Ultra low distortion signal sources

[Tubetec]

Did a bit of looking around into this subject , one guy on ebay does a special board, its set frequency ,but available to order at what ever frequency you like , four or five zero's in front of your % thd number , even the AP guys approved his boards, 

I looked at a few other approaches to low distortion ,and it all seems to come back to a special output stage configuration twin t filter and op amp , tuned carefully to a certain frequency . I found one design online  where the designer recomended a specific capacitor  dielectric type for absolute minimum distortion , the cheap and cheerfull ceramic disc caps showed some limitations but were down at  distortions of 140- 155db  below fundamental  (0.0000% or more).

Around two zero's is what my computer based REW set up can get me  with external A/D and D/A , with an external signal from an arb its about the same at a favourable frequency setting . Are we needlessly trying to use a zoom lens through the eye of a needle , distortion does become more objectionable by ear  as we ascend  the harmonic spectrum , still though is there any need  in taking in down to subatomic level just because we can ?

 

[JohnRoberts]

A pursuit for people with too much money or too much time on their hands.

When you start designing silly good gear, how do you measure it..?  :

I designed one piece of low end test equipment and that was hard enough.  8)

JR

 

[Tubetec]

Did you give REW a go yet John ?
Serious bit of kit
FFT/Spectrum I had 20 years ago in software ,
2nd-9th harmonic distortion readout in real time is nice to have .

 

[benb]

That oscillator on ebay, is this the "Victor" oscillator? There's been some talk about it on diyaudio in recent years, including the seller/designer stopping by and posting the schematic. It's fixed-frequency (I think available in different frequencies), lots of 0's in the THD figure, and remarkably low cost (ISTR under $40) for such a device.

Having lurked in both worlds, it does seem funny, recording people are all about which mic, preamp, compressor, EQ and whatnot give the "sound" they want regardless of distortion figures, but on the playback end hifi nuts generally (claim to) want as low distortion as possible, and too often believe their source material is absolutely as good as the medium (LP, CD, whatever) can give (refer back to all the recording equipment to get a "sound"). Often enough, the claim of wanting pure, very-low-distortion sound is canceled out by use of tube playback equipment. I've read where very-low distortion playback equipment has been called "too pristine."

What is thought of as a "straight wire with gain" is often used (sometimes subconsciously)  as a device for sound modification, and it's usually not as blatant as a cranked guitar amp.

 

[JohnRoberts]

Did you give REW a go yet John ?
Serious bit of kit
FFT/Spectrum I had 20 years ago in software ,
2nd-9th harmonic distortion readout in real time is nice to have .

no

40 years ago I had a spectrum analyzer I purchased used for around $700, back when that was real money (for me). It only measured up to 30 kHz with about 50 dB dynamic range... but was good for what I needed back then.

JR

[edit- back decades ago I extended the resolution range of my test bench by feeding the product output from my heathkit distortion analyzer into the input of that spectrum analyzer (old used Singer). This extra several tens of dB resolution floor allowed me to see distortion caused by the heathkit audio path itself. I worked around that path distortion by running the heathkit 100% 10 dB cooler.

Speaking of software several years back I coded up a crude FFT into a DSP microchip (DSPic) for a drum tuner prototype. I managed to parse levels from a FFT to shotgun multiple tones into a drumhead to find resonances. It was not easy to code the FFT in my favorite assembler langauge. I had to reverse engineer the FFT code from C routines.  It would be nice to revisit that super-tuner but drummers don't seem to want a better tuner, that isn't smaller/cheaper.  [/edit]

 

[Tubetec]

I found something interesting about caps in ultra low distortion oscillators of this type, the designer of one very low distortion signal reference noted a small difference depending on the kinds of caps used , foil caps showed a definate advantage over ceramic disc types in terms of achievable distortion , yeah ok its completely beyond what the ear can percieve as distortion , but its an interesting way of actually being able to see the distortion inherent in components themselves.

 

[JohnRoberts]

Our own Sam Groner did a bunch of work on low distortion oscillators and IIRC did a series of measurements of low level distortion from electrolytic blocking capacitors.

http://www.nanovolt.ch/resources/low_distortion_oscillators/index.html

I don't have a link to that specific capacitor study but a good website to review in general.

JR

 

[clintrubber]

I found something interesting about caps in ultra low distortion oscillators of this type, the designer of one very low distortion signal reference noted a small difference depending on the kinds of caps used , foil caps showed a definate advantage over ceramic disc types in terms of achievable distortion , yeah ok its completely beyond what the ear can percieve as distortion , but its an interesting way of actually being able to see the distortion inherent in components themselves.

FWIW, there was this series in EW & WW magazine a long time ago. I scanned it, but upload link seems to be lost/gone.

Ah, another source of that info here:

http://www.collinsaudio.com/Prosound_Workshop/Capacitor-Sound.pdf

 

[benb]

I found something interesting about caps in ultra low distortion oscillators of this type, the designer of one very low distortion signal reference noted a small difference depending on the kinds of caps used , foil caps showed a definate advantage over ceramic disc types in terms of achievable distortion , yeah ok its completely beyond what the ear can percieve as distortion , but its an interesting way of actually being able to see the distortion inherent in components themselves.

Ceramic capacitors are often REALLY bad as far as distortion. They're highly microphonic as well. NP0 (low capacitance drift with temperature) ceramic is usually not as bad, but I'd still use about any kind of foil cap for an audio or other 'critical' signal, even the tone rolloff in a guitar.

There are modern SMT ceramic caps that are 10uF, 10V in about a 1MM cube. These are great for bypass/decoupling in smartphones and such. but 'raw specs' can be deceiving. If you look through the datasheet, the 10uF rating is at 0V, and by the time it's charged to 10V, it's only 5uF.

 

[user 37518]

Ceramic capacitors are often REALLY bad as far as distortion. They're highly microphonic as well. NP0 (low capacitance drift with temperature) ceramic is usually not as bad, but I'd still use about any kind of foil cap for an audio or other 'critical' signal, even the tone rolloff in a guitar.

NP0/C0G capacitors are not bad at all I've used them for filters with great success, granted I prefer polypropilene caps thou. Ceramic caps are good for high frequencies, I've made very high frequency power amplifiers, and for those frequencies ( >2GHz) SMD ceramic caps are the way to go. Murata makes some great ceramics for high frequency applications, havent used them on audio thou.

 

[JohnRoberts]

Ceramic capacitors are often REALLY bad as far as distortion.

This is well known to those skilled in the art...

The dominant specification for simple distortion is "voltage coefficient" but there are other more subtle ones.

They're highly microphonic as well. NP0 (low capacitance drift with temperature) ceramic is usually not as bad,

NPO/COG are more than "not as bad", but quite good.

but I'd still use about any kind of foil cap for an audio or other 'critical' signal, even the tone rolloff in a guitar.

There are modern SMT ceramic caps that are 10uF, 10V in about a 1MM cube. These are great for bypass/decoupling in smartphones and such. but 'raw specs' can be deceiving. If you look through the datasheet, the 10uF rating is at 0V, and by the time it's charged to 10V, it's only 5uF.

Of course read and try to understand the data sheets.

JR

 

[john12ax7]

I would say modern NPO / C0G are as good or better than PP film.  So for those that prefer film still,  what aspect do you like better? Are you getting less distortion?

 

[user 37518]

I would say modern NPO / C0G are as good or better than PP film.  So for those that prefer film still,  what aspect do you like better? Are you getting less distortion?

What I like about film capacitors, sepecifically polypropilene capacitors (poliester do generate higher distortion) is that in general you can get them in larger capacitances than ceramics, this means that in applications such as filters, you can use a larger capacitor and a smaller resistor thus reducing noise. But I agree with you, NP0/C0G are great capacitors.

 

[john12ax7]

The film do win in terms of capacitance.  Kind of a tangent,  but  if anyone knows of any,  would like to find a good axial non-metallized  PP film  cap for turret board builds.

 

[ccaudle]

in general you can get them in larger capacitances than ceramics

I recently watched a YouTube video of a presentation by Bruce Hofer, retired CTO of Audio Precision, and he had a picture on a slide of a PCB with a bunch of C0G caps that was used to replace a large film cap that they could no longer get (original cap was EOL from the manufacturer). They had a jig to measure the distortion of the circuit, and the multiple parallel C0G solution was noticeably lower distortion than the original film cap.
This is the presentation on designing low THD circuits: Bruce Hofer presentation
This is the point near the end where he showed the picture of the C0G conglomeration that replaced the film cap: Film cap replacement
C0G parallel assembly was more than 20dB better distortion performance than the polypropylene and less unit-to-unit variation.

 

[user 37518]

I recently watched a YouTube video of a presentation by Bruce Hofer, retired CTO of Audio Precision, and he had a picture on a slide of a PCB with a bunch of C0G caps that was used to replace a large film cap that they could no longer get (original cap was EOL from the manufacturer). They had a jig to measure the distortion of the circuit, and the multiple parallel C0G solution was noticeably lower distortion than the original film cap.
This is the presentation on designing low THD circuits: Bruce Hofer presentation
This is the point near the end where he showed the picture of the C0G conglomeration that replaced the film cap: Film cap replacement
C0G parallel assembly was more than 20dB better distortion performance than the polypropylene and less unit-to-unit variation.

Yes, the problem with C0G is the high price and low capacitance

Thanks for the video link! looks interesting

 

[JohnRoberts]

Speaking of low distortion, Wayne just laid out a new PCB for his low distortion effort. Here is a link.
Wayne's low distortion osc

JR