|
Homebrew Equipment A place to show, design and discuss the weird and wonderful electronic creations from the hands of individual members. |
|
Thread Tools |
17th Mar 2020, 3:46 pm | #1 |
Heptode
Join Date: Jun 2019
Location: Oxfordshire
Posts: 738
|
Low distortion oscillator
Some years ago there was a design by C Bateman in Electronics World for a very low distortion 1KHz oscillator and filter. Does anyone know if the PCB layout for this project is still available?
|
17th Mar 2020, 4:52 pm | #2 |
Dekatron
Join Date: Nov 2010
Location: Oxford, UK.
Posts: 4,984
|
Re: Low distortion oscillator
I don't know if you are looking for a real circuit board, or just the physical layout.
If the latter, all of Bateman's Capsound articles including the layout is here https://linearaudio.nl/cyril-bateman...sound-articles Note that the VCA chip he used, the SSM2018, is alas long obsolete. That was characterised by a low distortion, 0.006%. Most currently available VCA's I've found have ten times higher distortion. An alternative approach is Samuel Groner's Master's thesis https://www.nanovolt.ch/resources/lo...ors/index.html . That has even lower distortion than Bateman's, and is variable frequency. less than -140dBc between 20Hz and 20kHz (~0.00001%) Some parts are now defunct - why OnSemi recently obsoleted the BC550C/BC560C heaven only knows - but there are viable substitutes, including for the defunct MAT04 (there is a pin-equivalent and lower noise device from THAT, the 300B). It is a big project - much greater than Bateman's. It is on the round to it project list! Good luck Craig |
17th Mar 2020, 5:14 pm | #3 |
Dekatron
Join Date: Nov 2010
Location: Oxford, UK.
Posts: 4,984
|
Re: Low distortion oscillator
Worth noting that Bateman's board is single sided with wire straps. It looks as if it was done with traditional taping, and not using a CAD package.
|
17th Mar 2020, 5:18 pm | #4 |
Dekatron
Join Date: Nov 2010
Location: Oxford, UK.
Posts: 4,984
|
Re: Low distortion oscillator
Key to Groner's design was to use a translinear cell multiplier instead of the higher distortion options (FET, opto coupler, or VCA). And a stupidly low distortion discrete op-amp.
|
17th Mar 2020, 7:58 pm | #5 |
Heptode
Join Date: Jun 2019
Location: Oxfordshire
Posts: 738
|
Re: Low distortion oscillator
Thanks Craig, in investigating the design alas as you say the SMM2018 is pretty much unobtainable, except at a high price from very dubious Chinese sources!
I later found this one - http://www.janascard.cz/PDF/An%20ult...0-140%20dB.pdf which claims -140dB, not sure if the LME49710 is available or a substitute, but other parts look viable. |
17th Mar 2020, 10:54 pm | #6 |
Dekatron
Join Date: Nov 2010
Location: Oxford, UK.
Posts: 4,984
|
Re: Low distortion oscillator
You can buy the LME49710 mil version from TI for $149. Yeah.
Or the LME49720, which is two LME49710 in one package. About three quid from Farnell (RS have zero stock). |
18th Mar 2020, 10:02 am | #7 |
Heptode
Join Date: Jun 2019
Location: Oxfordshire
Posts: 738
|
Re: Low distortion oscillator
I was going to try an OPA1611, which has a slightly better noise and THD spec, but hadn't noticed the LME49720 - thanks for pointing that out!
|
18th Mar 2020, 10:41 am | #8 |
Dekatron
Join Date: Nov 2010
Location: Oxford, UK.
Posts: 4,984
|
Re: Low distortion oscillator
I'd forgotten about the OPA1611. That is one spectacular opamp. With distortion products at the -136dB level, why go discrete? Although a paper in the AES a couple of years ago by Groner and Polak used two opamps in a particularly cunning way to get distortion down to -180dB.
I hear it on the grapevine that TI are about to release a range of ultra high performance audio op-amps. They have really got the bit between their teeth. They guy who runs this at TI has a thread on here on their latest FET opamp https://www.diyaudio.com/forums/vend...io-op-amp.html . That particular device is not suitable for the low distortion oscillator (too noisy), but TI is very very active in this area. |
18th Mar 2020, 2:05 pm | #9 | |
Dekatron
Join Date: Feb 2007
Location: Lynton, N. Devon, UK.
Posts: 7,082
|
Re: Low distortion oscillator
Quote:
|
|
18th Mar 2020, 2:26 pm | #10 |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,876
|
Re: Low distortion oscillator
The benefit of a low distortion oscillator is usually that it can be tuned around to suit needs. If only a couple of frequencies are needed, then the prospect of a less good oscillator tarted up by additional filters enters the fray.
Of course, that moves the low distortion criticality from the oscillator to either opamps or inductors in the filters. The HP8903 audio source/distortion meter uses a biquad oscillator, by the way, they moved away from wien bridge for reasons given in the HPJ article. In a similar vein, I read of work by Gorski-Popiel (of synthesisers book fame) describing greater spurs from DDS RF sources than could be explained by the usual analyses. THere was an investigation into simulating a biquad oscillator with digital integrators etc, which did not seem to exhibit this effect. My career changed before I saw much more of this path. David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done Last edited by Radio Wrangler; 18th Mar 2020 at 2:28 pm. Reason: Can't spell Popiel |
19th Mar 2020, 7:21 am | #11 |
Dekatron
Join Date: Jul 2009
Location: Sleaford, Lincs. UK.
Posts: 7,662
|
Re: Low distortion oscillator
How low distortion are you after? You can get 0.001 or lower with an opamp and lightbulb( wein bridge) , have the schematic if you want it, F is set by 2 R's & C's.
Here's some stuff I've collected over the years, inc JLH's, see attached. There's also an interesting thread, but a very lengthy read over on DiyAudio. Andy.
__________________
Curiosity hasn't killed this cat...so far. |
19th Mar 2020, 11:30 am | #12 |
Dekatron
Join Date: Nov 2010
Location: Oxford, UK.
Posts: 4,984
|
Re: Low distortion oscillator
Hi Andy
The first and third attachments are two designs we were discussing initially. Apparently Bateman's oscillator is nicked from a guy called Scott Wurcer's design (on DiyAudio low distortion oscillator thread) I'd forgotten entirely about JLH's design, that seems (at least if his measurements are right) good to -140dB distortion (0.1ppm, or 0.00001%). It is very difficult to measure because the sine wave generator has to have a distortion comparable to or less than the distortion analyzer - so say -150dB That actually exceeds the best and latest edition of the Audio Precision gear. With a few transistors and a (now long obsolete) R54 thermistor. |
20th Mar 2020, 7:13 am | #13 |
Dekatron
Join Date: Jul 2009
Location: Sleaford, Lincs. UK.
Posts: 7,662
|
Re: Low distortion oscillator
The osc I built had lower distortion than the osc in a HP3903B, I used a TL072, then a 5532 which was a tad better. It used a 727 28v 40mA miniature bulb for gain correction, and even worked well with the heaters of an ECC type valve. It had a sweet spot at 3v ish, dropping 10X THD, I was quite surprised. the best thing about it was it was very simple, unlike those more complicated designs above.
Another simple 0'004% wein bridge osc here - http://www.turneraudio.com.au/thd-measurement.html it's similar to the design I built. Elektor also did a spot frequency low THD osc too, but it was more complex than JLH's. the problem with any really low THD osc is measuring it for us home builders. The OP doesn't say what he's trying to measure, but one of the simple design's built on copperclad board and cut with a Dremmel gets you good results, the whole sealed in a cast ali box, oh, and lets not forget the PSU. Mine was pretty simple, 12 0 12v tfmr, 4 decent Nichicon 2200u caps, 2 in parallel for each rail, 7815/7915 as per datasheet, IE 0.1u caps on pins, protection diodes etc, very short wiring. Andy.
__________________
Curiosity hasn't killed this cat...so far. |
20th Mar 2020, 8:48 am | #14 | |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,876
|
Re: Low distortion oscillator
Quote:
Help comes in the form of exploiting the curvature of characteristics of analogue devices. If you run something at half-power, second-order products will usually fall 6dB (not 3dB) Third order will fall 9dB and so on. The device linearity improves at falling levels. The limitation becomes the noise floor and the rejection factor of harmonic selection filters. Narrower filters will reduce the noise floor. Low distortion measurement can be done, but it requires a bit of care and development to cultivate. There are eventual limits set by the linearity of resistors, and averaging times heading out to more than the time you have. David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
|
20th Mar 2020, 8:59 am | #15 |
Dekatron
Join Date: Nov 2010
Location: Oxford, UK.
Posts: 4,984
|
Re: Low distortion oscillator
Let's not forget that Bateman was trying to measure the distortion products of capacitors. So he needed a low distortion oscillator, and then a spectrum analyser to characterise the distortion.
Groner's Master's thesis distortion analyser manages <-140dB distortion measurement (0.00001%), and that is a very complicated beast indeed. David is absolutely right though - to generate a sine wave with that level of distortion, and then measure it, is very difficult. And why would you want to, other than a time-nuts type fascination for pushing the boundaries? Craig |
20th Mar 2020, 10:25 pm | #16 |
Nonode
Join Date: Aug 2010
Location: Resolven, Wales; and Bristol, England
Posts: 2,608
|
Re: Low distortion oscillator
Didn’t the Radford LDO use a twin-t oscillator for low distortion?
__________________
Richard Index: recursive loop: see recursive loop |
21st Mar 2020, 4:04 am | #17 |
Pentode
Join Date: Mar 2016
Location: Champaign, Illinois, USA.
Posts: 227
|
Re: Low distortion oscillator
Groner: "Master Thesis in Music"
! |
21st Mar 2020, 3:58 pm | #18 |
Nonode
Join Date: Jan 2009
Location: Bristol, UK.
Posts: 2,062
|
Re: Low distortion oscillator
How good is good enough for audio?
10ppm measurements are not difficult with a good 24bit sound card (it is possible to do a lot better), provided its frequency limits are acceptable. A signal generator ~ 10X better (0.0001%) would then be the gold standard for that. Just having a number for %THD doesn't tell the whole story so you want to see the harmonics. With Fourier analysis in the digital domain you get that automatically, but if you use a band-stop filter you really need a spectrum analyser on the back-end. I've used a TEK501A with Fourier analysis of the residual - all very interesting but even a simple op-amp circuit (like an 5534 with the gain lowish) can be so good the measurements are pretty much meaningless. I have seen the residual THD go DOWN when I connect the DUT (probably anti phase distortion that cancels?) dc |
21st Mar 2020, 6:47 pm | #19 |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,876
|
Re: Low distortion oscillator
A bottomless pit is opening up beneath you, Dave!
The audiophiles believe that no impairment can ever be so slight that it can't be heard, indirectly, if not directly. They will fight to the death to defend this view. Pragmatically, under everyday circumstances, we left perception behind long before we got down to quite these levels. Careful testing with efforts taken to make sure there are no give-away clues from anything other than the actual sound, shows that opinions are scattered randomly. It is very difficult to prove that anything is random. without citing the method of production. David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
22nd Mar 2020, 10:10 am | #20 |
Dekatron
Join Date: Nov 2010
Location: Oxford, UK.
Posts: 4,984
|
Re: Low distortion oscillator
Low order distortions are very kind to the ear. 2nd harmonic is an octave. Third harmonic is an equal temperament fifth up an octave to within 2 cents (100 cents is a semitone), and is inaudible.
The ones to watch out for are high order odd harmonics, which are rather a long way from accepted notes (7th, 9th etc) and if present are much more audible. That is why you need to be careful about crossover distortion in power amps, because being rather spiky in nature is rich in high order harmonics. But with care even that can be reduced to a few parts per million. I suspect that is why in the early days of transistor power amps they did not sound as good as the valve counterparts. Valve amps tend to have low order distortion spectra, which fit in with euphonic musical intervals, whereas early transistor power amps produced high order distortions. But as far as opamps are concerned, there are so many high quality ones now that cost a quid or less with distortions that are measured in sub parts per million (typically 0.1 to 0.3ppm) that are all totally beyond audibility. Well designed discrete op amps (Groner's are a case in point) can get beyond that by factors of ten. But is all a bit academic way down there. Back in the day I built a Texan (as I suppose other here did). That used 741 and 748 opamps which were an eye popping 60p each (getting of for a tenner in today's money). We've come a long way since then. Craig Last edited by Craig Sawyers; 22nd Mar 2020 at 10:18 am. |