Columns Retired Columns & Blogs |
October 14, 2009 - 8:30am
#1
Hello to everyone, I'm brand new here...
Loudspeakers Amplification | Digital Sources Analog Sources Featured | Accessories Music |
Columns Retired Columns & Blogs |
Loudspeakers Amplification Digital Sources | Analog Sources Accessories Featured | Music Columns Retired Columns | Show Reports | Features Latest News Community | Resources Subscriptions |
Hi Jim. Welcome aboard. A lot of posters here are in the, as you call it "esoteric" category. In certain circuits it's easy to hear the difference between caps.
Please remember that it does NOT take Golden Ears to ear differences your audiophile friends claim. It takes... practice! Listening over and over again to great recordings you love, not audiophile ones.
Surely you've gotten better at playing the trumpet and singing over the past 40 years. It would utterly insulting for me to pick up and 'play' a variety of trumpets for a few hours and tell you that that all of them sound the same. There are things you listen for in a trumpet, and educate others in regards to your knowledge. Obviously, if you know anything, and hear from a newbie that "you are full of it", you would be very insulted and disgusted with teaching anyone anything ever again. A neophyte knows nothing for a reason, and must look to those with experience for blind faith, until they are good enough to 'walk on their own' (or what to listen for).
This is the same approach that applies to critical listening, no matter what the recording. After listening to amps or speakers (or whatever component) in a controlled environment (and after spending hours comparing however you like), anyone can hear the differences, especially if you have heard your gear for several years upgraded with better parts. Unquestionably. This is how technology in general evolves, and musical instruments (even Stradivarius was better later in life).... slow, small changes over the years. Good listening!
Congratulations on your music talent and experience, and welcome. Because of your extensive experience in hearing sound/music played, you should be able to tell the improvement (or difference) between certain types of capacitors in the signal path. There was a thread in the Tweaks section about capacitors - you might check it out.
I too play music instruments and also have technical degrees, so it is very interesting to me to try to match the sound with the science. I also didn't believe those that claimed that different capacitors have different sound when replaced in the circuit - until I tried it. I repeat my caution that the original part was replaced by a new capacitor, but with different leads to the capacitor, different physical location and size of the part, different construction of the cap, different solder joint, and so forth. It's not clear what really was the reason, other than the discussion focuses on the type of capacitor dielectric material.
But with the aggregate replacement of caps throughout a pre-amp or an amp, or tuner, the effect is in the same direction of sonic change, whether or not the caps were smaller or larger or paralled with smaller values, etc. I remain convinced that this effect is real, and true for my own experience in my system. There were no controlled studies done in my case to make this a scientifically publishable result. I liken the difference to music being more like live with individual instruments more stand-alone and not as blurred together as it sounded before the cap mods.
If you have the resources, try to get two of the same equipment and modify one and compare. See if you really can tell which is which without knowing beforehand. This was an experiment I would like to have done myself.
I am a very firm believer in double blind testing of anything as subjective as musical quality, but I won't dismiss, out of hand, any claims from the more experienced listeners. I tend to trust proper instrumentation to give me information about distortion, noise, and frequency response more than I trust ears, no matter how well trained, though.
Forgive me for treading on anyone's toes, as I realize there are sensitive issues in any sufficiently passionate community.
I'll certainly check out the tweaks thread about capacitors, thanks.
Jim
Jim, when I first tried the caps mod, I performed it on the audio section of my FM tuner. The replacement caps were polycarbonate and polystyrene, replacing the electrolytic caps in the signal path. I mentioned this to a friend, and wrote specific instructions on what I did (he had the same model of tuner). When he was done with the mods, he said that the tuner sounded like a different machine.
Note that I could readily tell the improvement from electrolytic to film caps but I'm not sure how much a difference I would hear going from Mylar to polystyrene, for example.
I was curious as to why the FM tuner sounded different. Since that was already modded, I proceeded to the pre-amp. This time I ran square waves through the pre-amp and looked at the wavform from large signal down to the half-millivolt range on a oscilloscope. After the mods, I did the same square wave tests, and could not tell a difference, visually. But boy could I hear the difference in sound. Unfortunately, I did not have a storage scope, or other digital equipment that we now have (it was the mid-1980's), and the experiment wasn't rigorous enough. Nevertheless, I heard a startling improvement - in the sense that I did not expect that result. I'm sure there is a scientific explanation, and the square wave test may not have been the right test anyway.
Good to hear back from you and hope to hear more from you in the future. Enjoy!
Jim, Jim, Jim, double blind testing, basic mesurements. What are you thinking? Everyone here knows that science has no place in audio. No one here would ever admit that Psychology has a major input in listener bias.
Jim, I'm also a serious musician, trumpet and guitar, and an audiophile. I think that I easily hear differences between power cords, speaker cables, physical isolation devices, etc.
Part of the problem with traditional testing methods, like comparing square waves is that they don't take into account the complexity of music, with multiple complex wave forms that must be time aligned in relation to one another.
Nordost and U.K. based Vertex are working with sonar/radar engineers to measure what we hear, developing new measures. Hopefully a white paper will be forthcoming in 2010 so that others can validate their early findings. There's some discussion here:
Discussion of "new" measuring techniques
Dave
Hello! I'm new here too. Audio has been a hobby of mine for a long time as well. It all started with a Radio Shack boombox around 1983. That made one happy 3rd grader! Well- you know what they say, "The difference between men and boys is the price of their toys." The boombox is still around as a memento of simpler times.
Does that boombox have any tubes in it?
Lin
No- No tubes.
However, I saw a nifty Commodore tube type table-top stereo on Ebay once that looked like a boombox. It would be interesting to mod a boombox with nuvistors.
caps differences easily heard with this
http://www.libinst.com/Audio%20DiffMaker.htm#DiffMaker%20Download
and the .dyf file here
http://www.libinst.com/Poly%20vs%20NoCap.dyf
http://www.libinst.com/Z5U%20vs%20NoCap.dyf
I know from personal experience that coupling capacitors in amplifiers and preamplifiers make a major difference in sound quality.
I used an Audire Forte solid-state 120W amplifier for many years and it sounded (in retrospect) rather mediocre with the original input coupling caps which were a high-quality electrolytic type.
About 10 years ago I installed MIT Multicaps, a polypropylene film type, and the result was a much improved sound. Over the years, I experimented with various film caps from ASC and other companies, and each had its own sound quality. There is no question that they make a big difference in sound quality. Mylar caps are very inferior; all of the best-sounding types in my listening experiments were either metallized polypropylene or polypropylene with varous film dieletrics used.
Also, I used an Audio Research LS-2B preamp for many years. Mine was a fairly early model, and in discussions with a technician at my audio store I learned that he had been employed for several years as a tech at the Audio Research factory. He said that the engineers there found that one of the biggest factors in the sound quality of their preamps was the capacitors used for output coupling.
He said that Audio Research spends up to $100 just for the four output coupling caps because they are so important in the sound quality of the preamp; they are by far the most expensive single components except for the power transformer.
He also pointed out that the latest version they were using was considered to be much better-sounding than the ones that had come as original equipment in my preamp (1992 production). (he later returned to Audio Research, by the way, and later founded his own company that does retrofit modifications to improve the sound quality of various hi-end products)
The originals were MIT Multicaps. I replaced them with Dynamicaps, which is what Audio Research was using in the output stage of the latest LS16 at that time (2004), and the improvement in sound quality was stunning! For about $100, the LS-2B was transformed into a much much better-sounding unit than a stock one.
A person looking to upgrade their system cheaply could do worse than to go this route; a used LS-2B for maybe $600 and then $100 for the Dynamicaps will give you a very good preamp, and the Audio Research preamps pretty much last forever. They are built like a tank, and I have never had one fail for any reason in 25 years of using them.
The Mark II version of the LS-2B, by the way, actually did come from the factory with the Dynamicaps installed. If you look at the pictures of the Mark I and Mark II on the ARCDB website, you will clearly see the large yellow output coupling caps in the original and the red or white Dynamicaps in the Mark II picture (Dynamicaps are normally red, but Audio Research gets them specially made in a white generic outer wrap now for their production...apparently to somewhat disguise what they are...lol).
I now use the Audio Research LS-26 preamp, and it and the Reference 3 preamp also use the Dynamicaps in the output stage coupling.
I also later changed the Audire power amplifier over to Dynamicaps for the input coupling, and a major improvement was also heard there. I believe that the Dynamicaps are a metallized polypropylene type. As far as I know, there is no cap that equals the sound quality of the Dynamicaps; my ears tell me that they are truly superior to any others I have have tried.
So...that is my experience, for what it is worth.
Good listening!
P.S.-If you want to do your own experiments, my parts source for the best types available is partsConnexion in Canada; check out their website. I don't know what sort of amplifier or preamp you are using, but if it has coupling capacitors in the signal path and you replace them with Dynamicaps, you will have no trouble hearing the improvement in sound quality.
I do not disagree that square waves are not music waveforms. However, to get a square wave right, it has to have all its component frequencies correctly time-aligned and each and every frequency at the right amplitude too. That is why a good square wave performance can tell you information about 1/10 of the square wave frequency to beyond 10 times the square wave frequency. A 1 KHz square wave can provide useful information from roughly 100 Hz to 10 KHz.
However, square wave is a steady state signal whereas music is not steady state.
Edited for typos. 3:15 p.m., 11/7/09
music is made up of harmonics...harmonics are only sine waves. ALL audio is made up of sine waves..if the amplifiers/speakers/ancillary equipment passes all sine waves equally, i.e. has a flat frequency response, it will pass all the harmonics. audio isnt some mystical black art... it is quantifiable.
Exactly what I was getting at. However, there are other effects that are non-linear, and therefore different. In my case, I was trying to find a possible hint to what makes the caps sound different. That square wave test I tried did not provide enough information probably because of the way it was done. I had hoped that such an audible effect would show up more markedly there, but maybe not the right test.
What kind of other effects were you seeking?
No single stationary test signal will make all distortions and noise sources evident. It is quite possible, however, to use a variety of signals to create signals where different kinds of linear and nonlinear effects are exposed.
What were you trying to expose?
Although to be fair music is complex, meaning that while it is comprised of sinewaves the amplitude and funadmental-harmonics (up to and maybe over 30 fundamental-harmonics up to 20khz) can/are always changing in the ms timescale.
A steady state sinewave has nothing in comparison, I could be wrong but my understanding is that a squarewave is one of the most challenging ways to test an amplifier.
However this is not the same as say playing a trumpet with multiple notes, or complex chords on an instrument/multiple instruments.
The following link is more of interest I feel in showing fundamental-harmonics of a few instruments waveform in detail over FR and time (see section III and links for more instruments)
http://www.its.caltech.edu/~boyk/spectra/spectra.htm
I am just curious whether highly complex musical passages due to musical chords and/or multiple instruments create a limitation with the audio hardware, that could be fourier analysis or complex wave reproduction from an amp when considering the fast transients in time and FR-amplitude change (basic example is the attack-decay of an instrument, or multiple instruments-chords).
Anyone seen any research papers that say look at the use of simple sinewaves for testing audio equipment compared to complex notes-chords generated by instruments?
Thanks
Orb
Pretty good easy to read document by Bill Buxton explaining sinewaves and complex waves, psychoacoustics, etc.
Works at or used to for Microsoft Research (met him while our companies were discussing certain technologies).
http://www.billbuxton.com/AudioUI02acoustics.pdf
Cheers
Orb
That's nice. Do you have any questions about entry level audiophile equipment?
Of course, you're leaving out the detail that a sum of sine waves can represent your "changing by the millisecond" signal.
Not sure what you're trying to argue here.
Possibly because I am missing something here, so hoping you can expand a bit.
I can understand that a sum of sine waves can represent changing by the FR (ie complex wave and partials-harmonics), but wouldnt the sum change every ms as the transients do?
Meaning that fourier analysis (complex wave summing fundamental and partials) has to be recalculated every time a sound changed, because a note is never consistent due to fluctuations caused by amplitude changes over time (attack-decay-sustain-release) and also subtle changes in the sound (changes to the fundamental-harmonics).
After typing the above looked for the following to back up and explain more succinctly what I am saying;
Bill Buxton (did you ever get to meet him at Microsoft Research) sums it up nicely by saying:
That is on page 2.10
Another is:
http://www.sfu.ca/sonic-studio/handbook/Fourier_Analysis.html
So it seems if you want to look at fundamental-harmonics accurately in both time domain and frequency then you cannot use the sum effect and each harmonic must be shown individually as both Bill and the link suggest.
So this is why I posted a bit earlier regarding that music is a lot more complex than just a group of sinewaves that can be summed, which also for some seem to mean that a test tone-sinewave can be simple for testing equipment.
This ignores the potential (and potential should be emphasised) that the most complex of sounds in music and those generated by instruments can be limited by audio hardware and software, meaning a more strenuous test tone-complex wave should be used for measuring and testing audio equipment to be closer to reality of listening to music.
This is what I guess I am pondering and looks like an earlier poster was as well (WTL).
Bear with me as there may be mistakes in this - hard to type while being chatted to as well
Cheers
Orb
Ah, I see. You're not considering that one can take a Fourier analysis of any length.
You can take successive (usually overlapped) DFT's (or FFT's, of course) of a section of music,
OR
You can take the FFT of the whole thing.
Both work.
Both will contain exactly the same information, given some normalization, etc, and picking of nits.
The two will not be the same.
The point with Fourier analysis is that any signal that has finite energy and finite length can be represented by Fourier analysis (either discrete-time or continuous time, as appropriate). Finite energy alone is sufficient for continuous time, ditto finite length. For signals with discontinuities in the signal (as opposed to a derivitive), there will be a zero-energy amplitude error at the discontinuities. Of course, since no signal can have infinite frequency, there are no "real" discontinuities except in particle physics and astrophysics, and perhaps analytical P-chem.
jj, unfortunately, I had a very limited source of signals from test instrumentation. Since I was trying to complete the mods and listen, I tried the simplest multifrequency test I had on hand, a square-wave generator.
What I was trying to see was whether there would be any difference in square wave appearance on the scope before vs. after, not knowing apriori what it would sound like. I had a hint from modding the FM tuner, but the effect on instruments was 'different' in the sense of more noticeable on my preamp. Only after the mod did I try correlate the effect that is audible with the pre-mod signal, a square wave, by looking at a square wave afterwards.
The difference I heard is that the instruments are more separated by themselves than before the mods. Their locations in space seemed more stable horizontally, but I can't say that it was 'provably' better, just a listening impression. As I think how this might occur, it occurred to me that the better separation of instruments may have something to do with the time arrival of signals from the two channels being 'better' than before, but don't have any science to back this up or subsequent research to say one way or another. Looking back, the square wave test probably did not tell me what I was seeking to find.
I agree and understand your suggestions about other signals and combination of same will reveal more.
I was pondering the possible effect of a steady-state signal vs. a continuously changing signal. Pure sine waves are a single frequency. Square waves are odd harmonics. Triangle waves are even harmonics. (if I remember my Fourier stuff correctly, from decades ago). Pulses have a whole lot more. I understand that complex, non-steady signals can also be represented through Fourier methods.
What prompted my post is explained by the following. I re-mention the experiment that I read somewhere regarding the late Richard Heyser who brought a small device with an input and output for testing. When place into the audio path with standard test signals as input, the ouput was very good (flat frequency response, no distortion, etc.). When music was the signal, the output sound was distorted - badly. Obviously amps and pre-amps, etc., are not doing as poorly as that. Is it possible that electronics are doing a very small amount to music what the Heyser's device did, or something similar? The result is that the stuff measures well but subtly sounds different.
Now THAT is a problem I understand.
IF you have a known-reliable DAC, perhaps one you can run at 96kHz, you can generate test signals digitally and play them back over the DAC, and then capture the results with a similarly good ADC, and analyze them on your 'puter.
Something like "octave" or "SciPy" even has many builtins to make this kind of thing pretty easy.
It does require some mathematical understanding.
Indeed, any signal with finite time duration and finite power can be DFT'ed (i.e. FFT'ed). (Talking a sampled, quantized analog here, not a continuous-time, continuous-level analog.)
Thing is, a single sine wave may not excite the right nonlinearities. Ditto several. But I think we're agreeing on this.
If you do have the opportunity to try a digitally generated signal, do try the 250Hz + n*500Hz kind of nasty buzz tone.
These kinds of things will show up nonlinear effects, generally, by having a previously nonexistant 500Hz tone pop out of nowhere. (You might not be able to hear it, but you can measure it.)
Well I'm ready to throw some money and time into a new hobby. I am going to build a hybrid sys. Vintage amp preamp and a new squeezebox. I have been collecting flac music files for the past 3yrs and want to hear them at their proper quality and volume! I have already purchased a Mc2105 and am looking for a preamp,preferably a Mac. I am all over the place when it comes to the speaks. I have a decent pair of JBL 100 in need of some tlc, but I think I would rather build from the ground up.Any help or info would be greatly appreciated. Thanx.
Thanks JJ for giving a detailed response, much appreciated.
I should had been more articulate in my post and split it into seperate ones by me I feel, or at least not to try and do a post while trying to be attentive to someone chatting to me at same time here
I agree you picked up on what I was missing; this was down to my thinking of two different aspects to be honest and these should had been articulated seperately, specifically fourier analysis in real time (not a given sampled length with x periods), and also the incredible complexity of musical sound and its transients that vary very fast and affecting timbre-tone-etc.
My thoughts were around what we hear and presenting it accurately.
So with that in mind, wouldn't the time resolution have to be small, as in a window function of 1 to 2ms of the whole time sampled segment for accurate transients?
Just using that figure as an arbitary pick out of the air type, but would it not have to be pretty similar to what we ourselves can temporal resolve in the ear/brain?
Bearing in mind I think its been shown musician drummers can synch-phase down to 15ms and that requires listening and drumming to what they hear - think the average worked out around 30ms without using special technique.
This was what I meant by accurately representing the change of fundamental/partials and being fourier recalculated say every X ms.
If we look at the 1st diagram of a measured trumpet tone over 0.16sec, we can see that for the first seven harmonics it is never stationary, so the usual representation of the frequency domain for amplitude with fundamental-partial hz value would have to be re-iterated many times to be accurate.
http://www.sfu.ca/sonic-studio/handbook/Fourier_Analysis.html
Adding to this to make it even more complex would be not just major amplitude changes but also changes to tone and timbre caused by the change fundamental-partials.
I dont think I am saying anything that disagrees with your post, just you articulate the whole aspect of Fourier analysis where I was making my statement and did not consider the sampling of a segment of recorded music that as you shown can be represented by df * dt >= 1.
That is why I like how Bill presents time-frequency-envelope that is similar I assume to sonograph but presenting the fundamental-partials in an easier way to read.
And as you say, we cannot have it both ways in terms of resolution so there are compromises with FFT.
The following link gives a nice example (more for others than you JJ as you probably got your own practical examples); halfway down Frequency and Time Resolution Tradeoff: http://eamusic.dartmouth.edu/~book/MATCpages/chap.3/3.5.probs_FFT_IFFT.html
Thanks for the heads up
Orb
Yeah I really should had done two seperate posts as this is what I understood you to mean, in the end I jumbled it all together
My problem with the current tests involving signals is that they do not reflect real music.
They may show specific parameters in performance terms, but they will not show why two amps may sound different, especially with certain music.
This is why I also mentioned the three dimensional spectral plot shown by Bill Buxton and also in the other link.
I was thinking that it would be good if a hardware-software solution enabled for a segment of music to be accurately presented, measured, and compared.
So it would be both a simulator presenting the signal as does a CD player to the preamp, and also be able to passively measure in-line between amp and speaker and then compare what it presented to what was outputted.
In this situation we have an accurate look for amplitude,fundamental-partials, and time.
As per Figure 9 page 2.11 shown by Bob Buxton: A time varying spectral plot of a complex sound.
http://www.billbuxton.com/AudioUI02acoustics.pdf
Ofcourse this could never be a universal standard.
Being chatted to again so bear with any mistakes
Cheers
Orb
Try "help waterfall" in your copy of Matlab, presuming you have it. Not sure if it's supported in Octave yet.