Function Generator Sinewave distortion?

Can't help; with that part, but what would hapen if you just built a very sharp filter to filter out the sine wave ? Then looked at what ever was left.

Ideas:

1. Use twin-T notch filter to remove the fundamental so you can see all the harmonics on a scope. Piece of cake to make one.
2. Use the FFT function on a DSO to look at the harmonics.
3. Use both above together.
4. Set the generator to about 500Hz and use your ears - it is usually easy enough to put the speaker in spot that almost nulls the fundamental where you are sitting making the harmonics more obvious.

.... Phil

Isn't that how a distortion analyzer is made? I dont' recall needing an oscillator.

Michael

If there's a bunch of pots, likely they are generating a triangle wave (easy to do) and then using a set of diodes to "sculpt" it into a sinewave. At the best of times, it will not be a perfect sinewave.

HOw much fussing depends on how much you need the sinewave, and even then, I think there will be more distortion than if you set out with an oscillator that generated a sinewave directly. Function generators are for general purpose, an oscillator with a wide range of frequency and different output waveforms. But if you need a good sinewave, you look elsewhere.

Maybe stuff about the 8038 would give some ideas about adjusting the sinewave. It generated a triangle and then used diode-resistor shaping network, but there was little external adjustment of the waveform. But the old articles might have given some hints on getting a better sinewave.

As someone pointed out, if you need a good sinewave and it's a relatively fixed frequency, you could just pass it through a low pass filter, likely clear up things relatively well.

Michael

Since you are writing this on a PC, you already have the instrument of your dreams.

Connect the "sinus" to the sound card, run a free FFT software and - bingo!

w.

(easy to do)"

Are you thinking Wavetek ? I lik those. They claim to do 0.5 % harmonic dis tortion. I think tha tis some damn good engineeriong myself. It is not just diodes, it is also using the knee of the beta curve of amost common bas st ages. I have a Wavetek 111, and I thihnk for being back in the 1960s, the t hing is fantastic. for what they diud at the time I tink they were like Tek tronix. Not as complex but just as anal about certain things.

" and then using a set of diodes to "sculpt" it into a

The Wavetek 1XX series used three stages. Emitter fed by the triangle throu gh the diodes, each biased a bit differently. Totally push pull, positive a nd negative sides matched to a "T". Those pots adjusted the parts of a sine wave that the circuit acted upon, and it acted upon it. Wavetek found a wa y to make it a bit better. It hadd negative current feedback and brought th e low current hfe of the transistors into play. It has it just coming into conductance, and nulling it. there is a set of precision resistors that if not right, you would NEVER be able to align it down to 0.5 % harmonic disto rtion. Ever.

My buddy has a 103 that has a distortyd sine output. I found it to have a D C offset From the triangle output to the sine convertor. This is consistent with the fact that his "sine" is compressed at the top and ponty at the bo ttom. I temporarily fixed it after gettijng aggeavated working on it by put ting a blocking capacitor between the triangle generator and the sine conve rtor. It worked fine but he decided to be anal and get it workin right so h e took it apart and now we have a basketful. He frigged someting up and now it won't even oscillate.

That's OK. It is his. He did it. No problem.

Anyway, I was never impressed by generators tha tneed AGC and all that to s tay flat response. And thought I REALLY respect GOOD analog design, when it comes to this, digital is probably good enough now. Although I will not se l my old Wavetek, thsat doesn't mean I don't want something else.

See how we are ?

Some have it included, others do not. And some that do have it included give you the option of using an external oscillator.

Most are purely manual while more expensive models adjust the notch frequency automatically.

My home brew, Wien Bridge notch filter job has three fixed frequencies ( 66, 998 and 5280 Hz) plus two 10 turn WW pots to trim the notch frequency and depth. With a little patience, it can resolve down to 0.002%.

... Phil

Hear, hear!

The dynamic range is more than 120dB and most function generators using a triangular wave will be around 60dB from memory.

Either use your soundcard and run the software yourself, or get FFT programs.

I was impressed with the simplicity of the GUI done by Bob:

Bob Masta

DAQARTA v7.60 Data AcQuisition And Real-Time Analysis

Notice he now has an LCR meter built-in! shows you Resr of caps.

Thanks for the plug! The OP can probably do his calibration for free, before the 30-session / 30-day trial expires. (I say "probably", because I don't know if there is a good strategy for adjusting a multiple-diode wave shaper; I seem to recall reading a *design* article back in the Bad Old Days, though.)

Anyway, after the trial period, Daqarta's signal generator keeps working, forever. (Plus a whole bunch of other stuff, but not external signal *inputs*.) So even if he can't get his old signal generator working he'll have a replacement that not only produces lower sine distortion, but has a lot of exotic modulation options. 2 channels are standard, with up to 4 independent streams per channel. (Sneak preview: Daqarta v8.00 will allow 8 independent output channels.)

One shortcoming of sound cards, however, is that they are all AC-coupled so you can't get decent results below a few Hz without circuit modification. (Some high-end cards seem to claim DC coupling, but it's an "audiophile lie": They just mean that there are no capacitors in the signal path. They use a servo circuit instead, but still have an AC-coupled high-pass response.)

Regarding sine-from-triangle generators, I think the average function generator is around 1% (-40 dB). The "overdriven differential pair with feedback" approach can do that easily, and as Phil Allison notes it is easy to adjust by ear. You have only 2 adjustments, and once you are close it comes down to trading between 2nd harmonic and 3rd harmonic, both pretty audible if your fundamental is in the 100-500 Hz range. I have seen one (ancient) article where by careful design and oven control of the differential pair, they got down to -60 dB.

Best regards,

Bob Masta DAQARTA v7.60 Data AcQuisition And Real-Time Analysis

Interesting ideas. So what twin-T, RC values would you recommend for a notch frequency of 5Khz to observe distortion down to say 0.3%? . .

--- news://freenews.netfront.net/ - complaints: snipped-for-privacy@netfront.net ---

Plenty of info on the www about Twin-Ts.

You will need one op-amp, 3x 6.8nF caps and 3 x 4.7kohm resistors.

F = 1 / (2.pi.R.C)

... Phil

You will need an op-amp, 4 x 6.8nF caps and 4 x 4.7kohm resistors. Two caps and two resistors have to be paralleled to get R/2 and 2C values.

F = 1 / (2.pi.R.C)

... Phil

Yes, they use 15 resistors and 10 diodes. There are 3 voltage dividers of 5 resistors each and wierd values. The center divider has the triangle input which is already calibrated for symetry and offset. 5 diodes on one side are used for half the waveform and the other 5 diodes for the other half. There are 4 pots to set the DC voltage at each end of the 2 dividers. And a couple more pots for other minor adjustments. I think this thing does less than

--- news://freenews.netfront.net/ - complaints: snipped-for-privacy@netfront.net ---

Why the op-amp? This generator puts out 20 volts p-p open circuit, or 10 volts p-p into 50 ohms. Why not just feed the generator into a passive twin-T notch filter and look at the other end of the filter?

--- news://freenews.netfront.net/ - complaints: snipped-for-privacy@netfront.net ---

... Phil