That new Atmel Tiny4

Especially true with this part. I don't recall _ever_ seeing that high of current requirement (at voltages I can compare across.) I think the PIC12's BOR is half that, or less.

Would that be Microchip's XLP variety? (I haven't looked, but I remember reading way-too-good-to-be-true figures for the required current -- in tens of nanoamp range.)

A lot of systems do little other than use the 'free' watchdog that comes with them. It's a little nicer to have something dedicated to the purpose on the outside of all that. And if you don't trust that or feel that cosmic rays might yield a false result, stack them four deep with one having one pin 'lifted' so it knows it is the monitoring one and use it to generate the voting result... maybe? ;)

Jon

They can achieve 20 nA, but the chip isn't actually much use for many applications in that mode.

Leon

At what current level do they become useful?

Jon

s.

They retain the RAM contents at 25 nA, typically, which is some use. RAM contents are lost at 20 nA.

They take 500 nA with the RTC and calendar operational.

Leon

And that's with the BOR operating, I gather. More than acceptable.

Now if I can get that in SOT23-6.

Jon

is

rned

ipes.

I've got some of the 16-bit XLP chips, but haven't tried them yet.

Leon

When you are in sleep , you do not neccessarily have to have the Brown-Out detector turned on all the time.

The Brownout detector is mainly there to block the CPU from executing those instructions that will mess up the chip, like write to EEPROM or Flash.

The behaviour of the part during brownout is depending on current draw, If you are in sleep, very little current is going through the device.

I am told that the AVR, in sleep mode, will have a RAM retention/ D-Flip-Flop retention voltage of around 0.4V. The Power On Reset of the AVR will trigger around 1.0V so you *have* something watching your watchdog.

Jon Kirwan skrev:

Brown-Out detectors are really simple analog stuff. You have a trade-off between speed and current. A low current Brownout detector will be sloooow. A slow BOD will not trigger until after you have a chance to start overwriting your EEPROM.

If you want something to react below 10 us,then you are going to see 10s of uA.

The AVR sleeping BOD combined with the 1.0V power on reset will allow you, in the PicoPower devices, to reach 0,6 uA with system level BOD as well as 32 kHz oscillator running.

However, that's not Tiny4 :) Looking at the SiLabs devices, they claim 50nA for LowBat, but digging deeper shows that's just for a 1V level, and the Vcc monitor is 7uA.

Separate Vcc Detectors, are under 1uA

Problem with the vanilla 1V 'BOD' trip points, is they hit a black hole in the datasheet, as Clock operation is usually completely unspecified there.

Try it on some real devices - can be educational...

Still leaves the problem of who is watching your watchdog ?

-jg