In other words, "How to find out what matters and what doesn't".
I've done things like this before, especially when dealing with high-end home audio repair, modification, and design, and that is build a circuit or alter an existing piece of gear to allow swapping out of various parts under question. Why? If you think the discussion on what kind of JFET to use in a x0x can get crazy, get any two "serious audiophile" people started talking about capacitors. (And I mean REALLY crazy, could you imagine spending nearly $2,000 on a 3.3µf cap? They exist.)
In testing various things, I've come to find that some things people will argue for hours on end doesn't make a difference at all, while some things that are largely ignored make a huge difference in what you hear. With this in mind, I've started building a x0x main board (From Willzyx, after 3 tries of home-etching I got impatient) with a machined-pin socket in every semiconductor position. Don't worry about expense, every socket is either salvaged from dead equipment (Most of the ones in the pic actually), or a bulk purchase from a surplus dealer.
When done, I'll be able to switch out any semiconductor, or group of semiconductors I want, just to see if/what any difference is made. Not only do I work on high-end audio, I also repair/restore/mod analog synths, so I have a semi-decent stash of normally hard-to-find parts to choose from. Building a board this way, I'll even be able to see how much difference, if any, there would be in using a BA662 or BA6110 in the VCA. I also intend to play around with various JFETs, and different NPN transistors.
Once I get the build going, I'll try to at least get audio samples posted if not videos. Also, if anyone wants me to try a specific swap, post here and we'll try it out. I'm even open to trying stuff that SHOULDN'T make a difference. Here's why: I recently went through and did a full restoration of a Yamaha SK-15. On the "brain" board, there is a CMOS 4011 chip that controls the sound out relay. The relay wasn't engaging, so I decided to try swapping that chip before looking elsewhere. Three versions of the SAME chip, only the manufacturers were different, were tried. None worked at first, so the repairs were made elsewhere. Only the original chip would start the synth up, even though the others were supposed to be "identical". So, if a simple logic chip makes that much of a difference, nothing is off the table as far as testing semiconductors goes.
One more thing: Knowing a bunch of my audiophile friends and customers, I don't want this thread to become a venue for unsupported opinion and/or arguement, or a "BANNED match" for people trying to prove they know more than anyone else, me included. I think one thing about me may be somewhat of an advantage, or maybe not, but I have NEVER HEARD a "real" Roland TB-303. I do have an old Roland SH-09 I'm very fond of, but that's a very different beast. In other words, I don't know what the hell I'm talking about when it comes to a TB-303 sound , but I can share what I hear changing in this experiment!
If anyone wants to run a parallel experiment and needs transistor sockets, let me know, I have 0.75 gazillion of those things.
Don Taylor, a.k.a. BrassTeacher
An Attempt to Quantify Semiconductor Differences
Moderators: altitude, adafruit_support_bill, adafruit, phono, hamburgers
Please be positive and constructive with your questions and comments.
-
- Posts: 248
- Joined: Fri Oct 08, 2010 4:49 am
- antto
- Posts: 1636
- Joined: Thu Apr 15, 2010 3:21 pm
Re: An Attempt to Quantify Semiconductor Differences
just one thing:
don't use your ears.. record with a decent audio device, same conditions be careful, write down condition details when doing such tests
i'd say: use a high sampling rate (88.2kHz is minimum, altho i'd use 192k) and at least 16bits
don't use your ears.. record with a decent audio device, same conditions be careful, write down condition details when doing such tests
i'd say: use a high sampling rate (88.2kHz is minimum, altho i'd use 192k) and at least 16bits
- westfw
- Posts: 2010
- Joined: Fri Apr 27, 2007 1:01 pm
Re: An Attempt to Quantify Semiconductor Differences
I don't see how you can expect to get any meaningful results with SOCKETS in there, and especially USED sockets!!!! Everyone know that sockets can completely ruin the characteristic sounds produced by a circuit, even if you can't quite see how on a scope!!!
-
- Posts: 248
- Joined: Fri Oct 08, 2010 4:49 am
Re: An Attempt to Quantify Semiconductor Differences
Oh, I plan on doing just that. It'll give me a good reason to use my (formerly) expensive Sony Mini-disc recorder, and it's powered mic. Expect 16-bit minimum. People that tell you that mp3 doesn't sound any different from an un-compressed .aiff or .wav are listening to their music through very sub-standard gear!antto wrote:just one thing:
don't use your ears.. record with a decent audio device, same conditions be careful, write down condition details when doing such tests
i'd say: use a high sampling rate (88.2kHz is minimum, altho i'd use 192k) and at least 16bits
That said, you'd probably be surprised just how much of a living I have made off using my ears!
-
- Posts: 248
- Joined: Fri Oct 08, 2010 4:49 am
Re: An Attempt to Quantify Semiconductor Differences
Gotta love well-placed sarcasm! Seriously though, I've yet to be able to afford a scope that was capable of displaying differences that are inaudible, if one even exists. In other words, sometimes you can hear differences that a scope can't display. Perhaps a new type of instrument needs to be developed that is capable of doing so. What, I have no idea.westfw wrote:I don't see how you can expect to get any meaningful results with SOCKETS in there, and especially USED sockets!!!! Everyone know that sockets can completely ruin the characteristic sounds produced by a circuit, even if you can't quite see how on a scope!!!
You'd be amazed at the number of audiophile types I know that would agree with your statement! There would be an equal amount that would give you all kinds of crazy reasons that "prove" you are dead wrong! One of my favorites from the outer-fringe of the high-end audio world is that even a well-made wire will conduct an AC signal better in one direction than the other, and that you have to make sure the wire is running "the correct direction" to get a decent sound. I wish I could say I was making this up, but I've heard it from more than one person even...
Don Taylor, a.k.a. Brassteacher
- richms
- Posts: 558
- Joined: Tue Jan 20, 2009 3:05 am
Re: An Attempt to Quantify Semiconductor Differences
The directional cables came from the screen only being connected at one end, as is the norm. I guess the marketing department took it and ran with it using their pseudo-science dictionary to make the copy for the ad's.
My dollar store RCA cables have an arrow on them despite only having a single conductor so the screen is connected to the ground of the RCA at both ends...
My dollar store RCA cables have an arrow on them despite only having a single conductor so the screen is connected to the ground of the RCA at both ends...
-
- Posts: 221
- Joined: Mon Nov 23, 2009 9:34 pm
Re: An Attempt to Quantify Semiconductor Differences
SH-09's are coool man
Look forward to seeing some results.
I use the LM358 for the op-amp and they perform better. I don't like the idea of buying re-stamped crappier op-amps for high prices and the an6562 being re-stamped as more expensive amps!! You never know..
You have to watch those rca cables , I plugged my BANNED in wrong once and the tech police came.
RCA cables ?!? what are you talking about man?!?
Look forward to seeing some results.
I use the LM358 for the op-amp and they perform better. I don't like the idea of buying re-stamped crappier op-amps for high prices and the an6562 being re-stamped as more expensive amps!! You never know..
You have to watch those rca cables , I plugged my BANNED in wrong once and the tech police came.
RCA cables ?!? what are you talking about man?!?
- antto
- Posts: 1636
- Joined: Thu Apr 15, 2010 3:21 pm
Re: An Attempt to Quantify Semiconductor Differences
whut?Brassteacher wrote:...
In other words, sometimes you can hear differences that a scope can't display. Perhaps a new type of instrument needs to be developed that is capable of doing so. What, I have no idea.
...
your ears are not measurement devices
you can't "compare" two similar sounds with your ears..
even if you play the same sound twice - you'll hear a different thing every time (especially when it's complex sound material such as music or voice)
- altitude
- Posts: 995
- Joined: Wed May 11, 2005 5:17 pm
Re: An Attempt to Quantify Semiconductor Differences
I support this thread
I have done some work alread in this direction using modern parts (i.e. J201 vs 2SK30). A good resource to look at is the BANNED TM-3030 manual (http://www.oakleysound.com/tm3030.pdf) where he explains what differences there are and why he used what..
I have done some work alread in this direction using modern parts (i.e. J201 vs 2SK30). A good resource to look at is the BANNED TM-3030 manual (http://www.oakleysound.com/tm3030.pdf) where he explains what differences there are and why he used what..
-
- Posts: 221
- Joined: Mon Nov 23, 2009 9:34 pm
Re: An Attempt to Quantify Semiconductor Differences
I agree antto , my ears work differently on how I feel , my ear to brain circuit works quite well , much better than my friends etc.. but sometimes they don't feel like working. So I can't rely on them - scientifically speaking.
It all comes down to the brain , on how it processes the sound.
True the end result is for ears/brain and not a scope / electrical testing tools, so using your ears is probably the best way to hear if it sounds good. But a scope would be wise for checking the current / signal as you change parts.
You might think it sound fat and funky but to somebody else not, plus what might sound like a saw wave might not be on the scope.. and those sounds just might sound fun at the time, later not so fun....
I think scientific tools are just as important as well as the aural neuro tools.
It all comes down to the brain , on how it processes the sound.
True the end result is for ears/brain and not a scope / electrical testing tools, so using your ears is probably the best way to hear if it sounds good. But a scope would be wise for checking the current / signal as you change parts.
You might think it sound fat and funky but to somebody else not, plus what might sound like a saw wave might not be on the scope.. and those sounds just might sound fun at the time, later not so fun....
I think scientific tools are just as important as well as the aural neuro tools.
- richms
- Posts: 558
- Joined: Tue Jan 20, 2009 3:05 am
Re: An Attempt to Quantify Semiconductor Differences
Plus with it socketed it gives more ability to put "broken" and damaged components in to see what they do for sound, and totally wrong ones. Friend swapped some random mosfets into his guitar pedal and had some really angry sounding thin distorted sound which was "unique" and they were stuffed ones from a car amp of all things.
- westfw
- Posts: 2010
- Joined: Fri Apr 27, 2007 1:01 pm
Re: An Attempt to Quantify Semiconductor Differences
IIRC, there was a ETHERNET CABLE marketed as directional, for "audiophiles." (yeah. http://usa.denon.com/us/Product/Pages/P ... 63642a6206 )One of my favorites from the outer-fringe of the high-end audio world is that even a well-made wire will conduct an AC signal better in one direction than the other, and that you have to make sure the wire is running "the correct direction" to get a decent sound.
(Edit: OK, "Denon-link" does not appear to be ethernet. Some other "digital" tech, though.)
Your eyes are not measurement devices either. I don't have any problem believing that you can hear differences between audio signals that seem to look the same on a scope. I'm not sure I believe that the scope is INCAPABLE of displaying whatever the difference is, but not knowing what to display is effectively the same thing. A scope is not an ideal instrument for comparing two complex audio signals (music, rather than a pure waveform, for example.)your ears are not measurement devicessometimes you can hear differences that a scope can't display. Perhaps a new type of instrument needs to be developed that is capable of doing so.
you can't "compare" two similar sounds with your ears..
-
- Posts: 248
- Joined: Fri Oct 08, 2010 4:49 am
Re: An Attempt to Quantify Semiconductor Differences
Oh, I know exactly what you're saying. I use AudioQuest cables in my home rig (NOT the insanely expensive kind though...) And they are dual conductor, and made with a mesh shield connected only to one RCA ground. THAT is legit, not because of the "direction" of the wires, but by giving you the ability to basically "star ground" the shields, or by connecting all the shielded ends to one component that happens to have a three-conductor power cord that actually grounds the chassis.richms wrote:The directional cables came from the screen only being connected at one end, as is the norm. I guess the marketing department took it and ran with it using their pseudo-science dictionary to make the copy for the ad's.
My dollar store RCA cables have an arrow on them despite only having a single conductor so the screen is connected to the ground of the RCA at both ends...
No, what I'm talking about are some guys that if you give them a brand-new piece of metallurgically pure copper wire, they will play it one way, then flip it to find which way that they say it "sounds better". That, or (And I SWEAR I am not making this up) like one acquaintance I have that will look you dead in the eye, and seriously declare: "When you get a brand-new cable, the first time you use it, make sure you mark which direction the signal is going on the cable so you don't plug it in backwards the next time, because it won't sound as good once the wire 'learns' which direction the signal is supposed to go."
Yes, these guys make these declarations not realizing that an audio (or most all) signals are ALTERNATING current. I would imagine the myth comes from the fact that a piece of copper that either naturally, or is chemically treated (usually with a compound containing selenium) to develop a patina on one side only actually does become a rectifier.
-
- Posts: 248
- Joined: Fri Oct 08, 2010 4:49 am
Re: An Attempt to Quantify Semiconductor Differences
Ah yes, I've read this many times. It's part of the reason I decided to try this, and the entire reason I plan on ordering a set of BC212L and BC182L transistors. The BC548 and BC549 transistors will be salvaged from my small horde of boards from old Bang & Olufsen gear (Speaking of, wait until you see the chassis this particular project will live in ).Altitude wrote:I support this thread
I have done some work alread in this direction using modern parts (i.e. J201 vs 2SK30). A good resource to look at is the BANNED TM-3030 manual (http://www.oakleysound.com/tm3030.pdf) where he explains what differences there are and why he used what..
-
- Posts: 248
- Joined: Fri Oct 08, 2010 4:49 am
Re: An Attempt to Quantify Semiconductor Differences
Actually, I beg to differ. I've seen piano tuners nail the pitch of every string dead-on equal temperament with nothing but an "A" tuning fork, their ears, and a stopwatch. If what you say is true, it wouldn't matter which chips or transistors we put in our gear as long as they could handle the voltage and current required, and were the correct polarity, because they would all sound the same. Also, if what you say is true, you wouldn't be able to tell the difference between a well played soprano saxophone, an oboe, or a trumpet with a metal straight mute inserted in the bell. Almost anyone can train their ears to do seemingly amazing things. One day, when working with one of the groups I taught a few years back, I stopped the ensemble (a group of 64 brass players, upper high school and college aged), and called a specific trumpet player out on a small error. He chose to question my judgement in the middle of rehearsal. I didn't say a thing. I just turned my back to the group, told the trumpet section, "Ok guys, in random order, you pick, and you can even change the order in which you're standing, I want each one of you to play a "G" one at a time. With my back still turned, and I even closed my eyes, as each kid played a note I called the name of the kid that played it. I called one kid's name twice (pretty clever trick on their part). I stopped after about ten of the 22 trumpets played, turned, and asked, "Ok how many did I get right?" Ever looked at 64 kids with their mouths all just hanging open? I said, "Well?" One of them finally answered, "You got them all...". "Ok, let's start again at letter...". There was no more discussion ever again when I asked a specific person to make a correction. To be honest, I wasn't 100% sure on a couple of them, they were playing on a matched set of instruments. Also, to be honest, I had been working with them for quite some time, and their individual sounds were as different to me as their voices (There is a correlation between those two by the way).antto wrote: whut?
your ears are not measurement devices
you can't "compare" two similar sounds with your ears..
Do I have "magical ears"? Hell no! I wish I did. You are very correct in that ears can't put a calibrated number value on things. Can I tell you if two "A"s played one after the other were specifically 440Hz and 439Hz? No way. Could I tell you the second one was lower pitched than the first? Easily. Believe it or not, to just about anyone that 1Hz difference (at least in that frequency range) is pretty big. Can I tell you if two notes are in tune or not even if their "beat" frequencies are more than ten seconds apart? Yep. Give me two similar sound sources, and I would bet a good meal of your choice that I could teach you how to do it in an hour or less. Give me those two sound sources, a microphone, and an oscilloscope, and I bet two good meals I could teach you how to do it in 15 minutes!
What the human ear can distinguish, discriminate, compare, etc. is pretty amazing to me. And, once you know what it can do in numerical data, you can train yourself to do some seemingly impossible things. For example, do you think you could teach yourself how to tell if the front edge of a sound wave hits one of your ears approximately 0.0004 seconds before it hits the other ear? Sound impossible? It's easy. Even if the wave hits one ear 0.0002 seconds before the other it'd be easy to figure out. Here's how: If you were to hear a sound and turn your head toward it (not just your eyes, but your whole head), and you just happened to turn your head right at 90 degrees, that sound wave hit one ear about 0.0004 seconds before the other. Assuming your ears are about 6 inches apart, it takes sound about 0.0004 seconds to travel that distance, meaning that it met your head on an imaginary straight line drawn through your ears, which means you would have to turn your head 90 degrees to hear where it came from. Of course, for it to be really accurate you would have to know the temperature and humidity, elevation, the exact distance between your ears, the exact angle that you turned your head, the difference in perception in one ear over the other as far as dB or SPL is concerned, etc.
Now, all that said:
You're quite correct, listener fatigue plays a big role in perception of sound, especially when timbre and loudness are what you are focusing on. However, it wouldn't affect our piano tuner that much at all (Seriously).antto wrote:even if you play the same sound twice - you'll hear a different thing every time (especially when it's complex sound material such as music or voice)
That's why I (and others that I hope join in) plan to lay all this out with both oscilloscope measurements (photos mostly) and carefully documented sound recordings. I wish I had a spectrum analyzer! I have a software-based one that can do frequency sweep measurements on speakers, but it only works when IT is generating the frequency sweep that is sent through the speaker.
It's getting late, and not only am I rambling, but my brain is rapidly losing active neurons, so I will shut up now, and hope that everything I just typed agrees with what I THINK I typed when I read it tomorrow .
Oh yeah, the board is more than halfway complete now, with the analog section 90% finished. I need to decide what set of NPN transistors to start with, as I don't have nearly enough 2SC536F transistors to place everywhere that they are called for. Plenty of 2SC945's are available from one of my suppliers that's only a short drive away, not sure if they have the "P" suffix yet.
Planned initial setup will be using all vintage, known working, of the following: LA4140 (even though the signal won't be going through it for the test), 2SC2291, 2SC1583, 2SK30A-Y, 2SK30A-O, and a BA662-A. LM358 will be substituted for AN6562, and 2SA733P subbed for 2SA733AP.
Please be positive and constructive with your questions and comments.