Is the external pull-up on reset really needed?

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

In the ATMEGA162 datasheet, the "DC characteristics" section states there is a 30k to 60k ohm pull-up internal resistor on the reset. After looking through the datasheet and application notes, I could not find it stated anywhere that a 4.7K or 10K ohm external resistor is needed on the reset.

So, is the external resistor really needed and is anyone aware of any designs that work without the external resistor on reset?

Thanks,
Matt M.

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

A "stronger" pull-up might be needed in noisy environments, plus a capacitor.

Leon

Leon Heller G1HSM

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

Will it work without a "proper" reset circuit?

Yes, probably, on the bench/lab under good conditions.

Would we ever release a production design to field conditions without a "proper" reset circuit? No. (the opinion is that the internal pullup is too wimpy for field conditions)

(For our many production AVR designs over the years, we use 4k7 pullup to Vcc, and a 100nF cap to Vcc. Others may put the cap to Gnd.)

Lee

You can put lipstick on a pig, but it is still a pig.

I've never met a pig I didn't like, as long as you have some salt and pepper.

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

Quote:

So, is the external resistor really needed and is anyone aware of any designs that work without the external resistor on reset?

The ones where RSTDISBL is set perhaps? (but they will have BOD fuses set)

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

External noise sources, electrostatic and electromagnetic, induce current into conductors. Even conductors that have one end unconnected. If this current is larger and in the opposite polarity to your pullup, then you get a spurious reset.

The answer is, "Do ya feel lucky punk? Well, do ya?".

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

I don't feel lucky. not.at.all!

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

theusch wrote:

(For our many production AVR designs over the years, we use 4k7 pullup to Vcc, and a 100nF cap to Vcc. Others may put the cap to Gnd.)

As the cap goes to GND in AVR application notes, what is the reasoning behind putting it to VCC?

When power is turned on, the cap has no charge and thus keeps reset and VCC voltages the same. Resistor makes sure the cap does not accumulate charge either. It's almost like connecting reset directly to VCC, except someone can still pull it down (pushbutton or ISP programmer) after overriding the pull-up and capacitance. Who is then making sure the power supply rises fast enough so the AVR can start?

I have always figured the cap is connected to GND because it keeps the reset voltage low until pullup resistor has charged it enough so AVR is left out of reset state after sufficient time when power has risen to safe level. But then again, it might call for a real reset chip like max709.

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

There should be no real need for an external reset chip. The internal brownout and the right startup delay should work in most cases. So there is no need to create an initial reset pulse externally.

If there is not much external circuit and no long line on the reset pin, either a resistor or the capacitor should be enough. If you want to use debugwire there should be no capacitor.

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

Quote:

As the cap goes to GND in AVR application notes, what is the reasoning behind putting it to VCC?

Our grizzled hardware designer has a reason. I can't explain it to you. (Pick a reason: 1) I don't remember; 2) I don't understand; 3) If I tell you I have to shoot you.) He's been laying out micro designs for many years, and continues to do it his way. Perhaps once /RESET high to keep it high during a short noise spike?

Lee

You can put lipstick on a pig, but it is still a pig.

I've never met a pig I didn't like, as long as you have some salt and pepper.

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

Won't help Lee... Cap parallel to resistor isn't going to filter any noise in Vcc. Parallel RC's like that are filtering noise in the Current domain, not Voltage domain.

Writing code is like having sex.... make one little mistake, and you're supporting it for life.

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

Off-topic, but...

dbvanhorn wrote:
The answer is, "Do ya feel lucky punk? Well, do ya?".

A copuple of weeks ago your misquote was listed as the third ranked movie misquote in this list:
Top Ten Lasting Movie (Mis)Quotes http://movies.yahoo.com/feature/...
Quote:
3. "Do you feel lucky, punk?"
Dirty Harry (1971) - The actual line in this movie is far less catchy: "You've got to ask yourself one question: 'Do I feel lucky?' Well, do ya punk?"

The full correct quote is like this: http://www.imdb.com/title/tt0066...
Harry Callahan wrote:
I know what you're thinking. "Did he fire six shots or only five?" Well, to tell you the truth, in all this excitement I kind of lost track myself. But being as this is a .44 Magnum, the most powerful handgun in the world, and would blow your head clean off, you've got to ask yourself one question: Do I feel lucky? Well, do ya, punk?

Maybe thanks to your help this misquote move up to the second spot :wink:
I think I have misquoted this many times myself too before I read that.

Last Edited: Tue. Jun 2, 2009 - 07:26 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Question for folks -

Since the reset input is often used for ISP, what is the feeling about that extra cap? Won't it slow down the assertion and release of the ISP condition? Is this a problem? Or, do you successfully do ISP with an 100nf extra cap in place?

Thanks
Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

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

ka7ehk wrote:
Question for folks -

Since the reset input is often used for ISP, what is the feeling about that extra cap? Won't it slow down the assertion and release of the ISP condition? Is this a problem? Or, do you successfully do ISP with an 100nf extra cap in place?

Thanks
Jim

The cap won't affect assertion of RESET very much, it only affects the release of RESET (WRT to ISP) This is because the ISP programmer pulls the RESET line low, rapidly discharging the cap (no limiting series R). But when releasing RESET, it let's it float high, thus the RC time constant comes into effect. Since programming only relies on the assertion of RESET, there should be no problems with having the cap. (The ISP programmer should not care how long it takes the AVR to come out of RESET) In practice I have never had problems programming an AVR, and having a 100nF cap on RESET.

Writing code is like having sex.... make one little mistake, and you're supporting it for life.

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

By connecting the cap to VCC instead of GND, once running the AVR resets when a short voltage-drop spike occours on the VCC line.
If the cap is connected to GND a short voltage-drop spike wont trigger the reset because the cap has to be discharged through the resistor first.

For noise injected at the reset line the cap will act as a low pass filter regardless if it's connected to VCC or GND, as VCC act as GND for AC signals.

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

right, pretty much the same effect if the cap was not there at all. (I'm not convinced it would increase sensitivity, but it certainly does not reduce it either)

Lee: While your designer may have been doing it for years... my assertion would be that he's been doing it wrong for years. I'm willing to bet that with that cap there, or not, your designs are equally susceptible to resetting due transients on VCC.

Writing code is like having sex.... make one little mistake, and you're supporting it for life.

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

Do you guys also add a ESD protection diode between /RESET and VCC like Atmel recommends alsways to do if you don't use high voltage programming?
All other port pins has ESD protection diodes to both VCC and GND built in, but the ESD diode to VCC is left out on the /RESET pin because otherwise you couldn't pull /RESET above the VCC level for high voltage programming. But high voltage programming is normally not used in-circuit, so a diode can be added externally to ESD protect this pin the same way all the other port pins already are internally.

AVR042: AVR Hardware Design Considerations http://www.atmel.com/dyn/resourc...

Quote:
If not using High Voltage Programming it is recommended to add an ESD protecting diode from RESET to Vcc, since this is not internally provided due to High Voltage Programming. The components should be located physically close to the RESET pin of the AVR. Figure 3-1 shows the recommended circuit on the RESET line.

Attachment(s): 

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

I only add ESD protection to any nets that come into contact with the outside world. As I typically don't have a RESET button, or other external access, I do not add additional ESD protection to the RESET line. I would however add a pair of ESD diodes if I had a panel access RESET button.

Writing code is like having sex.... make one little mistake, and you're supporting it for life.

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

glitch wrote:
Cap parallel to resistor isn't going to filter any noise in Vcc.
AndersAnd wrote:
the AVR resets when a short voltage-drop spike occours on the VCC line
I'm just thinking out loud here, not trying to contradict. I mean no disrespect or offense. So please don't take any.

I don't think the goal at that point is filtering VCC noise nor is it sag protection. There are other parts present for doing those jobs and 100N does nothing for either of those effects. I think the assumption is that VCC is solid and the 100N is a shunt for high freq. energy. That high freq. energy doesn't care if it is shunting to VCC, GND, EARTH, Etc. They all look the same.

So, why put it to VCC instead of GND or EARTH? I don't know. A WAG would be that by putting the cap in parallel with the pull-up ensures a low-impedance impedance match for line effects. Whatever that means.

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

With a cap conneted from /RESET to VCC spikes from VCC is transferred to /RESET, as it takes a while to charge the cap that's has oV across it before the spike. But the AVR has an internal low pass filter on the reset pin so it might not reset even if a short spike is transferred to the /RESET pin.

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

glitch wrote:
I only add ESD protection to any nets that come into contact with the outside world. As I typically don't have a RESET button, or other external access, I do not add additional ESD protection to the RESET line. I would however add a pair of ESD diodes if I had a panel access RESET button.

What about if ISP is used? Then the /RESET pin will have connection to external circuit through the ISP connector.

Why would you add a pair of ESD diodes? The /RESET pin already has an internal ESD diode to GND, but not to VCC. wouldn't it be enough with just one diode?

Last Edited: Tue. Jun 2, 2009 - 08:37 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

AndersAnd wrote:
glitch wrote:
I only add ESD protection to any nets that come into contact with the outside world. As I typically don't have a RESET button, or other external access, I do not add additional ESD protection to the RESET line. I would however add a pair of ESD diodes if I had a panel access RESET button.

What about if ISP is used? Then the /RESET pin will be have connection to external circuit through the ISP connector.

Why would you add a pair of ESD diodes? THe /RESEST pin already has an internal ESD diode to GND, but not to VCC.

I typically don't protect production & debug interfaces, as attachment to them is done in a ESD protected environment.

Writing code is like having sex.... make one little mistake, and you're supporting it for life.

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

Why would you add a pair of ESD diodes? The /RESET pin already has an internal ESD diode to GND, but not to VCC. Wouldn't it be enough with just one diode to VCC? A didode to GND would just sit in parallel with the internal ESD diode.

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

Sure, in this case one, to VCC, would be enough. However, when protecting nets, I prefer to add my own full protection instead of relying on internal circuitry. Also I may have other devices on that same net that do not poses ESD protection on their pins. I also prefer to place my ESD protection as close to the source as possible, meaning right at the connector, or switch.

Writing code is like having sex.... make one little mistake, and you're supporting it for life.

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

Quote:
once running the AVR resets when a short voltage-drop spike occours on the VCC line.
And why would that be? There will be no difference in voltage to trigger anything as the chips' VCC will be the same as the reset voltage thanks to the cap. The diode added as Atmel shows discharges the cap quickly, therefore providing some improvements in the reset behaviour, in a fast VCC drop. In practise chips with BOD do not need it but will not hurt.

Now let's go back to the good ol' days when chips did NOT have BODs. With a cap connectec to VCC rather than ground you would NOT get a reset pulse, the only reason Lee's boards do work, if in fact they are designed the way he says, is because the BOD supplies the reset pulse otherwise one would be in all sorts of unpredictable strifes.

John Samperi

Ampertronics Pty. Ltd.

www.ampertronics.com.au

* Electronic Design * Custom Products * Contract Assembly

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

js wrote:
Quote:
once running the AVR resets when a short voltage-drop spike occours on the VCC line.
And why would that be? There will be no difference in voltage to trigger anything as the chips' VCC will be the same as the reset voltage thanks to the cap.

Yes you are right the /RESET voltage will not be lower than VCC. They both drop to the same voltage.

js wrote:
The diode added as Atmel shows discharges the cap quickly, therefore providing some improvements in the reset behaviour, in a fast VCC drop.

Yes if the cap is connected to GND the diode will discharge the cap quickly if VCC drops. If the cap on the other hand is connected to VCC then the diode will have no effect on discharging the cap. It's already discharged when VCC is powered up. If VCC drops the /RESET side of the cap will drop equally.

This might be the reasoning behind why some connect the cap to VCC instead of GND if they have not attached a protection diode from /RESET to VCC. Because without the diode and the cap conected to GND you could have a higher voltage at /RESET than at VCC when VCC powers down. While powering down, when VCC has dropped from say 5V to 2V there could still be almost 5V on the /RESET pin if the cap has not been discharged quick enough to follow the VCC drop down to zero. An overvoltage at the /RESET pin should however not damage the AVR, since /RESET can handle 12V for high-voltage programming. But maybe the AVR can enter into high-voltage programming mode if VCC drops to a lower voltage than /RESET? Or the /RESET voltage could potentially damage other devices connected to the /RESET line if VCC dropped below /RESET.
At all the other AVR port pins there's ESD protective diodes connected to VCC, these diodes also make sure to discharge any capacitors connected to these pins when VCC drops during power down, so that the AVR and other devices connected to these lines wont be exposed to voltages above VCC that might otherwise damage the devices.

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

Quote:
While powering down, when VCC has dropped from say 5V to 2V there could still be almost 5V on the /RESET
Correct, but it would be for a very short time with a 4K7 (or 10K) discharging the cap. In fact the resitor would try and power VCC off the cap, so the voltage difference could be minimal.

John Samperi

Ampertronics Pty. Ltd.

www.ampertronics.com.au

* Electronic Design * Custom Products * Contract Assembly

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

is it really required to connect a reset circuit with our project circuit to a microcontroller?

is it possible for microcontroller to work without reset circuit?

i am using atmega16

so if anyone can help

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

Depends on what you mean by a 'reset circuit'?

The AVR has a reset circuit built in, however you may need to add protection to the reset pin - be it just a pullup resistor to ensure it doesn't get reset by transients.

The datasheet tells all.

 

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

Aman Tirpude wrote:

is it really required to connect a reset circuit with our project circuit to a microcontroller?

is it possible for microcontroller to work without reset circuit?

i am using atmega16

so if anyone can help

 

So let me get this very clear... you searched the topic and found this extensive thread where the topic has been discussed at length and still find it necessary to ask the question again. Is there a comprehension problem here?

 

Ross McKenzie ValuSoft Melbourne Australia

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

You might look in the spec sheet for "Brownout" and "Watchdog". Those are built into the Mega16.

 

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net