Electrical Properties of an AVR switched off....?

Go To Last Post
10 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I'm working on a schematic for research at school.. I am being paid to do this, this is a learning experience for me and they realize this.

It's quite the project for me, more complicated than things I've done in the past because it involves multiple systems, multiple power-rails, and power switching. It's for a battery operated robot. The main electronics will run at 3.3v, with 5v for the main computer and 5v for RC servos...etc.. etc..

One requirement of this board is soft switching and battery gas gauging. So the gas gauge IC (non AVR), will be powered all the time, a second AVR (like AtTiny13 for instance) will also be powered all the time, living mostly in sleep mode and brought to life by interrupt when a switch (or other) is closed. This IC will be able to turn on the rest of the robot.

The servos will also be shut on or off in software. I've been planning on doing this via low-side NMOS, and I think that will be fine.. I didn't have any problem w/this on the bench anyways. Keep in mind, these are digital servos.

So, with all that explanation, here's my dilemma. ICs, AVRs, whatever are largely black boxes (at least in my mind), especially concerning their behavior when powered down, and in what way powered down.

My problem is, if I have parts of a system that are powered down (either by low-side or high-side switch) and parts that are never powered down, how can I be sure that the powered down parts won't be damaged, given that these parts are connected together via I/O pins or I/O buses.

Like, what could I expect to happen if a HIGH or LOW appeared on an AVR's I/O pin given that it's ground pins were Hi-Z? Or, what could I expect w/a HIGH or LOW on an I/O pin given that it's Vcc was Hi-Z.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Remember the there are 2 diodes at each I\O pin to make your life miserable in this case.

So draw those diodes in at each I\O pin and follow the current path.

hmmm robots with gas problems, may need some medications... :?

John Samperi

Ampertronics Pty. Ltd.

www.ampertronics.com.au

* Electronic Design * Custom Products * Contract Assembly

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

That's what I've been doing, and I don't see any problem w/clamp diodes when using low-side switching. e.g. 3.3v logic, with ground Hi-Z and

But, obviously clamp diodes are just the tip of the iceberg.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You could use tri-stated buffers like the 74HC125 (quad) or 74HC365 (hex) on the concerned IO's, connected in such a way that the lines are effectively disconnected when the module is powered down.

Last Edited: Thu. Sep 17, 2009 - 09:55 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I posted an article recently about using a tiny13 as an RFID tag. The guy simply connected a 100mH (uH?) inductor to the clk pins and was able to power and clock the uC via the transmitted power overloading the clamping diodes. So, it seem to me that an AVR might power on if a high(ish) voltage is placed on one of the I/O pins.

There is likely a threshold value, but you might want make a simple blinky program and see what happens in different configurations (high disconnected, ground paths from other pins, etc.) and help us out by posting your findings here.

"A common mistake that people make when trying to design something completely foolproof is to underestimate the ingenuity of complete fools."
-- Douglas Adams

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

With some micros, and other chips, it is certainly possible to partially power them via current drawn from their signal lines, while their V+ is "disconnected". This typically gives very unpredictable and erratic behavior of the improperly powered device.

JC

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The AVRs and most other chips (e.g. CMOS logic) want always, that the VCC pin see the power at first.

Powering an signal pin with VCC = 0V may cause ignition of a parasitic thyristor inside and then burning the chip.

So I would suggest, let the AVR always connected to VCC and switch the AVR into the power down mode to save power.
In power down the AVR need only below 1µA.
The battery self discharging would be higher.

Peter

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

danni wrote:
So I would suggest, let the AVR always connected to VCC and switch the AVR into the power down mode to save power.
In power down the AVR need only below 1µA.
The battery self discharging would be higher.
I am glad somebody has pointed out what is surely the obvious. Lots of the AVR range is designed to have low or very low power consumption in certain special sleep/power-down modes whilst VCC and GND are still connected. So just use the chip as it has been designed to be used.

Christopher Hicks
==

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

danni, that's true. But it's not as simple as that.

This is a 12v system, feeding a 5v SMPS, which will likely go to 3.3v LDO for AVR/3.3v stuff (SMPS is possible, but prob no necessary)..

It's not that the small current consumption of the AVR powered down is a problem, it's that I need to shut everything else off too.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

..And, even if I used an efficient low-current 3.3v SMPS to power the AVR only, that solves no problems because the stuff it connects to (powered by 5V SMPS) will be off.

What do you guys think of transmission gates as a solution for bus isolation? I2C, UARTs, I/O etc..