Transistors

That's CLOSED LOOP... portends to some nasty OPEN LOOP responses.

...Jim Thompson

The transistor.

John

All the graphs, 13 through 22, suggest something going on around 10 MHz. But "hang up" and "catch fire" aren't mentioned anywhere.

John

That's all right, with extra credit for the oscillation. But it's impressive how many people with fat resumes can't explain this fairly simple case.

John

It's a mix of actual NPN and PNP models and macros. It claims to be derived using "PSpice Model Editor v9.2"... one of the worst pieces of crap ever turned out by PSpice.

...Jim Thompson

That would be noise sources... unless you happen to have a device so high in fT that it's unstable in its own right ;-)

...Jim Thompson

I will admit to involvement in some "hang up" and "catch fire" events... taught me to evaluate loop gain and phase during abnormal bias conditions, for instance during power-up.

...Jim Thompson

They don't know the answer?

BTW, John, I'd check the supply bypass caps if I were you. I've spent the morning exploring some NASTY resonances in common popular type-3 multi-layer 0805 ceramic caps, after seeing the plots in OS-CON's lit saying they are better.

And, yep, guess what, all hell breaks loose over a narrow frequency range, then everything's fine again. Such as 6-milliohms at 795kHz, rising to 18-milliohms at 796kHz and 48-milliohms by 796.8, then up and down till 805kHz, then back to 8-milliohms at 809kHz and then rapidly down to 1 to 2-milliohms above 810kHz. Sheesh!

The cap has to be under dc bias to see the problem.

I put my 4192A impedance meter on slow high-res mode and scanned in 100-Hz steps to explore this.

What do your supply rails look like while the opamp is in trouble? You could also have a nice undamped bypass- cap resonance going.

Most small-signal discrete NPNs will oscillate in this situation. A typical 2N2222, for example, will scream a volt or two p-p at the emitter at roughly 100 MHz. That can lead to some very weird DC behavior.

Hey, we still have zener and avalanche behavior to discuss! Not to mention the Pease optical transfer effect.

John

Black phenolic transistor encapsulation passes IR like the phenolic isn't even there.

Solved an IC problem once by adding an extra layer of metal that covered all junctions.

...Jim Thompson

The old GE plastic transistors used a bownish epoxy and were photosensitive. That caused some strange dc offset and hum effects.

John

They'd pick up the fluorescent lights in our lab.

...Jim Thompson

With parasitic inductance in the collector of Lc,

Frequency: 1/(2*pi * sqrt*(Ccb * Lc))

Amplitude: somewhere between 0 and a couple of volts peak depending on the transistor and how crappy the layout is.

Best regards, Spehro Pefhany

I've seen transistors oscillate fiercely when the collector was hard bypassed, and when the collector had a resistor and/or a ferrite bead in series. Seems like the wirebonds and various capacitances make a colpitts-like oscillator. The sure fix is some base resistance.

I got burned a while back doing a classic opamp-pnp closed-loop current source. I figured the output impedance of the opamp would be sufficiently resistive to damp the pnp's tendency to oscillate. Wrong.

John

_PHENOLIC_? Really? Dinosaurs were walking the Earth then, right? TO-92s and all subsequent epoxy packages have been made of Novolac, which is nice and opaque at all wavelengths of interest. Remember a few years back when there was a fire at a Sumitomo plant and nobody could get Novolac for awhile? That was a mess.

I remember hearing about some early plastic-packaged op amps that used dye to colour the plastic package. Organic dyes are nearly all transparent in the near-IR, so there were some interesting offset voltage funnies. (The NIR transparency of dyes is why you don't use exposed colour film for viewing eclipses--exposed B&W film is pretty good.)

Cheers,

Phil Hobbs

Told you I was "old as dirt" ;-)

...Jim Thompson

On a sunny day (Sun, 18 Nov 2007 09:09:36 -0800 (PST)) it happened Winfield Hill wrote in :

.....OK, :-) if you say so.

Oh well, I had to import some Zetex spice models in swcad, and these are not the ones I want. I get 30 degrees phase shift at 10MHz, with 250MHz ft transistors, there are much better ones. Removed some stuff too (makes little difference). ftp://panteltje.com/pub/amp1_ac.jpg It does not have so many, what do you call it, 'ripples' as that chip. I am sure with the right transistors this will do 50MHz. Actually I used it in the past several times.... for other things. So what do you have against that configuration? This is with lower emitterio resistors: ftp://panteltje.com/pub/amp1_ac2.jpg

Jim Thompson a écrit :

On my 200pV/rtHz preamp I first had a huge unexpected 1/f noise with very unstable level. Scratched my head for almost one day long. Then the problem suddenly disappeared when clouds came masking the sun. Gasp! The path was through the window, reflecting on the lab walls, then reflecting up on the PCB and finally entering the IF3602 jfets through the glass beads sealing can.

Don't "mystery failures" really drive you nuts?

Had a problem with a JFET OpAmp breadboard just up and literally smoking. Took me a week of failures before I determined that the Moto RF guys had moved into the lab next door... every time they keyed their transmitter my OpAmp breadboard went bye-bye :-(

...Jim Thompson