what kind of capacitor to use on reset pin?

What criteria did you use to decide between a mylar reset cap and a polypropalene or polystyrene cap? Once you answer that, you can likely answer your own posted question.

If the uP has a schmitt type input, then any cap type will work. But some processors need a clean, fast edge, so a uP reset chip is the best way to start them up.

The other problem with a simple RC is that you may be able to tease the power supply (apply various brownout conditions) and hang things up or crash the code. A good reset chip is immune from being fooled like that.

Which uP are you using?

We just did a new product that wouldn't start up. Seems we went from a

John

Why not?

John

"Larry Brasfield" wrote in news:PqWxe.36$ snipped-for-privacy@news.uswest.net:

To be entirely honest - I was just told "film capacitor" - so I found a film capacitor of the right value that looked like the one in the picture - a small yellow box - and got it. I've been using the same part ever since.

I should mention that the reason for my question is this: I'm converting a circuit I have breadboarded to an entirely surface mount board - and I can't find a film capacitor in a 0805 package of the right value (Digi-Key doesn't carry any at least). There were plenty of ceramics of the right package and value though - so I'm wondering if it would be OK to use one of those. Instinct tells me yes - but it also tells me I should ask to be sure before I do something silly.

-Michael

Either ceramic or mylar will work not electrolytic

Dan

--

Dan Hollands

The only engineering consideration in this case would be the capacitor leakage current. It must be negligibly small so as not to induce a voltage drop across the 100k and thereby prevent complete pull-up. Film and ceramics are good, electrolytics go from risky to bad.

Sorry, but that's just plain silly.

John

A reset capacitor is a non-critical time delay application. You don't care if the reset pulse is 5% longer or shorter on a warm day, you're not concerned with series resistance or high-frequencies. You can select a capacitor for its physical size and price.

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

I'm using an Atmel ATMEGA162. (an AVR)

It does have built in brownout protection I believe, though the supply should be fairly constant so I don't think that will be an issue.

Hello Michael,

Most of the uC circuits I came across use a simple ceramic SMT cap. But I am always suspicious when it comes to such simple resets. Unless the uC has a proper (and that's easier said than done) reset function or better yet a brown-out reset it may be better to build you own. Or use a reset chip but in that case you'd have to trust that the designers of that chip knew what they were doing. They usually do but not always.

Regards, Joerg

MAX809, and about a million clones from TI and National and such. The low-end parts cost $0.19 or such.

John

Even the crappiest aluminum electrolytic series should be able to handle a 100msec time constant with typical levels.

BUT- I strongly advise against using RC circuits for reset in all but the most non-critical AND (not OR) cost-sensitive applications, for reasons that have already been mentioned. An open-collector level detector (a few discretes) and an RC will be reliable in many cases, but the reset chip is smaller, will probably use less power, and maybe even will be cheaper.

Best regards, Spehro Pefhany

snipped-for-privacy@nowhere.net wrote in news: snipped-for-privacy@4ax.com:

Can you reccomend such a chip? I've never used anything except for an RC circuit on the reset pin. Thanks,

-Michael

On Sun, 03 Jul 2005 19:54:50 GMT, Fred Bloggs wroth:

Of course, one could always *engineer* a solution. I know that's a novel concept here, but it's true.

The micro's data sheet will specify the voltage levels that constitute valid ones and zeros. They are seldom Vcc and ground.

They also specify the minimum time that the reset level must be asserted.

Once you have that data, you can choose an RC product that meets the time and choose a maximum capacitor leakage and minimum capacitor value over the expected temperature range.

Show me the calculations that say that almost any tantalum capacitor can not be used.

Jim

On Sun, 03 Jul 2005 13:27:07 -0700, John Larkin wroth:

Not to mention that a SOT-23 reset chip will probably be smaller than the RC network.

Jim

the

You have to think about the frequency response of the caps otherwise high frequency glitches will give you unwanted resets. A crappy electrolytic won't handle as high a frequency as the uP will respond to. I agree, go with a reset chip. It should help get the power supply rejection you need.

Paul C

It's just like decoupling caps. Above a certain frequency the lead inductance dominates and the cap acts inductively. That's why they add those 0.1uF or 0.01uF film caps in parallel with the electrolytics in power supplies. I've seen it happen. Evey time the relay clicks the microprocessor resets because of some high frequency spike and I had to go back and add filtering to kill the spike.

Paul C

Okay, so say we've got a 1uF electrolytic in series with a 100K resistor to Vdd and maybe 5pF of capacitance from the input.

In what frequency range(s) do you think that the impedance of the e-cap becomes significant compared to the 100K?

Best regards, Spehro Pefhany

You really have to look carefully at how well the internal reset circuitry actually works. Some AVR chips are (in)famous for EEPROM corruption during power cycling, and some PICs with BOR seem to occasionally get themselves into a state that can't be reset without power cycling. It was clearly stated in data sheet in the former case that the BOR tolerance is outside the guaranteed range of operation of the chip, so you'd only have yourself to blame for trusting the big print.

Best regards, Spehro Pefhany

Any type of cap will work because there are no constraints. The rc constant is not critical and polarity is always positive.

Harold