Yeah, This is a state-variable oscillator which are not known for low distortion. Wien bridge oscillators with a good amplitude control can be much better. Jim Williams discovered this in his paper “Max Wien, Bill Hewlett and a rainy Sunday afternoon”. He tried to get better and better distortion figures by using different amplitude control circuits. When he got to the J-FET he found that the control voltage modulated its channel length which limited his efforts to improve it. I also found this but did better eventually with a homemade voltage-controlled resistor made from an incandescent lamp and a Cadmium-Sulphide photocell pushed into opposite ends of piece of small tubing. I got the THD down to 0.00018% at best using a notch filter to see the residue and measured on a HP 3580A spectrum analyzer.
I purchased the same unit you have but I got the assembled one. Interesting mine have multiturn trim pots for R6 and R7. I bought it in 2023. They are using those two pots as rheostats for the R values. Using a +/- 15 volt power supply and the QA403 24-bit Audio Analyzer I measured a frequency for the sinewave of 996 Hz. The THD was -82.00 dB or 0.00794%. So I would rate this as a very good low distortion sinewave oscillator. Maybe you have a defective part or possibly an error in assembly of your unit. Or the one they build, they trim R6 and R7 for best performance.
I did the same measurements, THD of 101dB, and THD+N of 93dB. Replacing the op-amp with OPA1612 improved the THD by 2dB The power supply is critical to improve THD+N
Classical Wien Bridge oscillator with incadecent lamp in feadback gives me 0.0017 % THD. Distortions starts at -96dB. All "measurments" was made with Visual Analyser software and creative SB soundcard.
Why not? I have proof, video can be searched by keywords " Creative SB0100 THD & Noise floor. How good Sound Blaster Live! in record mode " And sory about my english
VR adjusts that JFETs linearity as a resistor. What the circuit seems to be is a high Q (high enough to self oscillate) state variable filter. J1 controls Q of the filter through that peak detector around the diodes and TL071. Matching R6 and R7 should help, same with C1 and C2 because they control the frequency and those two integrators should be as close to each other as possible. After that adjusting R13 or R12 might lower THD (by lowering the overall output level that starts lowering Q).
In Linear Technology Application Note AN132 - Fidelity Testing for A to D Converters, Jim Williams and Guy Hoover present a low-distortion oscillator circuit that eliminates or reduces the distortion that is present in several features of the eBay circuit. For instance, the FET that is a source of nonlinearity is replaced with a cadmium sulfide cell that is illuminated by an LED, and the LED bias source is designed to contain very low ripple from the oscillator. 🙂
@@IMSAIGuy Okay. I presume the authors used the very best available from Linear Technology at the time, and those remain expensive, particularly since there is very limited production volume. Fortunately, there are modern high-performing audio op amps that are much less expensive that could be substituted for A1 through A3, and the remaining op amps could be any of a wide variety of modern inexpensive parts.
@@analog_guy How do you know there are exact, or good enough, part substitutions. Sometimes it isn't possible, even using the specified parts, to get to the spec without binning.
@@willthecat3861 I read the AN135 document, I considered the differences in the design techniques used in the two schematics, and I looked at the data sheets for some op amps. In this case, there was no firm spec given for the oscillator, but IMSAIGuy recognized that the distortion performance was disappointing and the circuit did not meet his expectation. Even run-of-the-mill op amps should deliver better performance when used in a well-designed circuit. Good design techniques can often mitigate issues that would otherwise result from component variations and nonlinearities. In this case, presuming the component values match the schematic, I suspect that the gain-control FET that was used has a different I-V characteristic from the one used when the circuit was first developed, but eliminating that nonlinear device altogether, as was done in AN135, gets rid of that particular problem and yields better performance without resorting to binning. You can also see the comment by @basinstreetdesign5206 for more specific information.
I use an app on my iPad to generate clean audio range sine waves on the headphone output. Since it is not referenced to earth, there is no risk of any earth loops.
I built something similar (found it in the Dutch electronics magazine "Elektuur", sometime in the '80s). It was originally designed to be adjustable from about 20 Hz to 20KHz, but I extended the range from about 1.5 Hz to 120 KHz. It has low distortion, around 0.01% at 1 KHz. The Elektuur design uses one Quad TL084. The circuitry to control the gain was better than this design, resulting in lower harmonics.
The capacitors on the board that control the oscillation frequency are selected and paired using a digital meter. The resistors that control the oscillation frequency are a 3.3K resistor and a 200 ohm 3296 multi-turn potentiometer in series. The pairing is measured and adjusted using a digital meter to ensure the accuracy and low distortion of the output frequency.
@@willthecat3861 that "any video" part was just rude.. Dude uploaded 2k+ videos of him actually doing stuff that is actually practically teaching in a very timesaving format compared to e.g. yapping for an hour long video of the same thing over and over on modern eevblog.. There is plenty of quality info on the channel, and only parts of low effort I've encountered here are "cheat day" lazy _real-time_ assembly videos of moderate level solder work, that's it..
@@Mr.Leeroy well you said more on my part than I said for myself. It's my opinion, after watching since The Guy started. If you are learning stuff great. That doesn't mean everyone else is, or isn't. Everyone is different.
From the view on your FFT I just estimated that the first harmonic was 50-60dB below the fundamental, which agrees quite well with a few 0.1% of THD. For a circuit board which is populated with probably not selected and paired components, even mixing in 5% tolerance carbon film resistors and ceramic capacitors, I would say that 0.3% THD is quite decent. To compare it to a 8000+ dollar modern scope, which few amateurs ever will be able to afford even though almost every youtuber seems to be awarded by Keysight with this beast lately, is more than unfair. But yes, a Wien oscillator as pointed out in other comments, would probably have a better performance at the same price level.
Quick question... Are R8, R9 & VR the correct values? I can't really see the part numbers on the PCB as the part probably covers them but I think I'm seeing 10k, 10k and 0.5R. Not likely I know and maybe the colours aren't translating well (or I'm just thick! ;) Thanks... Clark FYI: I'm looking at 8:41 & 9:33
Low distortion coefficient sine wave oscillator circuit board, oscillation frequency: 1 KHz, ultra-low distortion, distortion coefficient 0.000064% (-124dB) line from "Transistor Circuit Design and Production" (Japan) written by Toru Kuroda.
No doubt. Mind you 10% distortion is intolerable to listen to, it is the norm for car audio when used at maximum power output. At least, the audio chips I've seen that are designed for simpler automobile systems seem to be rated like that.
@@MinceWalsh I was not talking the electrical distortion, but the mechanical one from the speaker chassis. You need to be very good in (air)fluid design to bring that down. I remember something like low single digit percentage with a horn for impedance match. But some gold ears hear 1/100% electrical difference.
What if, forgive me my ignorance, if you build two of the same audio oscillators , but one of the two is delayed, the outputs are tied together. The delay is related to the frequency, and it is always one complete phase behind, the harmonics would then be out of phase and therefore always less then what it used to be.
@@IMSAIGuy The Total Harmonic Distortion goes down when the signal of the oscillator is lead through a delay and added to the original signal, the two true sinussoes add up when in phase, but all distortions and harmonics will be out of phase and therefore be less then before. To be frankly honest I have no clue how to delay the signal for exactly one complete phase, others can do that. I should have studied harder. But reducing the distortion is possible, in the seventies there was an article in Elektuur (a Dutch electronics monthly) a low-noise pre-amplifier for a gramophone, they used parallel ten cheap transistors to amplify in the first step. Every transistor had his own hisses and hums, but all hisses and hums were out of phase to each other and did not add up. I am confidant that you are able to reduce the THD, I can't do it myself, I am only an admirer of the true geniuses.
@@IMSAIGuy I have basic knowledge in electronics. I know what components do, how they work and why. What I'm missing is someone who can teach me more and with their knowledge show the most common mistakes you make / think as a newbie in their construction of something. My question is can you not make a series of videos where you go through different circuits, show why not to do this or that. I really miss this kind of stuff on YT. What I find is the anitingen at too low a level for me or on the contrary, at too high so I do not understand at all.
you have it wrong. It's a $25,000 scope. and yes I'm as poor as us. Keysight just gave me the thing. amazing. I guess the moral to the story is if you do good things you get good things.
You need a decent pairing of capacitors and resistors in the upper part of the circuit, matching R6,R7; C1, C2; R1, R2 A guy from Czech Republic has done a lot of work on this sort of circuit; he even sells kits of his adapted design. He has a PDF of schematic and a lot of commentary as well, you should be able to find it if you search "An ultra low-distortion oscillator with THD below -140 dB" Part of it is building an equally decent notch-filter just to be able to see any residual spurs! A decent modern replacement OPAMP to make these kind of circuits with should be OPA1656 by Ti, it won't break the bank at least. Harder component to find is a really good linear optocoupler.
Matching and pairing R6, R7, C1 and C2 is rather pointless. The only result of unmatched compts is a slight difference in amplitude of the outputs of the 2 integrators, which won't affect anything. One integrator generates a cos signal, the other a sin signal. If you display them as X and Y on a scope, you might be able to see any non circularity which represents distortion. I would expect most of the distortion arises from the auto levelling part of the cct acting on the JFET.
Yeah, This is a state-variable oscillator which are not known for low distortion. Wien bridge oscillators with a good amplitude control can be much better. Jim Williams discovered this in his paper “Max Wien, Bill Hewlett and a rainy Sunday afternoon”. He tried to get better and better distortion figures by using different amplitude control circuits. When he got to the J-FET he found that the control voltage modulated its channel length which limited his efforts to improve it. I also found this but did better eventually with a homemade voltage-controlled resistor made from an incandescent lamp and a Cadmium-Sulphide photocell pushed into opposite ends of piece of small tubing. I got the THD down to 0.00018% at best using a notch filter to see the residue and measured on a HP 3580A spectrum analyzer.
I purchased the same unit you have but I got the assembled one. Interesting mine have multiturn trim pots for R6 and R7. I bought it in 2023. They are using those two pots as rheostats for the R values. Using a +/- 15 volt power supply and the QA403 24-bit Audio Analyzer I measured a frequency for the sinewave of 996 Hz. The THD was -82.00 dB or 0.00794%. So I would rate this as a very good low distortion sinewave oscillator. Maybe you have a defective part or possibly an error in assembly of your unit. Or the one they build, they trim R6 and R7 for best performance.
I did the same measurements, THD of 101dB, and THD+N of 93dB. Replacing the op-amp with OPA1612 improved the THD by 2dB The power supply is critical to improve THD+N
Classical Wien Bridge oscillator with incadecent lamp in feadback gives me 0.0017 % THD.
Distortions starts at -96dB. All "measurments" was made with Visual Analyser software and creative SB soundcard.
-96dB would be the base dynamic range of your Creative SB soundcard -- in other words, it's undetectable with your equipment.
Why not? I have proof, video can be searched by keywords " Creative SB0100 THD & Noise floor. How good Sound Blaster Live! in record mode "
And sory about my english
JurekOK may very well be right. How about using one of those 1-bit DSD recorders? They can sample at around 10 MHz if I remember correctly.
I too have a design and comparable numbers, with my incandescent based oscillator.
Based on GPT Quadrature (Sine-Cosine) Oscillator can go down to 0.00003% THD, lot lower than Wien
VR adjusts that JFETs linearity as a resistor. What the circuit seems to be is a high Q (high enough to self oscillate) state variable filter. J1 controls Q of the filter through that peak detector around the diodes and TL071. Matching R6 and R7 should help, same with C1 and C2 because they control the frequency and those two integrators should be as close to each other as possible.
After that adjusting R13 or R12 might lower THD (by lowering the overall output level that starts lowering Q).
Ah, but your assumption abut using good components is the issue. The circuit was designed only to work with crappy components.
I built one it works if all you need is a quick input signal. Fun to assemble. That’s pretty much it.
In Linear Technology Application Note AN132 - Fidelity Testing for A to D Converters, Jim Williams and Guy Hoover present a low-distortion oscillator circuit that eliminates or reduces the distortion that is present in several features of the eBay circuit. For instance, the FET that is a source of nonlinearity is replaced with a cadmium sulfide cell that is illuminated by an LED, and the LED bias source is designed to contain very low ripple from the oscillator. 🙂
I looked at that app note and thought I would make one. oh. ouch. very expensive opamps in that thing. I don't need it that bad.
@@IMSAIGuy Okay. I presume the authors used the very best available from Linear Technology at the time, and those remain expensive, particularly since there is very limited production volume. Fortunately, there are modern high-performing audio op amps that are much less expensive that could be substituted for A1 through A3, and the remaining op amps could be any of a wide variety of modern inexpensive parts.
@@analog_guy How do you know there are exact, or good enough, part substitutions. Sometimes it isn't possible, even using the specified parts, to get to the spec without binning.
@@willthecat3861 I read the AN135 document, I considered the differences in the design techniques used in the two schematics, and I looked at the data sheets for some op amps. In this case, there was no firm spec given for the oscillator, but IMSAIGuy recognized that the distortion performance was disappointing and the circuit did not meet his expectation. Even run-of-the-mill op amps should deliver better performance when used in a well-designed circuit. Good design techniques can often mitigate issues that would otherwise result from component variations and nonlinearities. In this case, presuming the component values match the schematic, I suspect that the gain-control FET that was used has a different I-V characteristic from the one used when the circuit was first developed, but eliminating that nonlinear device altogether, as was done in AN135, gets rid of that particular problem and yields better performance without resorting to binning. You can also see the comment by @basinstreetdesign5206 for more specific information.
you need a Masters degree to understand this comment.
One man's low-distortion is another one's trash. Thanks for the video.
I use an app on my iPad to generate clean audio range sine waves on the headphone output. Since it is not referenced to earth, there is no risk of any earth loops.
I built something similar (found it in the Dutch electronics magazine "Elektuur", sometime in the '80s). It was originally designed to be adjustable from about 20 Hz to 20KHz, but I extended the range from about 1.5 Hz to 120 KHz. It has low distortion, around 0.01% at 1 KHz. The Elektuur design uses one Quad TL084.
The circuitry to control the gain was better than this design, resulting in lower harmonics.
The Ebay listing for this board claimed a distortion level of 0.0064%.
Wow...
a copy cat gotcha - good use of the Keithley 🤔saves the vid for me
The capacitors on the board that control the oscillation frequency are selected and paired using a digital meter. The resistors that control the oscillation frequency are a 3.3K resistor and a 200 ohm 3296 multi-turn potentiometer in series. The pairing is measured and adjusted using a digital meter to ensure the accuracy and low distortion of the output frequency.
Way more work than The Guy, will probably put into any video. It's only show and tell level, for this kind of video.
@@willthecat3861 that "any video" part was just rude..
Dude uploaded 2k+ videos of him actually doing stuff that is actually practically teaching in a very timesaving format compared to e.g. yapping for an hour long video of the same thing over and over on modern eevblog.. There is plenty of quality info on the channel, and only parts of low effort I've encountered here are "cheat day" lazy _real-time_ assembly videos of moderate level solder work, that's it..
@@Mr.Leeroy well you said more on my part than I said for myself. It's my opinion, after watching since The Guy started. If you are learning stuff great. That doesn't mean everyone else is, or isn't. Everyone is different.
@@willthecat3861 What do you watch it for all those years?
Made of nothing but the best Chineesium.
From the view on your FFT I just estimated that the first harmonic was 50-60dB below the fundamental, which agrees quite well with a few 0.1% of THD. For a circuit board which is populated with probably not selected and paired components, even mixing in 5% tolerance carbon film resistors and ceramic capacitors, I would say that 0.3% THD is quite decent. To compare it to a 8000+ dollar modern scope, which few amateurs ever will be able to afford even though almost every youtuber seems to be awarded by Keysight with this beast lately, is more than unfair.
But yes, a Wien oscillator as pointed out in other comments, would probably have a better performance at the same price level.
ruclips.net/video/oOFYrZH1ECw/видео.htmlsi=LZIUN4FbgF7QcQRN
Quick question... Are R8, R9 & VR the correct values? I can't really see the part numbers on the PCB as the part probably covers them but I think I'm seeing 10k, 10k and 0.5R. Not likely I know and maybe the colours aren't translating well (or I'm just thick! ;) Thanks... Clark
FYI: I'm looking at 8:41 & 9:33
By the comments, eager to see Part2. ;)
I built one as a Chinese kit and got .025%
I have my 36k$ APx 555 to design my audiophile amplifier 😊😊.
Low distortion coefficient sine wave oscillator circuit board, oscillation frequency: 1 KHz, ultra-low distortion, distortion coefficient 0.000064% (-124dB) line from "Transistor Circuit Design and Production" (Japan) written by Toru Kuroda.
where did you found this book? I can't :(
Please give a link it is imposible to find...
FET attenuation has some distortion .
It would be interesting to measure the THD, using the same opamps but with better quality: the issue should be there.
❤
Compared to the 10% of a typical speaker, 0,3% distortion is fine.
No doubt. Mind you 10% distortion is intolerable to listen to, it is the norm for car audio when used at maximum power output. At least, the audio chips I've seen that are designed for simpler automobile systems seem to be rated like that.
don't tell that to the audiophools....
@@MinceWalsh
I was not talking the electrical distortion, but the mechanical one from the speaker chassis. You need to be very good in (air)fluid design to bring that down. I remember something like low single digit percentage with a horn for impedance match.
But some gold ears hear 1/100% electrical difference.
That's not considered low distortion! Even a simple twin T oscillator using an op amp can easily hit fractional three digits distortion.
@@jim9930That little r19 thermistor can still beat the pants off all but the best digital stuff even today!.
The real ultra low THD ones are designed, made and sold by Viktors on eee bay.
What is the output frequency of the board? The harmonic distortion test should be checked according to the test frequency before testing.
The keithley measures the frequency and harmonics automatically
What if, forgive me my ignorance, if you build two of the same audio oscillators , but one of the two is delayed, the outputs are tied together. The delay is related to the frequency, and it is always one complete phase behind, the harmonics would then be out of phase and therefore always less then what it used to be.
THD stands for TOTAL harmonic distortion. ALL harmonics are summed, no matter what frequency or phase.
@@IMSAIGuy The Total Harmonic Distortion goes down when the signal of the oscillator is lead through a delay and added to the original signal, the two true sinussoes add up when in phase, but all distortions and harmonics will be out of phase and therefore be less then before. To be frankly honest I have no clue how to delay the signal for exactly one complete phase, others can do that. I should have studied harder. But reducing the distortion is possible, in the seventies there was an article in Elektuur (a Dutch electronics monthly) a low-noise pre-amplifier for a gramophone, they used parallel ten cheap transistors to amplify in the first step. Every transistor had his own hisses and hums, but all hisses and hums were out of phase to each other and did not add up. I am confidant that you are able to reduce the THD, I can't do it myself, I am only an admirer of the true geniuses.
I think you should use a wireless mouse with the new keysight scope.
I have one on it now
How to mesurment audio Amplifiers damping factor and THD
I have built a lamp oscilator, it was not stable, never again.
Instead of giving a 5 minute critique of the circuit, you could have talked and shown how you would have built it to make it really good.
you might like this: ruclips.net/video/oOFYrZH1ECw/видео.htmlsi=NivgK1ZV0i1NiFIo
@@IMSAIGuy I have basic knowledge in electronics. I know what components do, how they work and why. What I'm missing is someone who can teach me more and with their knowledge show the most common mistakes you make / think as a newbie in their construction of something. My question is can you not make a series of videos where you go through different circuits, show why not to do this or that. I really miss this kind of stuff on YT. What I find is the anitingen at too low a level for me or on the contrary, at too high so I do not understand at all.
...and R1, R2 &R6 seem to be orange / orange / red / red / white, what's that? My novice mode / screen might be betraying me...
carbon film 3.3 kΩ ±2% - but I don't know what the white ring means...
what are you doing with a $15,000.00 scope? i thought you were as poor as us. lol
you have it wrong. It's a $25,000 scope. and yes I'm as poor as us. Keysight just gave me the thing. amazing. I guess the moral to the story is if you do good things you get good things.
@@IMSAIGuy :)
@@IMSAIGuylove that they comp'ed you a scope.
You need a decent pairing of capacitors and resistors in the upper part of the circuit, matching R6,R7; C1, C2; R1, R2
A guy from Czech Republic has done a lot of work on this sort of circuit; he even sells kits of his adapted design. He has a PDF of schematic and a lot of commentary as well, you should be able to find it if you search
"An ultra low-distortion oscillator with THD below -140 dB"
Part of it is building an equally decent notch-filter just to be able to see any residual spurs! A decent modern replacement OPAMP to make these kind of circuits with should be OPA1656 by Ti, it won't break the bank at least. Harder component to find is a really good linear optocoupler.
Matching and pairing R6, R7, C1 and C2 is rather pointless. The only result of unmatched compts is a slight difference in amplitude of the outputs of the 2 integrators, which won't affect anything.
One integrator generates a cos signal, the other a sin signal. If you display them as X and Y on a scope, you might be able to see any non circularity which represents distortion.
I would expect most of the distortion arises from the auto levelling part of the cct acting on the JFET.