Best way to ensure signal does not exceed 5v?

One of the simplest ways: send the output to ground through a resistor and a 5V zener diode. This is a kind of voltage divider, and so the resistor has to be chosen properly with regard to the impedance of the next device. You take the output from the top of the Zener (between the resistor and Zener).

How your Zener works is that it does not conduct current (or conducts a very small amount of reverse leakage current) until its Zener voltage is reached, and then its starts to conduct copiously. At this point, current is limited by the resistor. Whatever voltage is in excess of the Zener, that excess voltage is dropped across the resistor.

If your resistor is too high for the input impedance of the next device, this Zener circuit will not exhibit a stiff voltage. For instance suppose that the next device is disconnected and you have a 100K resistor with the Zener, and the analog interface is programmed to output 4V. This means that the Zener is not turned on (it is a 5V Zener, recall). No current flows through the 100K resistor, and so the output (top of Zener voltage) is 4V. Very good. Now suppose you hook up the input, and the input has only a 1K resistance. Now the Zener becomes irrelevant! Current flows across the 100K resistor, and through the 1K input, forming a 100:1 voltage divider. Your actual output voltage is now about 0.04V, not 4V. Oops!

Choose this resistor so that it's 5 to 10 times smaller than the input impedance of the next device. But if the input impedance of the next device is small already, this means that the resistor will be small, which means that a lot of current can flow when the Zener diode is turned on (10V worst case).

Without being able to assume anything about the next device, or about the analog output device for that matter, this can't be 100% correctly designed using just a couple of passive components.

If you cannot assume anything at all, you can end up way over-designing it. Gee, let's see, 0 to 5V, rock solid, noise-free, into anything from 0.1 ohm, to 1Mohm. No problem, we will just need a couple of powerful transistors bolted to a heat sink ...

Hmm, page says: "Limited quantity available, not recommended for new design".

Invert?

Emitter to +5V, collector to resistor to ground: PNP!

"Obsolete item", unfortunately. But there many such devices in the catalog.

yeah, if you need the ruggedised one.

96c for singles. the SMT version is slightly cheaper.

dunno where you got LM286, he wrote LM386 (which is a low power loudspeaker driver recently discussed in another thread and perhaps the cause of the error) it's fairly obvoius he meant LM336 (which is described in the PDF)

not that cheap unless you buy 5000

--
?? 100% natural

--- Posted via news://freenews.netfront.net/ - Complaints to news@netfront.net

Hahaha! *rubeyes* I can't belive I just wrote this bull.

Sorry about that! Well, Back to sleep.

Actually, there is a way to have a single PNP non-inverting switching stage for this, which is what I must have been thinking of, since I worked with such a circuit not long ago. Suppose we want 0V output, until the input is somewhere above 3V, in which case our circuit is to swing to +5V.

One trick to get non-inverting behavior in a single transistor stage is to convert the input voltage, or portion thereof, to current and sense it by the BE junction. A PNP transistor is required because it has non-inverting behavior (under a positive supply) with respect to current. Since the load is ground-side, the more current flows through the collector, the higher the load voltage rises above ground. We just have to make collector current follow input voltage.

What we do is bias it like this: the input drops across a current-sensing resistor, which is placed across the PNP base-emitter terminals, then another resistor to help limit current, and finally across a 3.3V Zener to ground. (Or maybe a stack of regular diodes to get some lower voltage, as desired.) When the input voltage is below

3.3, no current flows across the resistors and so the PNP is cut off. As the input voltage rises from there, current flows across the resistors. When about 0.7V builds in the current sensing resistor, the PNP turns on. We put no feedback resistor between the transistor's +5V and the emitter, so it saturates easily. When it does saturate, it pulls the load to +5V (minus the small saturation VCE). Thus load voltage polarity follows input polarity: no inversion.

He said LM136. That's what I searched on. Still, $1 vs $.004; you pick.

Typo. Yes, but John's post was so screwed up there's no telling what he meant.

At $.002 each that's a whole $10 for a reel. Wow! ...like you'll never use another.

smaller.

--
How so? You can't get around a typo? ;)

It's just a low-power shunt reference to ground being used like a
Zener.

5.1 volt zeners have a soft knee, and rarely do what you described.

--
You can't have a sense of humor, if you have no sense.

I guess speed wouldn't matter since I doubt the DAC converter is going to respond any faster than this circuit can handle..

If would be simple to add a R load if needed.

Jamie

Right, might just as well use a Schottky clamp to the 5V rail or use the input's ESD clamp, if available (though I don't do it unless it's suggested in the spec).

smaller.

If there is any capacitance at the input, it could hang for some time. There is nothing but leakage to bleed the charge outta there. Who knows what the leakage is, or even direction.

Unless we know a lot more about the circuit, the R is needed. Unless it's a low impedance input, there is no need for the transistor at all. Both are a bad idea for a TTL type input, though.

smaller.

Typo? Sure, but it wasn't a typo.

Sure, but it's a crap idea. As I said, expensive and an all-around PITA.

So? What happens if he sets it to 2.5V? Think!

That's a *great* idea! (I was the first to suggest a Schottky. ;-)

I consciously chose this approximation of the Zener because it fits the application.

The knee behavior is not important, because there is no requirement to clamp voltage to exactly 5.000 volts.

OK, not quite like it in exactly that way -- but there's tons of pre- biased transistors out there to be had.

--
www.wescottdesign.com

smaller.

An

--
Well, since there aren't any LM386-5.0's around and there are _lots_
of LM336-5.0's out there, I'd have to say it appears you're wrong.

e

ew

t,

bit

o a s

Wow, I left town for a meeting and come back to an EPIC THREAD! ;0)

Lots of info here, this thing has taken on a life of it's own. After re-evaluating my requirements against my knowledge/abilities and time constraints I have selected the voltage divider solution: #1 my primary goal was to avoid hardware damage - check! #2 If myself and my team can't read code comments and follow instructions to output 10v then I have bigger issues to worry about.

I appreciate the input from ya'll, thanks and have a good weekend.

-Steve

ce

ol

o s

DAC,

FYI, current requirements are exactly 2mA so I should be good. Thanks for the information.

Kaz' post has some good points and I would like to try my best to answer them. However all my family just showed up for the holiday weekend and I don't have time yet. I will try to post Monday, I just felt that I should say this so that no one felt ignored. ;0)

A low voltage zener has a lot of other failings, but if you want to half ass something you can't say that people didn't try to tell wy it was a bad idea.

--
You can't have a sense of humor, if you have no sense.