Ping Jim Thompson Sine Shaper question ????

Nope. You need one in the sine shaper to avoid even-harmonic distortion.

See above.

You don't design a lot of oscillators, I gather. There are a lot of ways to skin that particular cat. Just wrapping an ALC loop around a normal RC oscillator will do a good job. That plus a PLL will handle a pretty wide range of cases.

Cheers

Phil Hobbs

All analog sine generators/shapers are junk compared to what can be done with a $1 uP

The sine shaper has a lot of distortion. A few tenths is not going to make much difference. One poster even showed a version using 1N4148's.

Actually, much of my work centers on stable low phase noise oscillators.

The classic Wein bridge oscillator that started HP is a good example of an RC oscillator with feedback. However, it was noisy, has a lot of jitter, and is restricted to fairly low frequencies.

A pll that can cover a frequency range of 2:1 can be used to cover a wide frequency range using binary dividers. However, it requires good filters to make a sine wave. A DDS is often a better choice, but you have to watch out for spurs. Both approaches can suffer from poor phase noise.

Wien bridge oscillators work pretty well. Double-integrator (i.e. quadrature) oscillators work pretty well too.

Read the PDF I posted. -70 dBc is possible with a simple circuit.

Sounds interesting. Tell us about one of them, in enough detail that we can all learn something.

There's nothing preventing you from building a 1 GHz Wien bridge. It's just that at higher frequency you have good inductors available, so you aren't stuck with a Q of 0.5. Higher Q helps a lot.

It's not difficult to make an LC oscillator with a 2:1 frequency range. You take a hyperabrupt varactor with inductors in series and parallel, build any of the classical oscillator configurations, and wrap them in an ALC loop. Works great.

Even a Y5V capacitor has a tuning range big enough for that.

A DDS is often a better choice, but you have to watch out

Or not. It's a matter of design.

So what sort of oscillators have you designed recently?

Cheers

Phil Hobbs

I've got several tubes of icl8038's do you want a bunch?

George H.

Hardly any. (Rigol DDS serves most needs) But the last was an RC Wien bridge with diode limiting. (Oscillators are fun.)

George H.

Wien bridge is RC.

I see the bus has reached my stop. This is where I get off.

The traditional lamp-as-gain-control-element Wien bridge does produce a really nice low-distortion sine wave. But if you don't want a lamp on your PCB the AGC schemes one can do with other discretes that come in SMT get pretty complicated quickly and AFAIK there's no simple way to make the topology frequency-adjustable over a wide range.

Ah, I got it. Use a uP in the feedback loop running a digital model of an incandescent lamp.

ne

Narrow range isn't too bad. If the in-phase amplitude adjust that you need to for amplitude stable output doesn't contribute much distortion, a quadra ture frequency adjust using the same mechanism can give you a frequency sta bilised output with the same - limited - extra distortion.

The core gain in the Wien bridge can contribute very little distortion, and the extra feed-back element just aren't as good, so any significant correc tion starts degrading your harmonic content. Keeping amplitude and frequenc y spot-on is easy enough, but getting much range isn't.

That gets into the oscillator via a D/A converter. They aren't exactly dist ortion-free. You tend to use the D/A as a digitally programmed multiplier, which doesn't help.

I would think that small signal transistors would work best, as they would be closest to the geometry of the PNP/NPNs on the IC.

...Make sure that BITS of tungsten do not corrupt the program.

No reason you could do it with LC too. (trickier to make it tunable.. dual ganged varible cap?)

George H.

Diode clipping can be OK... Does give more harmonics. In theory you can tweak it to be as low as you like. \ (With infinite voltage head room.) I was going to post a 'scope shot, but all the instruments are packed for shipping.

George H.

For those who don't know, I do still laser shows. For certain analog abstr act effects, we use 8038 and 2206 to generate sine waves for making the vec tor images. I actually WANT the distortion, as it gives a scanned vector im age a certain classic "look".

DDS are rarely used for abstract images for laser shows because the scanne rs can show the "jaggies". Low pass or reconstruction filters are rarely used because abstract patterns are all about phase.

Yes, there is still demand for the odd laser show on the planetarium dome.

Yes, I can do this digitally, and have rather sophisticated software (Pango lin) to do so. Digital stuff, even state of the art digital stuff, looks bo ring. Digital oscillators actually have very fixed phase steps and images o ften look like they are constructed of right angles.

Small thermal drifts in older analog gear add "personality" when your creat ing images in real time.

I do see frequent reports of counterfit and poorly cloned 8038 and 2206 on another forum. As well as production seconds showing up these days.

I like the cascaded sine shaper over the two diode model.

Funny thing about the human eye, when looking at patterns scanned over a 50 foot dome, very subtle things in your base waveforms show up.

It is an artistic thing, and no I'm not being audioPhoolish...

I have one of the original live abstract laser image consoles used by Laser media, it toured I'm told, with ELO and was used at a few Floyd shows..

Its all done with smoke, mirrors, and oscillators.. Yes, I have quadarature oscillators for the baseline circles, too..

Think sum, multiply and modulate, creating real time drifting spirograph an d Lissajous patterns, with six different colors of laser lines..

Yes, people will still pay money to watch forty minutes of live "played" im ages to popular music, provided content is created in real time. The digit al stuff is so perfectly repeatable that it looks canned and gets boring. W hich costs you repeat customers.

Yes, we record baseline tracks in studio sessions and then modify things in real time.

Yes, I have quadrature oscillators..

Maybe this weekend I'll take some pics..

Steve

Please allow me to clarify. I took a few hours off from work this morning and have had time to refine my thoughts.

What I should have asked Jim, is how small signal NPNs and PNPs on 1970s/1980s chips typically differ in performance from their three leaded small signal counterparts.

In terms of Ic, Hfe, BVceo etc.

Steve

Or you could learn something about electronics.

It seems that the shaper is not fussy about the transistors used. Just garden variety NPNs and PNPs. See LTSpice file with 2N3904 and 2N3906:

Version 4 SHEET 1 2264 1512 WIRE 32 32 -384 32 WIRE 304 32 32 32 WIRE 704 32 304 32 WIRE 1088 32 704 32 WIRE 1440 32 1088 32 WIRE 32 96 32 32 WIRE 304 160 304 32 WIRE 32 208 32 176 WIRE 240 208 32 208 WIRE 1936 208 480 208 WIRE 32 240 32 208 WIRE 352 256 304 256 WIRE 416 256 352 256 WIRE -384 352 -384 32 WIRE 704 352 704 32 WIRE 32 400 32 320 WIRE 640 400 32 400 WIRE 1760 400 880 400 WIRE 1936 400 1936 208 WIRE 1936 400 1840 400 WIRE 752 448 704 448 WIRE 816 448 752 448 WIRE 32 480 32 400 WIRE -384 528 -384 432 WIRE 1088 544 1088 32 WIRE 32 592 32 560 WIRE 1024 592 32 592 WIRE 1760 592 1264 592 WIRE 1936 592 1936 400 WIRE 1936 592 1840 592 WIRE 2176 592 1936 592 WIRE 1136 640 1088 640 WIRE 1200 640 1136 640 WIRE 32 672 32 592 WIRE 1440 784 1440 32 WIRE 32 832 32 752 WIRE 1376 832 32 832 WIRE 1760 832 1616 832 WIRE 1936 832 1936 592 WIRE 1936 832 1840 832 WIRE 1488 880 1440 880 WIRE 1552 880 1488 880 WIRE 32 960 32 832 WIRE 1936 976 1936 832 WIRE 352 1024 352 256 WIRE 752 1024 752 448 WIRE 1136 1024 1136 640 WIRE 1488 1024 1488 880 WIRE 32 1168 32 1040 WIRE 352 1168 352 1104 WIRE 480 1168 480 304 WIRE 752 1168 752 1104 WIRE 880 1168 880 496 WIRE 1136 1168 1136 1104 WIRE 1264 1168 1264 688 WIRE 1488 1168 1488 1104 WIRE 1616 1168 1616 928 WIRE 1936 1248 1936 1056 WIRE 1936 1248 -352 1248 WIRE -352 1296 -352 1248 WIRE -352 1456 -352 1376 FLAG 352 1168 0 FLAG 752 1168 0 FLAG 1136 1168 0 FLAG 32 1168 0 FLAG 480 1168 0 FLAG 880 1168 0 FLAG 1264 1168 0 FLAG 1488 1168 0 FLAG 1616 1168 0 FLAG -384 528 0 FLAG 2176 592 Out IOPIN 2176 592 Out FLAG -352 1456 0 SYMBOL res 16 80 R0 SYMATTR InstName R1 SYMATTR Value 5.2k SYMBOL res 16 224 R0 SYMATTR InstName R2 SYMATTR Value 200 SYMBOL res 16 464 R0 SYMATTR InstName R3 SYMATTR Value 375 SYMBOL res 16 656 R0 SYMATTR InstName R4 SYMATTR Value 330 SYMBOL res 16 944 R0 SYMATTR InstName R5 SYMATTR Value 800 SYMBOL npn 240 160 R0 SYMATTR InstName Q1 SYMATTR Value 2N3904 SYMBOL pnp 416 304 M180 SYMATTR InstName Q2 SYMATTR Value 2N3906 SYMBOL res 336 1008 R0 SYMATTR InstName R6 SYMATTR Value 33k SYMBOL npn 640 352 R0 SYMATTR InstName Q3 SYMATTR Value 2N3904 SYMBOL pnp 816 496 M180 SYMATTR InstName Q4 SYMATTR Value 2N3906 SYMBOL res 736 1008 R0 SYMATTR InstName R7 SYMATTR Value 33k SYMBOL npn 1024 544 R0 SYMATTR InstName Q5 SYMATTR Value 2N3904 SYMBOL pnp 1200 688 M180 SYMATTR InstName Q6 SYMATTR Value 2N3906 SYMBOL res 1120 1008 R0 SYMATTR InstName R8 SYMATTR Value 33k SYMBOL npn 1376 784 R0 SYMATTR InstName Q7 SYMATTR Value 2N3904 SYMBOL pnp 1552 928 M180 SYMATTR InstName Q8 SYMATTR Value 2N3906 SYMBOL res 1472 1008 R0 SYMATTR InstName R9 SYMATTR Value 33k SYMBOL res 1856 816 R90 WINDOW 0 0 56 VBottom 2 WINDOW 3 32 56 VTop 2 SYMATTR InstName R10 SYMATTR Value 10k SYMBOL res 1856 576 R90 WINDOW 0 0 56 VBottom 2 WINDOW 3 32 56 VTop 2 SYMATTR InstName R11 SYMATTR Value 2.7k SYMBOL res 1856 384 R90 WINDOW 0 0 56 VBottom 2 WINDOW 3 32 56 VTop 2 SYMATTR InstName R12 SYMATTR Value 800 SYMBOL voltage -384 336 R0 SYMATTR InstName V1 SYMATTR Value 10 SYMBOL res 1920 960 R0 SYMATTR InstName R13 SYMATTR Value 800 SYMBOL voltage -352 1280 R0 SYMATTR InstName V2 SYMATTR Value PULSE(0 3.3 0 0.5m 0.5m 0 2m 0) TEXT -256 912 Left 2 !.tran 10m TEXT 304 1344 Left 5 ;Positive half of ICL8038 sine shaper, with discretes

The LM13700 is a good candidate.

Cheers

Phil Hobbs