what kind of capacitor to use on reset pin?

Sure that's another solution if you need such a regulator and can find a fit that doesn't cost a fortune. National has a bunch. But you generally want the type with a timer as well as a comparator and reference to avoid the need for the RC network. You want time for the oscillator/PLL etc. to stabilize before removing the reset both on power up and in any other case where Vdd has gone out of spec. Some micros may provide this function.

Best regards, Spehro Pefhany

It seems to (now) be properly specified. See doc2513.pdf PDF page 49 Note 1. They say they test the micro operation at a voltage below the (otherwise) specified range to guarantee reset if the BOD level falls below the minimum voltage. *Presumably* (????) they add a factor to allow for BOR and minimum operating voltage changes with temperature at the *maximum* clock frequency.

Minimum reset time of a couple of us could conceivably be a problem since it's only set by hysteresis. If there's a lot of supply current available it could charge the capacitance on the supply rails up in less than that time if power is interrupted then restored just after falling below Vbot.

Best regards, Spehro Pefhany

the

The original question dealt with a 10ms time constant- that is 69% of Vdd if you recall. Why don't you show us how to calculate the minimum power supply turn-on time required to reliably reset the uc when the minimum reset duration is say T- that is duration of time for voltage across C to be less than 0.2Vdd for Vdd > Vdd,min. Obviously, there will be no reset when the voltage across C essentially follows the instantaneous Vdd.

Hmmm- tons of LDOs with logic compatible dropout indicator signals available these days...

You don't need an external supervisory chip- the ATMEGA162 contains an internal power-on reset, an internal programmable brown-out protection reference and comparator, and the external /reset. The external /reset is most often used to bring the chip out of sleep mode, although it can also be used as a general hardware reset.

The ATMEGA162 BOD is based on an internal 1.1V bandgap reference and comparator. The datasheet is screwed- my download is missing page 46 and some other critical stuff relating to reset times. Oh well- "most of it" is there and that is good'nuf for the programming world.

See that's just it- the e-cap leakage lowered his noise margin with 20uA leakage current..../reset was hanging near Vdd/2:-)

That's what's nice about the open drain /LBO type outputs- couple with an RC on /reset and you slow on/ fast off. This is all battery application stuff anyway- so the Vdd holds up well above the minimum operating voltage after /LBO has gone low. The RC takes over on power up. I don't really see where the external reset chip should have any advantage over the exact same circuit internal to the AVR- both should behave in exactly the same way.

Could have been that the smaller physical size of the tantalum was coupling less of the interference to the reset input. EMI can be a PITA. An 0402 ceramic sitting right over a ground plane is nice.

Best regards, Spehro Pefhany

John Larkin wrote in news: snipped-for-privacy@4ax.com:

Hi - if I'm reading all the datsheets properly (for the MAX809 and for the ATMEGA162, the uC I'm using) - the MAX809L is the chip that I want, as it's threshhold voltage is 4.63V, and the ATMEGA162 can't be run at 16Mhz (the speed I'm running it at, at anything below 4.5V, according to page 265 of the datasheet.

But what I'm confused about is the extrnal resistor that is drawn in. On page 6 of the MAX809 datasheet it says that this resistor is needed for high impedance CMOS inputs - is that what the ATMEGA162 RESET' pin is? It says that's only needed for if the supply voltage dips below 1V, which I suppose should never happen (the AVR won't even run at that low of a voltage anyways).

But is all this even necessary? The AVR has built in brown out protection at three voltages - 4.3, 2.7, and 1.8.

Thanks for the all the help,

-Michael

I see what you're saying. I'd need a pretty big ESL to cause a problem. Still I had a spike coupling in from 12V circuit into a 5 volt circuit I thought I had isolated and one of the changes I made was to change a 10uF electrolytic to 1uF tantalum in the reset circuit. It worked.

Paul C

I've used electrolytics for decades on 8051 family designs. As recommended in fact.

Your distaste for them is unscientific I suspect.

More important is whether you include a diode to avoid the charge pump effect on power-down that can over-volt the reset pin. I note that a few datasheets actually mention the issue.

Graham

Electrolytics may have been that bad 30-40 yrs ago. Today it just isn't true.

Graham