Sziklai Squared ??

That's basically a polarity-inverted version of the NFET + PNP wraparound trick that I sometimes use (but certainly didn't invent). The main difference is that you're using a fed-back voltage source in place of the tail current source.

Normally I think that trying to use feedback to turn an emitter into a high impedance is a mistake, but in this case at high frequency you get the source and emitter impedances in parallel.

It's an interesting idea, if it doesn't oscillate fiercely. Got a few words for us about the circuit details?

Cheers

Phil Hobbs

Actually it improves the effectiveness of the PMOS gm.

It looks like it might take-off > 100MHz, so I'll just kill the loop-gain before then... I just need a solid X1 gain at 2MHz.

Can't... NDA. ...Jim Thompson

As an aside, did I just read that _Jim_Thompson_ is building a Sallen-Key filter, using a few transistors instead of one of those ones that needs 12 op amps?

The crowd gasps at this unexpected display of sanity. ;)

Cheers

Phil Hobbs

I wondered who would catch that ?>:-}

It's a simple 3rd-order Butterworth situation with +/-20% corner accuracy no big deal... the shape will always stay Butterworth and be well-behaved due to the ratiometric advantages of chip design.

Plus I have a real-world power budget. ...Jim Thompson

Here it is as bipolar... needed some compensation...

(No attempt to optimize... probably needs a zero as well.) ...Jim Thompson

Looks reminiscent of this gem from 1965, designed for the same purpose:

I used the Darlington inside a LM13700 to make a Sallen Key one time, for filtering out the sum frequency from a four quadrant multiplied signal. Works good...

Maybe he could make the capacitors out of gyrated inductors.

Heh, it's a complementary differential pair, strapped as a follower. :)

Tim

-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Contract Design Website:

"Jim Thompson" wrote in message news: snipped-for-privacy@4ax.com...

Not a bad idea. The LM13700's Darlington has constant bias, unlike the one in the late lamented LM13600, whose bias came from I_ABC. That wouldn't have worked as well.

Cheers

Phil Hobbs