BOD level near operating voltage

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

So I haven't really done anything using a proper battery operated design, thus I'm wildly ignorant in regard to the proper use of the BOD.  The question I have for you guys is how close can the operating voltage be to the BOD level without risking extraneous resets?  I'm currently playing with an attiny13a, and I see that the 1.8V BOD level ranges from 1.7-2.0V between chips, and there's 50mV of hysteresis.  Would that mean I'm fine operating at 2.1V using a plain old linear regulator, or would I be asking for pain and suffering?  How much wiggle room should I allow myself?  Might it be possible to go as low as 2.0V if I can guarantee the voltage is allowed to rise past 2.05V when the MCU is in reset, or would that just be madness?  I realize this may be splitting hairs to try to go that low, but I figure as long as I'm already asking the question, let's go nuts!

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

Rezer wrote:

... I see that the 1.8V BOD level ranges from 1.7-2.0V between chips, 

 

  Would that mean I'm fine operating at 2.1V using a plain old linear regulator

 

Is the 2.1V from the regulator the absolute minimum? (Don't forget output variation over temperature range.)

 

Personally, I like to have at least 0.3 volts between the BOD maximum trigger voltage (2.0 volts in your case) and the regulator output minimum voltage (2.1 volts in your case?).

 

So, for the example you gave, I'd say it was too close. 

Remember, there will be noise on the V+ supply line due to switching currents and other noise sources.

 

 

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

That's kind of what I had figured, it's good to get some sort of idea how big that margin of error should be.  Is it common to actually operate that close to the BOD threshold, or would that be more of an edge case?

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

Rezer wrote:

Is it common to actually operate that close to the BOD threshold, or would that be more of an edge case?

 

Here are the Brown-out detection levels from the ATtiny13A datasheet:

 

As I understand it, the listed Typ values correspond to standard Vcc voltages -

    1.8 for 2.5V

    2.7 for 3.3V

    4.3 for 5.0V

 

The Vcc voltages are nominal and typically have a tolerance of +/- 5%, so:

    2.5V - 5% = 2.375  --> 2.375 - 2.0 = 0.375 volt margin.

    3.3V - 5% = 3.135  --> 3.135 - 2.9 = 0.235 volt margin.

    5.0V - 5% = 4.750  --> 4.750 - 4.5 = 0.250 volt margin.

 

From the above numbers, it looks like 0.235 volts is the lowest worst case margin (comparing maximum brown-out threshold and minimum Vcc voltage).

 

 

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

From the above numbers, it looks like 0.235 volts is the lowest worst case margin

I am using BOD 2.7V at Vcc 5V.

There were frequent resets when I tried BOD 4.3V. (Atmega88, 100 microFarad on Vcc)

I think it is reasonable to use as low BOD as possible.

 

I am not sure if the difference of 0.3V is enough.

 

 

Last Edited: Sat. Jul 9, 2016 - 05:55 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Visovian wrote:

From the above numbers, it looks like 0.235 volts is the lowest worst case margin

I am using BOD 2.7V at Vcc 5V.

There were frequent resets when I tried BOD 4.3V. (Atmega88, 100 microFarad on Vcc)

I think it is reasonable to use as low BOD as possible.

 

I am not sure if the difference of 0.3V is enough.

 

 

 

To use BOD 2.7V you can't use a cpu clock above 8MHz.

 

Does this mean that one can't use the BOD if running at 16MHz?

Why did Atmel provide the 4.3V BOD?

 

I'm not disputing your experience, just curious.

What type of peripherals were you using?

 

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

That...makes a lot of sense, thanks. I hadn't considered that each level has a specific "target" voltage, in no small part because the datasheet never mentions it.

Here's a related question though:. If the BOD is problematic below around 2.4V, how does one safely operate at lower voltages? Of course it's speced to run all the way to 1.8V, but is there any safe way to actually run near that voltage if the BOD can't be trusted? Or do you just have to accept that eeprom will occasionally be corrupted and SRAM will get clobbered as a trade off for operating so close to the limits of the MCU?

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

Rezer wrote:

That...makes a lot of sense, thanks. I hadn't considered that each level has a specific "target" voltage, in no small part because the datasheet never mentions it.

Here's a related question though:. If the BOD is problematic below around 2.4V, how does one safely operate at lower voltages? Of course it's speced to run all the way to 1.8V, but is there any safe way to actually run near that voltage if the BOD can't be trusted? Or do you just have to accept that eeprom will occasionally be corrupted and SRAM will get clobbered as a trade off for operating so close to the limits of the MCU?

 

The 5.0V, 3.3V, and 2.5V values were just my interpretation - not from the datasheet or any apnote.

 

To safely operate at lower voltages, my guess is that you need to disable the internal BOD and if BOD is necessary, provide an external BOD circuit.

 

Where did your specification of 2.1 volts come from?

 

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

It's not a spec by any means, I just have little need for more than a couple mhz and I noted current falls linearly with operating voltage, so I want to know what I can "get away with".

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

Not necessarily a solution for you, but I picked up a few thousand Seiko S-80128 rail monitors in SOT23 package for less than a tenth of a cent each. 2.8 volts +/- 2%, 1.3uA quiescent.

Ross McKenzie ValuSoft Melbourne Australia

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

Oh neat, I didn't know such a thing exists at such a low quiescent current. If I'm not mistaken, I could put something like that before the regulator and it'll catch when the battery current is really getting low, without having to worry about false triggers from noise by trying to operate right on the edge of the bod threshold?

Btw, I searched for that exact chip you mentioned and it seems to be around 25 cents in quantities of 3000. How'd you get the 99.6% discount?