Oscillators in LTspice

Oscillators in LT Spice often need to be goosed with an injected pulse or some initial condition setting, if they are to start in a tolerable time, or at all.

Or skip the initial conditions solution, or bring up the power supplies after sim start.

In a multivibrator, some tiny component value tweak can make the asymmetry to kick things off.

Actually, the classic astable mv usually has a genuine hangup state, both transistors saturated.

Return the base resistors to the collectors of their respective transistors instead of the +ve rail. That will prevent them from sarurating and will ensure reliable starting.

Right. It makes the transistors go linear if it's not oscillating.

Yes, I discovered that in the early 1970's when I first started using multivibrators. I don't remember seeing it documented anywhere.

I invented that too! Roughly the same time.

The supply coming up should be enough of a goose to start the oscillator.

boB

They don't usually self-start. Rounding errors in the arithmetic do act as a source of wide-band noise, but for any practical simulation you set up the initial conditions to launch the oscillation at an appreciable fraction of the steady state amplitude.

That’s actually the problem—some circuits with lockup states will work fine on the bench and only fail in the field, due e. g. to wimpier supplies and brownouts.

For LC and crystal oscillators, startup problems occur when the small-signal (class A) gain is too low. Circuits with that problem start up fine if the turn-on transient is large and sharp enough, but not otherwise.

In a simulation, the very small-scale nonlinearity of rounding error can make the circuit seem to have startup problems that the real circuit doesn’t, sort of like how you can stand an egg on end if you’re careful. (Or sprinkle the bar with salt first, but I digress. )

t’s perfectly legit to goose an oscillator simulation as long as it’s firmly in the small-signal limit, i.e. the goose is << kT/e, so that it’s always class A.

Cheers

Phil Hobbs

It's sometimes hard to understand your pun spelling meanings, maybe because you are not a native English speaker. Foreigners who try to pun in English are often ludicrous. I was at a giant gathering at a US national lab and some French bigwig tried to pun in English. There were about 2000 instances of dead silence. I almost felt sorry for him.

LT Spice is wonderful. Ususlly I can go straight to a production PC board from a good sim.

I have noticed that the Dutch tend to be cold and rude.

The food is medicre too.

ASML is the biggie in semi fab, but they bought the technology from Cymer.

The do tend to be rational and direct. If you expect to be flattered non-stop you might find that rude.

My experience in the Netherlands was that they were mostly warm and friendly. I once got dragged into some tricky correspondence with UK suppliers, to add some native speaker polish to the English language letters we were sending them, and got told off for trying to express the demands more diplomatically (strange as that may sound).

There are exceptions, as there are even in the USA.

And made it work. That seems to have taken some help from Phil Hobbs.

I'm not fond of ASML - they wouldn't hire me when I was job-hunting in the Netherlands. It's a Philips spin-off and their human factors department has the same unjustified high opinion of their own expertise as every other Philips human factors department.

Philips has got a lot of high technology into the market - the Compact (optical) Disk format is an obvious example - and their pure research mob were in the Bell Labs class (if quite a bit smaller).

The supply may be slow to come up - perhaps it's a NiMH cell with a photovoltaic panel charging it.

I'm working with a PCB layout guy who is great with placement and routing and such, but terrible at following written directions and with spelling and reference designator placement. In other words, very visual but not very verbal. Not a native English speaker.

I suspect that the Roman character set and decimal number notation affect the way people think and get some credit for the Enlightment and for the technical progress of western countries.

How do Chinese companies do their reference designators? All the stuff I've seen looks very western, not Chinese characters, except that the prefixes are arbitrary, like TR for transistor, or TR for transformer, whatever.

The chinese and japanese and some other char sets need high resolution too. Our smallest ref desigs are 50 mils high, which works well with modern inkjet printing.

There are native speakers of English who are more visual than verbal.

I knew one brilliant engineer who was so verbal that he couldn't see the point of spelling "their","there" and "they're" differently because they were all pronounced indentically. I once offered to check his text for him, and he rejected the offer because he couldn't see the point.

Not an opinion shared by the kind of visual psychologists who think about that kind of stuff.

Korea, Thailand, and the Arab countries all have their own character sets. Japan has three. The Cyrillic alphabet is pretty close to Latin and Greek (because it evolved from them).

The basic fact that human languages all use a limited number of phonemes (drawn from a common stock of about 200 which pretty much every human can produce and distinguish) pretty much defines how writing works.

The practical details get messy.