|
Homebrew Equipment A place to show, design and discuss the weird and wonderful electronic creations from the hands of individual members. |
|
Thread Tools |
14th Aug 2021, 10:29 am | #41 |
Nonode
Join Date: Jun 2016
Location: Bognor Regis, West Sussex, UK.
Posts: 2,288
|
Re: Low Distortion Oscillator output puzzle
I have now received some non polarised electrolytics from the HiFi collective (thanks Craig for the link) and will use one of these, at least initially, when the boards arrive.
Peter |
15th Aug 2021, 9:29 am | #42 |
Hexode
Join Date: Apr 2007
Location: W Yorks, UK.
Posts: 406
|
Re: Low Distortion Oscillator output puzzle
|
15th Aug 2021, 5:10 pm | #43 |
Dekatron
Join Date: Nov 2010
Location: Oxford, UK.
Posts: 4,941
|
Re: Low Distortion Oscillator output puzzle
It is an unrepresentative set of curves - no one would put 10V rms at low frequencies into a 1uF electrolytic feeding into 1k.
Bateman's results show the effect of bias voltage on distortion for three types of 100uF 25V electrolytics: "With a 0.3 volt test signal, measurement noise floor improved to -123 dB but the PET reference capacitor harmonics remain buried in noise. Second and third harmonics of the polar capacitors are now clearly visible, their distortions having increased much faster than the test signal level. With a 0.3 volt test signal and no bias, the Silmic, at 0.00098%, produced the least distortion of the three electrolytics. It’s second harmonic measured -100.6 dB, Black Gate -98.5 dB and YXF -89.1 dB. This is the best electrolytic of those I tested with no bias, however it still produced more than three times the distortion of the PET assembly. Fig.7 With 6 volt DC bias, the Silmic and Black Gate, with second harmonics around -90 dB, produced similar 0.003% distortion. The YXF second harmonic was -87.3 dB for 0.0043% distortion. With 18 volt DC bias the Black Gate develops fifteen times more distortion than the PET assembly but now distorts less than the other two electrolytics. Its second harmonic at -84.1 dB was some 3 dB better than the Silmic and 4 dB better than the low cost YXF type. Distortions now measured 0.00637%, 0.00840% and 0.00951% respectively." |
16th Aug 2021, 9:48 am | #44 |
Hexode
Join Date: Apr 2007
Location: W Yorks, UK.
Posts: 406
|
Re: Low Distortion Oscillator output puzzle
The value doesn't matter; the same general effect exists with bigger values, more capacitance just pushes the traces off to the left.
|
17th Aug 2021, 10:25 pm | #45 |
Pentode
Join Date: May 2015
Location: Hemel Hempstead, Hertfordshire, UK.
Posts: 132
|
Re: Low Distortion Oscillator output puzzle
Here is a link to an ultra low distortion oscillator.
https://quantasylum.com/blogs/news/notch-filters I hope you all find it interesting. |
18th Aug 2021, 9:38 pm | #46 |
Octode
Join Date: May 2009
Location: Linkoping, Sweden
Posts: 1,463
|
Re: Low Distortion Oscillator output puzzle
How does the Wien Bridge oscillator designed by Jim Williams, can be found here: https://www.analog.com/media/en/tech...otes/an43f.pdf on page 33, Figure 48, compare to the designs discussed in this thread?
__________________
Martin, Sweden |
18th Aug 2021, 11:05 pm | #47 |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,800
|
Re: Low Distortion Oscillator output puzzle
Jim Williams' design is right in there with them. His common mode suppression technique targets the remaining most significant mechanism creating distortion. To go a bit further it would be worth buffering the feed to the detector diodes to remove non-linear loading from the oscillator loop itself.
Notice the use of an LED/CdS photoconductor as the gain adjusting device. Much more linear than attempts with FETs But at low frequencies, the detector will progressively start to follow the signal and this will self-AM the oscillator amplitude == even order distortion. HP went over to a state-variable filter as the tuning mechanism rather than the wien network. For noise content issues, the Q of the frequency determining network becomes important. Good RF designs can hit Qs of a few hundred. and this translates into faster cutoff of noise sidebands. The poor old Wien bridge has a calculable Q of one third if my memory is right. David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
19th Aug 2021, 6:45 am | #48 |
Nonode
Join Date: Jun 2016
Location: Bognor Regis, West Sussex, UK.
Posts: 2,288
|
Re: Low Distortion Oscillator output puzzle
The pcbs I refered to in Post #1 have arrived. I will assemble a single frequency oscillator at 1kHz to test the circuit before making the full variable frequency version.
I will report results but I may be delayed a tad due to getting ready for my Son's wedding. Peter |
19th Aug 2021, 7:08 am | #49 | |
Dekatron
Join Date: Nov 2010
Location: Oxford, UK.
Posts: 4,941
|
Re: Low Distortion Oscillator output puzzle
Quote:
Which uses the LED/CdS technique. Fixed frequency. The oscillator board, and a twin-T notch to verify performance was the princely sum of Euro29. Distortion <-150dB, which is tricky to verify even with a finely adjusted notch to get rid of the fundamental. Craig Haven't built it yet. |
|