LT Spice schmitt oscillator

John Larkin ha scritto:

Hello John, try this:

Version 4 SHEET 1 2100 1172 WIRE 848 -416 752 -416 WIRE 1040 -416 928 -416 WIRE 752 -304 752 -416 WIRE 848 -304 752 -304 WIRE 1040 -304 1040 -416 WIRE 1040 -304 912 -304 WIRE 1104 -304 1040 -304 WIRE 1152 -304 1104 -304 WIRE 752 -272 752 -304 WIRE 752 -160 752 -208 FLAG 752 -160 0 FLAG 1104 -304 out SYMBOL Digital\\\\schmtinv 848 -368 R0 WINDOW 3 30 111 Left 0 WINDOW 123 34 145 Left 0 SYMATTR InstName A1 SYMATTR Value vhigh=5 vlow=0 trise=25n SYMATTR Value2 tripdt=5n vt=2.5 vh=.9 SYMBOL cap 736 -272 R0 SYMATTR InstName C2 SYMATTR Value 100n ic=0 SYMBOL Misc\\\\EuropeanResistor 832 -400 R270 WINDOW 0 27 56 VTop 0 WINDOW 3 5 56 VBottom 0 SYMATTR InstName R1 SYMATTR Value 100k TEXT 536 -392 Left 0 !.tran 0 200m 0 uic

Stefano

Helmut Sennewald once told me here in the NG. If you want to use the library Schmitt add this to value2:

Vhigh=5 Vt=2.5 Vh=1

and this in SpiceLine:

Td=5n

Accessible by right-clicking the part. Change those values according to the datasheet for the part you actually want to design around.

We don't do homework. ;-P

Cheers! Rich

Or use a real Schmidt model instead of behavioral ;-) ...Jim Thompson

Thanks, Stefano, that works. All those device params have to be set up.

Do you know what that extra pin is for? I doesn't seem to do anything.

John

Yep, for timing-critical stuff you have to. But a simple oscillator is often ok.

For accuracy and temperature stability far better than attainable with a Schmidt, use unbuffered inverters...

I mostly use them for PWM stuff where some "fuzz" on the slopes is desired, as long as the slopes are as steep as they can be. Sometimes adding even more fuzz is very beneficial at the EMC lab.

That's "device common". Typically connected to ground, and all unused terminals are to be connected to device common.

uuuhhh! Mike:-)

When you want to run a logic gate between 24V and 29V, it is nice to have a ground reference pin.

Others have already helped you out on your actual question.

The logic gate things are all infinitely fast or faster unless you put in the optional parameters for speed. The classic toggling flip flop doesn't work right unless you slow down the flipflop etc. LTSpice doesn't do well when a string of devices have gain and no delay.

Those nice folks at Linear did me a big favor when they put the logic gates in. I have used them to model fairly complex logic in with the analog circuits.

I want to make a low-frequency triangle wave, 1 KHz maybe, to FM spread-spectrum a bunch (namely 5) of switching regulators on this board I'm doing, so that they don't leak into the ADC and show up in the eventual FFTs we're going to do and make birdies in the spectra.

Neither the amplitude nor the frequency is very critical. An NC7S14M5 running at 3.3 volts will give me about a 0.9 volt p-p triangle.

John

Why so expensive? To spread things nicely you'll probably want five individual oscillators all running at different and non-correlating frequencies. Why not just use one 74HC14 for all of them and then use Schmitt inverter #6 for a blinkenlight or something?

I thought about splattering them separately, but don't really see any advantage. Each switcher has one resistor to set its nominal frequency and one more to set the FM amount. They should be all over the place.

John

It'll be a lot easier to test with only one oscillator, too. N-dimensional testing spaces are no fun.

Cheers

Phil Hobbs

There's a logical fallacy, common in things like cryptography, to the effect "If I make it so confusing that I can't understand it myself, it must become random." This attitude has, literally, sunk fleets.

If I set the base switcher frequencies different, and FM them from the same triangle but with different deviations, I doubt that any of them will lock. There may be beats and very brief lock-like intervals as all five of them are sweeping here and there, but the fallacy does include a bit of hidden truth: none of these interactions should make spectral lines anything like what I'd have with five CW oscillators.

Aside: I was recently thinking about N-dimensional optimization in general and of of the performance of a very complex scientific instrument in particular. The particular instrument I'm working on was designed by chemists, and has very bad electronics. Adding EEs increases N by adding parameters that the chemists never imagined existed, or adds some parameters that didn't matter before because their effects were buried in noise. So we show up and, in theory, make the sweet spot even more difficuly to find than it was when life was simpler... 4 more dimensions at least.

This one isn't too bad. I can bench test the system to verify that we've sort of spread the switcher lines and fiddle resistors by instinct. Then we can do some real samples with the ss on and off, just for fun. That doesn't interact with anything else.

Is *anyone* capable of optimizing, or even mildly improving, something as complex as a social system?

John

Sure. Just shoot all "liberals" ;-) ...Jim Thompson

Well, the spread spectrum guys don't think so :-)

The last sentence kind of sums it up. Your switcher base frequencies are all different but if you use only one modulator then they will all scoot up and down in unison. IOW it's like a clump of you-know-what sloshing around with the waves near a beach. Takes forever to disintegrate.

Yes, but that would go OT here. OT as in Old testament :-)

John Larkin a écrit :

Hmmm, you mean like bringing democracy to a foreign country?