I cobbled together a discrete wien bridge oscillator using a Jfet and bipolar and a xmas tree light for AGC.
I get a reasonable looking sinewave over 2 decades and wondering what the distortion (THD) might be. Looking at one cycle on a scope, I see about 20 minor horizontal divisions on the down slope compared to 18 on the rising slope. So the symmetry is not perfect and some distortion is indicated.
What would you estimate the distortion to be, and how could it be improved?
The lamp is not doing much as emitter resistor. The DC is responsible for the resistance, and that should be ac only. The lamp should reduce the gain, when the ac voltage is to high, so it should be the top R in a 2R divider. That way the feedback voltage goes lower, and distortion decreases.
The ac voltage at the second transistor should be less than 30% of saturation, or even lower if you can make it oscillate at that level. Try to get a small pilot lamp for 24-48V, it will give you a better range of resistance for regulating. Btw, does that fet not need a bit of bias voltage?
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D Grab 128 or 256 samples of the waveform at 12 or more bits and post a table here. Go get a cup of coffee. You will get the FFT and THD analysis within minutes.
Back in the days when I used to build triangle-to-sine shapers, I found that I could hear 2nd and 3rd harmonic distortion better than I could see it on a waveform. (At least when there were adjustments to tweak, it was easier to find the best compromise.) Visual distortion usually was over 1% THD.
But nowadays you can do *much* better with a spectrum analyzer that uses your Windows sound card, at least if your signal is in the audio range. Even with a cheap built-in sound card, you can easily see distortion peaks 80 dB below the fundamental.
You are welcome to try my Daqarta package free for 30 sessions/30 days. It's only US$29 to buy, but if you finish up your project within the trial period and don't have a further need to look at input signals, you can still keep using the Daqarta signal generator (plus a lot of other features) forever, FREE.
Enjoy!
Bob Masta DAQARTA v5.10 Data AcQuisition And Real-Time Analysis
formatting link
Scope, Spectrum, Spectrogram, Sound Level Meter Frequency Counter, FREE Signal Generator Pitch Track, Pitch-to-MIDI DaqMusic - FREE MUSIC, Forever! (Some assembly required) Science (and fun!) with your sound card!
Yes, I missed that. I was switching the thing on and off and noticing it powered up with a distorted wave and then settled down as the lamp warmed up, and concluded the agc was working. Thanks for pointing that out. But it seems to require a few extra parts to drive the lamp from AC. Another idea I had was to use a cadmium sulfide cell in the emitter circuit and a LED to adjust the gain. But that would also require more parts to rectify and filter the AC and drive the LED. I'm trying to keep it simple. I've seen some circuits on the web using an op-amp with the lamp in the AC feedback loop, but it requires a sensitive lamp using only a few milliamps. I wanted to use a Xmas tree light since I have hundreds of them. The fet is operating on self bias. The vgs is about 3 volts negative, so with the gate grounded, the fet turns on until the source voltage rises to about 3 volts which yields about negative 3 volts on the gate relative to the source.
Ok. Then I would give the second stage a variable gain fot the time being, like R-collector ~1Kohm, and the R-emittor a potmeter of 0 to 1k. AC couple the collector to a emmitorfollower with a R-emmmitor of ~50 ohm and a bias of 6 V, and couple that AC to the lamp/R divisor, and feed that back to the first stage. Slowly increase the gain of stage 2, from 1(R-emm = 1k). If you choose the lamp divisor initially as lamp and R of equal value, you should get oscillation at about a gain of 2-4 of stage 2. Replace the potm. then with a resistor of the next lower value. I wonder what the cold resistance of the lamp is, it should be about
100-400 ohm, as high as you can get it. Thats why I advised the 24-48V pilot lamp. They are very tiny, and use but a little current, so their cold resistance is rather high. If you have an osciloscope, check that none of the stages is near saturation, it would be nice, when the voltage on the lamp/r divider is 1-3 volt, lower is better for your distortion. If you succeed to reach those goals, you will have a very nice oscillator.
Yes, that's an idea, I tried an emitter follower and a cap from emitter to the lamp/resistor and feeding back the junction of lamp/R. I found a small lamp with 30 ohms cold resistance and about 65 ohms at
7 milliamps. But the thing won't start with such a low R connected to the feedback network. It starts with the network connected to the collector and the lamp comes on with low intensity indicating the AC voltage is lighting the lamp. Might work if there was a way to start the circuit and then switch to run mode, but that's too complicated. I'll look around for a higher resistance lamp and try it again.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.