Differential-pair amplifer

ROTFLMAO! Dummies are good for business... except when they insist they know what they are doing :-(

...Jim Thompson

(snip)

Two properly adjusted stages can produce a damn good approximation of a sine wave from a triangle.

That is my understanding.

Well, the compressor part of the compander, anyway. You need another one in a feedback path to do the equivalent expanding, downstream.

I think it is useful to compress a wide range AC signal to be digitized by an A/D converter, having high resolution at small amplitudes, and lower resolution for large amplitudes, where the resolution is mostly wasted the A/D result can then be linearized with a look up table.

I think audio that any exceed the swing limits of a power amplifier sounds a lot better if the hard clipping is prevented by predistorting the audio with a tanh compressor so that the power stage cannot reach clipping level, or if it does, clips gently, so it doesn't generate such high harmonics of the signal frequencies.

If it is true that a pair of transistors, a differential pair amplifier, has a Tanh transfer characteristic, then at audio frequencies it behaves as a sort of amplitide or volume compandor.

At large signal inputs the output approaches a squarewaveform and hopeless distortion occurs. At intermediate inputs audio distortion increasingly appears.

What other so-called useful applications does the circuit have? I can't think of any myself?

you can convert a triangle wave to a pretty good aproximation of a sinewave with it.

Colin =^.^=

I guess you don't consider a circuit whose output is a function of the difference between two inputs useful, but a lot of op-amp makers find it convenient...

In article , Reg Edwards wrote: [...]

That distortion is not hopeless if it is what you need. In that case it is very nice distortion. Take a look at the circuit of a comparitor IC like the LM339. It distorts like crazy just like you would want it to.

If you combine a few in tricky ways, you can make an analog multiplier and or, a circuit that finds the RMS of an applied waveform.

You can use them to perform logical operations.

Kevin, Kevin... John was merely speaking tongue-in-cheek!

Now Kevin, I'm slaving away making accurate low-current MOSFET spice models, you can join the fray, too. The more the merrier. Also, just because a MOSFET model is available, don't assume it's accurate below say 3 to 5% of the FET's absolute-maximum current. In other words, it's likely to be inaccurate in the subthreshold linear-circuit region that interests us. However, it's not too hard to fix the models with a current-controlled current source feeding a diode directly in series with the FET's gate. Using n=3 works pretty good for a 2n7000, big power MOSFETs are around n=5.

See the Steven M. Sandler and Charles E. Hymowitz article in the May 2005 issue of Power Electronics Technology magazine. OK, I know it's not out yet, but... In the meantime, try this link,

Sandler's approach works well, after correcting his parameters.

Not at all. You just need to know how to design with them correctly:-)

If you want really high CMRR, they are the *only* way to go. And as we can see from the above, in the correct configuration, distortion can be very low indeed.

Kevin Aylward snipped-for-privacy@anasoft.co.uk

What? You don't want to deal directly with the device model ?:-)

As I showed a while back, the Supertex Level=3 model was pretty close, without adding on subcircuit elements.

I'm always leery of add-ons based on limited measured data, covers one particular case, but may bomb out on the next.

How does this model behave in regard to modeling gate charge?

Win, I don't receive PET magazine. Could you get me a copy of the Sandler & Hymowitz article when it comes out? Thanks!

...Jim Thompson

It s useful in analogue video when you are doing cross mix between 2 video source ISTR, (from a video servicing course at Ampex years ago) Sorry Reg, I cant remember the theory, I fell asleep during this part of the course

martin

After the first death, there is no other. (Dylan Thomas)

Current-mode opamps have a different topology.

John

Indeed. But we don't want any newbies to get the wrong idea:-)

Tell you what though, this discrete design lark is getting really difficult. All the components are disappearing. Its really hard to find

Kevin Aylward snipped-for-privacy@anasoft.co.uk

Hello- ever hear of something called feedback- can be used to pre-distort the input drive of the diff amp as well as maintain operation in small signal region. Is there such a thing as a monolithic op amp without a differential amplifier front-end?

You'd better explain that to all the semiconductor companies; they keep calling them 'opamps'.

John

Where'd you show that? I'd love to take a look at it.

Bring it on.

Actually, in this instance, it's general, because at the low subthreshold currents the MOSFET reverts to diffusion-driven n VT = n kT/q performance, very predictable and very general for all MOSFETS, with a little variation in n, natch. Takes over nicely in the region where Id otherwise drops like a rock because it's below Vthreshold.

No problem, you add the gate-voltage-adjusting diode *inside* the complex charge-modeling circuitry, stuff that's already carefully worked out by the manufacturer and is retained.

It's a free mag and I get it, but they also put up the articles on the web. The May issue must be overdue, but I'm watching for it and will alert you when it hits. Meantime Sandler's notes give the idea. It's a better approach than the successful one I had developed with components in the MOSFET's source. I had gotten that one working well over a 50pA to 100mA range, but I've now exceeded that with the new one from 50pA to 2A. Recommended.

Most people understand op amp to mean voltage mode- current mode only made a difference because people were too lazy to invent a different circuit symbol.

Originally posted to A.B.S.E on 4/6/2005, re-posted at....

Newsgroups: alt.binaries.schematics.electronic Subject: Win Hill - 2N7000 Spice Model Comparisons - Compare2N7000Models.pdf Message-ID:

...Jim Thompson