Slow VDD rise in a STM32F051x

Spending half a buck on a POR chip can prevent a lot of midnight phone calls.

Cheers

Phil Hobbs

replacing a regulator with one that has power good, might even be "free"

-Lasse

As one who has been burned... I test my designs for all kinds of power supply sequencing... some slow power-ups have latch-up problems. ...Jim Thompson

A good chip should be able to handle any voltage from 0 to the rated voltage. It isn't that hard to design. Assuming a CMOS design, you insure you have a supply rail of at least a VTP plus VTN. You use a simple timer to make sure the voltage has been at least this high for a while. From there, you can use a bandgap if you need a precise turn on voltage. Or a Schimdt trigger, etc.

Slightly harder is a design that assures you can short any pin to a power supply or ground. Often these chips will have a watchdog timer in them.

One test you need to do is to suddenly slam the power supply to ground, then ramp the supply back to the rated voltage. You need to check for charge storage in the chip. Good designs will have added parasitic diodes to the positive rail to insure the internal timing caps are discharged.

• Do not see that; read 3.51, page 13 thru 3.54 page 14.

A simple RC diode scheme works for the PICs.

Page 13 of the datasheet says the POR monitors VDD, maintaining reset until VDD > 2V, thus should work for any VDD ramp-up. ...Jim Thompson

I saw that. I was mostly wondering if anyone had found any counter- examples.

So far for me, if ST says it works then it works, but there's always a first time.

What it means is (and I've done this on chips myself) that the POR is essentially a bandgap and when it goes active (at 2V) reset is released. So it's robust. For instance, a recent chip design of mine released reset at +2.7V, and reapplied reset at +2.3V (hysteresis). ...Jim Thompson

"Jim Thompson" wrote in message news: snipped-for-privacy@4ax.com...

Joerg has mentioned before that ATmegas have crappy POR / BOD circuits, any thoughts?

Seems like it would be pretty obvious (put in a comparator and be done, as mentioned), but then, their onboard voltage references were pretty atrocious, too.

Tim

Naaaah! It's pretty easy to do references accurate to +/-10mV, and you don't just "put in a comparator and be done", you use the actual start-up mechanism of the bandgap to give you a clean/sharp POR. ...Jim Thompson

There are "bandgap comparators" and bandgaps with a comparator, with the first item being a bit better integrated, and the later usually more accurate because you can put more fu in the bandgap design. Since a bandgap needs time to settle, there is usually a crude comparator to insure the bandgap is at least above the VTN of a weak nfet.

Since the power supply could sag when the chips turn on, this does get tricky. What you want to avoid is a scenario where the whole thing starts to oscillate. Generally a timer after the comparator helps the situation, so that the power supply keeps rising before the chip turns on. Then if the supply sags a bit, you hope it won't sag to the point where the comparator trips off. The comparator will have a little hysteresis.

A well designed POR is pretty bullet proof.

Plus some time delay, usually. Really the 20 cent POR chips (e.g. the MAX809) are a good investment.

Cheers

Phil Hobbs

Easy enough to add, if you want. I did a chip a few years ago where the POR started an oscillator and a counter to give a 100ms delay ;-)

(Settling time for analog video processing.) Many of my analog designs need power sequencing control... particularly those with split supplies... otherwise you can get flame :-)

From whence need came my behavioral simulation parts MOSTPOS and MOSTNEG ("Bounding_Circuits.zip" on the Device Models & Subcircuits Page of my website), to simulate power supply sequencing protection. ...Jim Thompson

te

, as

ST also makes PORs so it should be possible for them to stick on the corner of their MCUs, but you never know

-Lasse

Second sourced by On semi and AD (ADM809). On is currently the cheapest at 8-9 cents each. They're bulletproof,, cheap and good.

Best regards, Spehro Pefhany

Yeah, I'd never buy them from Maxim. ;)

Cheers

Phil Hobbs

rote

is

ine

ts,

ne, as

isn't the question; would you be able to?

;)

-Lasse

I have used precisely this chip. We had a slog rising supply also, never saw any problems.

I actually did current versus voltage measurements (to have better current consumption data), never discovered any gotchas there either

Cheers

Klaus

If you dont mind the long delay they also work for cheap simple 3 pin switch debouncers. I know there are a boat load of ways to debounce but if I'm already dropping one on the board for uC reset using another one is not a big issue. I'm not Joerg cost constrained.