristics of the shake sensor. Because the shake sensor is directly connected to GND, the high-level state can only be determined by the GPIO state of the analog input (QDEC).

[haakonsh]

Why are you using the Quadrature DECoder to detect this signal and not LPCOMP, COMP, GPIO, or GPIOTE ?

When the sensor is connected to GND in static state, GPIO leakage is almost zero, and the power consumption of the whole machine is only 3ua. But 52810 also uses QDEC. When the sensor is connected to GND in static state, the leakage current reaches 400ua，

 I need a schematic in order to understand your problem, as well as some more information about your sensors, preferably a datasheet. 

 

Moreover, QDEC triggers are far less sensitive than competitors. I tried to adjust different internal reference voltage, but the result is not very satisfactory. According to the data, QDEC is a double input signal comparison, but the competition can also achieve high sensitivity input detection only by single input. I think 52810 should also be able to do so. Thank you very much.

 A QDEC is used to decode two digital quadrature-encoded inputs, you're using it way outside of it's intended operation. The fact that another MCU's QDEC performs better in your use-case is really arbitrary and it does not really reflect the performance of a QDEC with respect to its intended use-case.


Can you share some more information about your sensor?

[bevis]

After shake, 98% of them are in OFF and 2% are in ON.The competitive schematic is also shown in the following figure. I've described the characteristics of the sensor very clearly, because for GND directly, an edge pulse will be generated when the vibration occurs, and the voltage source of the pulse can only be the analog input level in GPIO. Basically, the principle of sensor is that when shake, the internal resistance of sensor will be from 0 - > infinite, or infinite - > 0 ohm.

[haakonsh]

That circuit requires a voltage source, how else can you measure the resistance?

And what do you mean by "After shake, 98% of them are in OFF and 2% are in ON.", are 2% of the sensors somehow different from the rest of the 98%, and if so are both OFF and ON states valid after a shake?

Do you have a datasheet for the sensors?

[bevis]

After the vibration is completed, 98% of the sensors are in closed state, which is like a button close direct GND. 2% of the sensors are in open state, which is like a button open. The datasheet is in Chinese. Actually, what I have described is very clear. That's what the datasheet says.

[bevis]

I measured it with a parallel multimeter.

[haakonsh]

Okay, I think I've understood now.

You will still need to connect one side to a voltage source, and you'll want to add some series resistance in order to reduce the current consumption in the "closed" state.

The QDEC needs a fairly clean square-wave input, so you'll want to add a small RC filter to reduce the noise.