Unused LM319 inputs

No, not ok, it can oscillate badly. One needs to go high, one low.

Or inverting to output and non inverting to ground.

Never both to ground.

A wee challenge, because of the open collector output. The other downside with this comparator is that, unlike with the LM339, the inputs cannot be farther apart than 5V. And that's already abs max so I'd say below that.

Phil, if you have some sort of reference or other voltage that is in the few volts range you could use that. It doesn't have to be stable and it can be noisy.

And also never to any other node together.

Thanks. I rearranged the design to put the 3.3V supply on this card, so I now have a bias voltage to tie the spare input to.

Your average fast comparator will oscillate if both inputs are in their linear range, for sure. My uncertainty is mainly about the input CMR, which doesn't include either supply. If I tie both inputs low, it'll turn off the input stage, which should prevent that section from doing anything, but may screw up the bias generator for the other section too. The equivalent schematic on page 7 of the National datasheet shows a fairly weak-looking bias supply driving the base of the input stage current source. If both input transistors are turned off, the current source transistor will saturate, which looks like it'll drag down the bias supply and turn off one half of the chip completely.

What I'm not clear about is whether that'll turn off the other one too. I might have to try it out, but I'm frantically busy this week.

Cheers

Phil Hobbs

Normally they don't influence each other unless you drive one section to latch-up. But to be sure you'd have to ask National about it. If in a rush just use a couple of resistors and make a lower helper or something for the other pin.

Old man Muntz would have said you are wasting money if you don't need the 2nd half :-)

Yessir, anything you say, Mr. Muntz, sir--I'll just replace it with a nice single LT1016, for ten times the price. ;)

Cheers

Phil Hobbs

The cheapest *fast* comparator around is an LVDS line receiver.

John

Nah, use the other half as an amp, gate, inverter, Schmitt or whatever you can replace in there. But make sure to have enough barf bags available at the design review when that part of the schematic comes up :-)

Some semi-famous person said "never use an opamp as a comparator, and never use a comparator as an opamp."

Wimp.

John

Unused LM319... ground inputs AND output. ...Jim Thompson

Cheapest one I see is DS90LV012ATMF, about 36 cents in 1000 pcs, vs. about 15 cents for an LM319. Of course 80 ns isn't exactly _fast_.

This is an optical datalink for an ad-hoc battlefield network (a DARPA subcontract). It runs with about a 300 kbit data rate at a centre frequency of 1 MHz, so I'm using a fancy photodiode front end running into an SA604A FM IF chip, with the RSSI output controlling a 2N7002 as the bottom element of an L-network attenuator on the input to get another 40 or so decibels of dynamic range. The LM319 is producing the

It winds up being pretty simple and cheap, works very well in direct sunlight (even with the Sun in the field of view), and is shot noise limited down to the medium nanoamps. The RSSI/MOSFET trick is quite pretty--not exactly AGC, because it doesn't do anything at low light, it just reduces the maximum signal without changing the mark/space amplitude ratio.

You can do a lot with photoreceivers if you get rid of the 300 kelvin resistors in the low-level stuff.

Cheers

Phil Hobbs

That's like putting a glass over a hornet, it'll keep on buzzing inside. Then uncle Leroy's TV quits working because ...

Also, chips that are internally chattering at a fast clip can begin to suck in quite a bit of power. Although it probably won't be extreme on this one.

I didn't know bologna made the grass grow green ;-) ...Jim Thompson

Careful with that SA604. Arrow has a healthy stock situation but Farnell doesn't even find it. Considering that it's a European chip, well ... stuff like this can go lalaland at times.

Turns out that the LVDS inputs of most FPGAs make very nice r-r comparators, too.

John

I did have garlic bologna on bread at lunch but it hasn't made anything green :-)

Still over there?

Yep :-( ...Jim Thompson

Thanks.

NXP has been introducing new variants for awhile, so I think it'll be okay for DARPA purposes. I love FM IF chips--the late lamented MC3362 had a 100-uA RSSI output that would give 40 ns edges if you cascoded it. I used it like that for a project to make a pocket 3-D scanner--a combination coordinate measuring machine and page scanner that was about the size of the palm of your hand. It got cancelled before it was finished, unfortunately--it would have been a pretty cool gizmo. I may have the opportunity to resurrect some of that stuff soon.

Cheers

Phil Hobbs

Thanks .

So it doesn't drag down the other section's bias?

Cheers

Phil Hobbs