Power on Reset

(relays)

In my first "biggy" semiconductor amplifier, 30W, circa 1962, I used a mechanically-delayed (?) Mercury-wetted relay to avoid the thump.

(?) Some kind of dashpot arrangement. ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     | 
| Analog Innovations, Inc.                         |     et      | 
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    | 
| Phoenix, Arizona  85048    Skype: Contacts Only  |             | 
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  | 
| E-mail Icon at http://www.analog-innovations.com |    1962     | 
              
I love to cook with wine.     Sometimes I even put it in the food.

Add the fact that the equipment had been on 24/7/365.25 for over five years to the crap reset circuit, and it was in bad shape.

Some used the old Amperite thermal time delay relays that were about the size of a 6AQ5 or 50C5.

But it will have a lot more flexibility in its next term.

That they *did* dig down to find the cause of the problem is why they know what is needed to assure the reset works for the "rest of the time" cases and not just "most of the time". I have also been lucky in my designs so that when I used a simple reset, usually a diode to help discharge the cap in addition to the RC, it worked well enough for that situation. But then I learned about the various failure modes from the experience of others and got to see other people's simple reset circuits fail in designs. I don't need to make the same mistakes I see others make.

Yes, if you are running from batteries a simple reset circuit might work just fine. But with a reset device not much more than the price of a transistor, unless you really are counting every penny, why skimp?

Exactly, what ever!

Rick

I need the fertilizer for my yard. :)

Poor quotation, it's "... and the pig enjoys it." ...Jim Thompson

-- | James E.Thompson, CTO | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona 85048 Skype: Contacts Only | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at

| 1962 | I love to cook with wine. Sometimes I even put it in the food.

Only conservative pigs. ;-)

Not as long as you are only working on the one "approved" design!

BTW, that's "1984", not socialism.

Rick

What's overkill about a single SOT23 reset chip that costs 10 cents or so? Your mosfet circuit is more expensive that that, and has hazards.

The dedicated reset chips are precise. You buy them to shut down at a specific voltage. They give you a solid reset for a couple hundred milliseconds after the supply exceeds that threshold. They handle brownouts right.

CMOS schmitts track; bipolars don't.

I'm not sure we are talking about the same thing. By track, do you mean

1:1? Can you give me a reference on this?

My understanding is that hysteresis is done with positive feedback. When the input crosses the threshold a small portion of the resulting signal change is fed back to push the input in the same direction. The amount of feedback determines the amount of hysteresis. The same mechanism results in the amount of hysteresis depending on the supply voltage and so the threshold. But it won't be a 1:1 change or the hysteresis would be the full Vcc/Vdd. No?

If I fail to understand the mechanism, please explain it to me.

Rick

I'm not sure we are talking about the same thing. By track, do you mean

1:1? Can you give me a reference on this?

My understanding is that hysteresis is done with positive feedback. When the input crosses the threshold a small portion of the resulting signal change is fed back to push the input in the same direction. The amount of feedback determines the amount of hysteresis. The same mechanism results in the amount of hysteresis depending on the supply voltage and so the threshold. But it won't be a 1:1 change or the hysteresis would be the full Vcc/Vdd. No?

Perhaps you are talking about the basic threshold of CMOS tracking half the power supply voltage while bipolar thresholds are set by the forward bias voltage? There is still some variation in threshold with Vcc/Vdd because of the feedback. To prevent that they would need to set the feedback with a voltage reference and I doubt they do that.

If I fail to understand the mechanism, please explain it to me.

Rick

The easiest CMOS Schmidt trigger can be described as follows:

Make a string of two inverters. Now take a 3rd inverter and place it backwards across the 2nd inverter. The first inverter then has to fight the third inverter, which causes the hysteresis.

If you ratio the inverters, then everything tracks the supply voltage.

So is this how they make hysteretic inputs on MCUs, etc? Or is this speculation?

I'd like to see a reference.

Rick

Take a look at the TI 74HC14 data sheet. Hysteresis increases with Vcc, not quite linearly.

Or if you buy them with the regulator that's powering the device, they're usually even cheaper and smaller (one pin).

On a sunny day (Sat, 27 Oct 2012 11:17:22 -0700) it happened miso wrote in :

(relays)

Then use a real newsreader, I wrote one, its free:

Been serving me well since I ditched win 32 for Linux :-)

I just mentioned that, to make that relay disconnect before power down is an other matter.

Yes.

On a sunny day (Sat, 27 Oct 2012 17:12:50 -0400) it happened rickman wrote in :

Yes, the world changes, I have suggested to make China the next US state. We must give the Russians that they got Sputnik up first, and never fell for the space-shuttle maddness. They seem to have a problem getting things to mars, and China may well walk on that planet first. US seems to have a hell of a problem with joint strike fighter and boeing's new plane. US too much focus on tronics, Von Braun did it with simple hardware. Soon we have electronic steering in cars, I do not want one :-) Politics has killed engineering, next WW will sort all that out. The best one will survive / win / whatever live in the following dark ages, earth flat again.