Ping Jim Thompson Sine Shaper question ????

Hi Steve, I'll see what I can conjure up. It's been awhile. My latest "sine shaper" circuit was more than thirty years ago...

How precise do you need it to be? Some years earlier I cracked ON-TV's horizontal sync perversion more accurately than their own box using just discrete devices >:-} ...Jim Thompson

If you're thinking of making a discrete-transistor copy of the '8038 circuit, you'll want a bunch of NPN+PNP trannies in the same package, and a small package too, I'd imagine. We don't list any in AoE's Table 8.1b, matched transistors, because NPN PNP pairs aren't be matched, exactly. But they are more or less, and hopefully are at the same temperature.

Take a look at Diodes' MMDT3946, a 2n3904 + 2n3906 combo in a sot-23-6 package. They also make MMDT2227M, MMDT4413 and MMDT5451, but these aren't small-signal types. NXP make a 2n3904 + 2n3906 combo, the PMBT3946YPN. THAT makes a matched NPN PNP array, but I doubt you'll be interested. Zetex makes ZXTD6717E6, NPN+PNP 1.5A trannies in sot-23-6.

Why not use a commercial unit? The LM311 will probably outperform anything you can make with discretes:

IC designers can throw lots of transistors at a problem; they are basically free. There's probably a better way to do sine shaping (or, actually, sine generation) than copying that circuit.

How about the _sine_converter_ portion? The comparators and current sources only create the triangle wave (ala 555 timer style)... which is then shaped to sine. ...Jim Thompson

If I understood your post correctly, you want to duplicate the ICL8038's internal design with discrete parts because the XR2206 is no longer in production. Why not duplicate the XR2206's sine shaper directly?

If this is for a one-off project, you can still get both types at AliExpress. Judging from the prices, I'd say there's a good chance that they're *not* Chinese knockoffs.

In fact, a couple of thousands of MAX038 are still available from Rochester.

On Thursday, September 21, 2017 at 11:12:26 AM UTC-4, snipped-for-privacy@gmail.com w rote:

sil 8038 sine shaper section? Since the 2206 is out of production, I don't think it would hurt to discuss it..

The Breakpoint Wave Shaper. the ICL8038, the AD639 and the XR-2206 are all discussed in Sergio Franco's Design with Operational Amplifiers and Analog Integrated Circuits. The book also discusses The Logarithmic Wave Shaper a nd the JFET Wave Shaper which may be more useful for a discrete design.

I don't know if this is useful for you but just amazes me that I remember b ooks I read long ago.

Unfortunately the thermal coupling between the two dice in the single package is hardly better than if you just mounted two single devices from the same reel next to each other.

Cheers

Phil Hobbs

Yup. Using a PLL with an AGC-stabilized oscillator, for instance.

Cheers

Phil Hobbs

A 311 in a sine shaper? Do tell.

(Sine shapers take triangle waves and turn them into approximate sine waves, usually with a BJT diff pair. Typical distortion is around a percent, but you can do better if you subtract off a small amount of the original triangle wave to get rid of the cusp in the output waveform.)

Cheers

Phil Hobbs

Just the sine shaper. Triangles are easy.

Steve

Actually, just a pair of diodes, clippers to ground, isn't too very awful.

There must be some old Wavetek function generator schematics around. I recall some reasonably simple circuits, all discretes.

Den fredag den 22. september 2017 kl. 01.53.31 UTC+2 skrev John Larkin:

google...

I'm interested in the sine shapers. I'll see if I can get Franco's book on intralibrary loan. ILL is one of the few nicer perks of doing my job for 26K $ less then what I'd make in the private sector.

Steve

ICL8038 Comparator usage Functional Diagram, Page 1 Detailed Schematic, Page 4

Triangle wave generator Schmitt trigger followed by integrator

RC oscillator generates linear triangle wave

Yah. But Steve was asking about the _sine_shaper_. The triangle/square wave oscillator is the easy part.

Cheers

Phil Hobbs

That last one is a TANH, mostly.

I was going to suggest that. Really.

Found my Mathcad doc from when we had this conversation back in 2010:

With an LM13700 and a bit of tweaking you should be able to get to 0.03% THD, which is 70 dB SINAD.

Cheers

Phil Hobbs