If i want precison without spending x10 money my choice is Vishay MRS25 - i measure them anyway and regard them excellent to nominal value.
MBB0207 should be good.
At these ultra-low distortion levels one can always run into all sorts of cancellation effects, so maybe the Dale are actually more linear (voltage coefficient and thermals, and non-magnetic) than the Yageo but are less optimal match for the given circuit with all its other parts, including the notch and whatever else there is in the signal chain.
One way to find out is to replace the single resistors with series-paralleled quads and see if that changes anything in the distortion profile. If not, the resistors could be considered as basically blameless.
One way to find out is to replace the single resistors with series-paralleled quads and see if that changes anything in the distortion profile. If not, the resistors could be considered as basically blameless.
If you want to avoid thermal effects (=higher THD), then use MMB0207, not MMA 0204. Dale is not suitable for audio.
They both come with the exact same temperature coefficient as well as tolerance?
Fact is that the 0207 is a bit bigger, so every so slightly more thermal mass.
https://www.vishay.com/docs/28713/melfprof.pdf
According to the top graph at page 8, the current noise voltage ratio is ever so slightly better of the 0207
This is a really interesting post, and it sounds like you have a great oscillator. At what frequency was the oscillator running in your measurements? Am I correct in assuming that this is a Wein bridge oscillator? What op amp were you using? Were you using a JFET as the gain control element? What kind of amplitude detector were you using to set the amplitude (e.g., half-wave, full-wave, or more sophisticated detector like sample and hold)?
Your experience with the resistors is intriguing. I have studied thermal resistor distortion due to power dissipation quite a few years ago; maybe I missed something. In my measurements the amount of distortion depended on the frequency and temperature coefficient of resistance, with distortion increasing al lower frequencies, since the thermal time constant of the resistor mass would average out resistance changes due to signal-dependent resistance swings at higher frequencies. If your measurements are at 1 kHz, these effects would likely be quite small. There is a graph of resistor distortion as a function of frequency in my Designing Power Amplifiers book. It was for a 1/4-watt ordinary metal film resistance. As a sanity check, you may want to measure the distortion of these resistors by themselves, rather than in an active circuit like an oscillator where numerous mechanisms may be at work.
Cheers,
Bob
Thanks everyone for each of your replies. I have the Holsworthy in my cart at Mouser and need to build another board anyway so I'll report back on them.
The Vishay MBB02027 look promising, Mouser show them to be available in 0.1% in all the values except the 909Ω where 1% will do so I have them in my cart as well.
@KSTR I see cancellation effects at these low distortion levels often and have considered that to be a possibility. I suppose if all the other brands measure bad it might point a finger at the Yageos adding a cancellation. When using them they are very consistently "good."
@Bob Cordell. Thank you. To answer your questions. It is a 1 kHz Wein chimera based on Viktor's design, and a composite op amp borrowed from "Flip Flip World using THAT340s and 5532s. The notch filter is from Janasek. The gain control concept, using a degenerated element, I give credit to you. (I used a THAT2180 subtractively rather than a JFET.) The detector is half-wave borrowed from Viktor which appears similar to Jung. The attenuator was adapted from Amber Instruments. Those are the various credits and the short answer.
The longer answer is here: https://proaudiodesignforum.com/forum/php/viewtopic.php?t=1265 and here: https://proaudiodesignforum.com/forum/php/viewtopic.php?t=887
Some unrelated work I've done on a compressor using a log-domain sin² + cos² sidechain may benefit those designing state-variable oscillators. In the compressor I use a Dome filter to obtain Quadrature and transistor arrays but a pair of THAT4305 could easily make a "ripple free" linear or log-domain output detector for SV oscillator stabilization sans Dome. You'll find that thread here: https://www.proaudiodesignforum.com/forum/php/viewtopic.php?t=1361
One interesting property of the Dome filter applied to audio is that Sum and Difference information appear at the output with equal weight when added algebraically to mono or power summed by the RMS detectors. Many recordings sound better when summed as I+Q rather than L+R because the L-R information is not lost.
@Bob Cordell While I'm giving you credit I use your rectifier design here: https://www.proaudiodesignforum.com/forum/php/viewtopic.php?t=1331
I make distortion too...
The Vishay MBB02027 look promising, Mouser show them to be available in 0.1% in all the values except the 909Ω where 1% will do so I have them in my cart as well.
@KSTR I see cancellation effects at these low distortion levels often and have considered that to be a possibility. I suppose if all the other brands measure bad it might point a finger at the Yageos adding a cancellation. When using them they are very consistently "good."
@Bob Cordell. Thank you. To answer your questions. It is a 1 kHz Wein chimera based on Viktor's design, and a composite op amp borrowed from "Flip Flip World using THAT340s and 5532s. The notch filter is from Janasek. The gain control concept, using a degenerated element, I give credit to you. (I used a THAT2180 subtractively rather than a JFET.) The detector is half-wave borrowed from Viktor which appears similar to Jung. The attenuator was adapted from Amber Instruments. Those are the various credits and the short answer.
The longer answer is here: https://proaudiodesignforum.com/forum/php/viewtopic.php?t=1265 and here: https://proaudiodesignforum.com/forum/php/viewtopic.php?t=887
Some unrelated work I've done on a compressor using a log-domain sin² + cos² sidechain may benefit those designing state-variable oscillators. In the compressor I use a Dome filter to obtain Quadrature and transistor arrays but a pair of THAT4305 could easily make a "ripple free" linear or log-domain output detector for SV oscillator stabilization sans Dome. You'll find that thread here: https://www.proaudiodesignforum.com/forum/php/viewtopic.php?t=1361
One interesting property of the Dome filter applied to audio is that Sum and Difference information appear at the output with equal weight when added algebraically to mono or power summed by the RMS detectors. Many recordings sound better when summed as I+Q rather than L+R because the L-R information is not lost.
@Bob Cordell While I'm giving you credit I use your rectifier design here: https://www.proaudiodesignforum.com/forum/php/viewtopic.php?t=1331
I make distortion too...
Last edited:
All HD3 means that distortion is either power or voltage related. (I mean absolute value of a voltage)
Thanks for the info. I have also explored compressors, including variable mu tube, THAT VCA and JFET. I had a nice conversation with EveAnna Manley at the AXPONA show last week.
Cheers,
Bob
Well usually its electric field strength, not power. You can parallel resistors to increase power without affecting distortion for instance, or use 2S2P array to increase power without changing resistance value and decreasing distortion.
And some distortion can be due to the end-cap attachment too.
And at low frequencies you also get thermal distortion (which does depend on power as well as frequency)
Try a cheap 1W 0.1% resistor perhaps?
That's been on my mind as well. We're in the 1 ppm to 100 ppb range.
I should add that while the 500 mW Dale were dissipating less than 1.5 mW I lightly freeze-misted them and the HD3 level and overall signature didn't change. My gut tells me this time it isn't thermal.
I can also attest that it's not the Yageo's cancelling any VCA HD3 because there's a jumper I opened to remove the VCA from circuit. It requires manually servo'ing the AGC with the calibration trim pot long enough to complete a low-average, low-point FFT.
I suppose it could be.
At my test level of +14 dBu the resistor was seeing 5.5V peak across a 250V-rated resistor but we're admittedly looking deep into the weeds at the 100 ppb level of the Yageo's and the approximate 500 ppb distortion of the Dales.
Tested a lot of dual op amps in this circuit and the 5532 out-performs both the OPA1612 and LME49720 in particular. I prefer NJM but TI 5532s are almost identical in performance. Have not tried an OPA1656 yet.
What is the voltage across the 10K resistors in question? Are the Dales new or old? I will check a handfull on the CLT-1 https://linearaudio.net/sites/linearaudio.net/files/CB-to-ES-V1-ref-2-IMG_0001.pdf of the older stock I believe I have. At those ultra low distortion levels all manner of things will get in the way. One of Radiometer's demos of the CLT-1 is to hold a large pencil led next to the component and watch the distortion rise.
The Dales were delivered from Mouser the day before I tested them.
The CLT1 is impressive.
Last edited: