An Attempt to Quantify Semiconductor Differences

[antto]

come on..
i'm trying to mimick the sound and behaviour of the 303 (in software) since almost 2 years or more (i don't know, it's all i'm doing)
i have a huge number of recordings from different sources (303s, x0xb0xes)
there are slight differences in lots of places
YOU CAN'T use your ears to tell the difference, no, you can't!
you can only tell "yeah in this recording the square is different than the square in that recording"
but that's if they are playing the same note (or same pattern)
..and if they are tuned identically (which is rather hard)
then, what's different in the two recordings? "they sound different" is not enough
there are quite a number of things that can differe in the square wave alone

wanna talk about the filter? it has a number of HP filters in which might vary a lot in frequency between different units (different capacitors and resistors)
how the hell can your ears tell that? - no way
what is needed is: scope/graph, sonogram, calculator, and someone well trained in using them, experience with comparing audio materials is a must too

blah
EDIT: try and tell me what do you "hear" in this recording
is it the same synth?
is it 2 synths?
is it 1 synth but recorded at different periods?
do you hear something different? - WHAT?
okay, i'm waiting

[Brassteacher]

blah
EDIT: try and tell me what do you "hear" in this recording
is it the same synth?
is it 2 synths?
is it 1 synth but recorded at different periods?
do you hear something different? - WHAT?
okay, i'm waiting

Having too much to type in way of a response, I recorded my impressions, since you're trying to test me...
Test Response

[phono]

blah
EDIT: try and tell me what do you "hear" in this recording
is it the same synth?
is it 2 synths?
is it 1 synth but recorded at different periods?
do you hear something different? - WHAT?
okay, i'm waiting

sounds like 2 different synths to me, played one after the other, first one perhaps a x0x the second software? just a guess but both sound a bit digital but that could be the mp3 compression, the second one sounds a lot less brighter. Some of the accents on the first one sound odd also, so it could all be software, tbh who knows..

okay, i'm waiting

[Brassteacher]

blah
EDIT: try and tell me what do you "hear" in this recording
is it the same synth?
is it 2 synths?
is it 1 synth but recorded at different periods?
do you hear something different? - WHAT?
okay, i'm waiting

sounds like 2 different synths to me, played one after the other, first one perhaps a x0x the second software? just a guess but both sound a bit digital but that could be the mp3 compression, the second one sounds a lot less brighter. Some of the accents on the first one sound odd also, so it could all be software, tbh who knows..

okay, i'm waiting

Similar to my first impression. Conditions: Only first 2 minutes of file listened to, not downloaded, played back through browser, internal speaker of my old G5 (which is the weak link of the machine ). I intend to download it, burn it to CD, and play it back through my main listening system later.

First impression up to 2:12 under conditions stated above: 1. Software synth (and/or this is a low bit rate mp3), 2. Same one each time, 3. It's a midi file as opposed to played live, 4. Either the filter is modeled based on a quite different machine than I've heard, or the resonance has been enhanced. 5. The decay sounds a little "stair-stepped", and/or an artifact of compression, and/or more resonance than i've heard so far (see #4).

Given the listening conditions, I could be way off base, I'll get a better read when I download and burn it. Also, I'll post the real (longer) response to this that I started typing earlier.

[antto]

i can try to upload the original .wav file (right before encoding to mp3)

but what i meant with this is a demonstration of how small the difference is
the fact that you can't compare _now_ is one aspect
you need specific speakers?
do you position your head in some special spot in order to be in the "center" of the speakers, or do you have some expensive room with _no_ reflections?

in other words, if you try a blind test, and someone plays you some sounds and you have to tell if you heard anything different - you'll fail pretty bad
listening to the same sound twice - you can hear 2 different things, it's just like the eyes, your brain fools you, you see what you wanna see, you hear what you wanna hear, the missing parts are automagically filled with something "appropriate" by your brain
do you also try the temperature with your hand (or other part of the body) in the oven? 180°C when you cook a cake?
i hope i made myself clear
i got good ears too, maybe not as good as yours (i can hear up to 17.9KHz) but this has nothing to do with sound analisys and comparison or measurement
your ears are not spectrographs.. even if you hear a "perfect square" which is one of the easiest sounds - you can't draw it like an oscilloscope, because your ears can't catch the phase, thus a square wave sounds the same even when fed thru an allpass filter, and we all know how it looks after the allpass filter, while it sounds the same

[the big bear]

Hi Dude
Personally i think what you are doing is a great idea, i have bc182l bc184l and 212l in stock also some fets bf244b if you want i can send them to you. I have been involved with synthesizers for 25 years and one thing i can say is that people hear with there eyes, i had a mate who put a 12db transconductor filter in a minimoog and pretty much every one who heard it said how great it sounded!! I've got another mate who has a black and orange odyssey and it does not sound like an arp and another mate has got a 303 from new and the filter oscillates, trust your ears because banned sounds banned and if you get a great combination no mater how clean it looks on a scope that's what to go for. And that's all I've got to say on the matter. TBB

[Altitude]

Don, don't forget the other production parts like the 945 and the other one that has been mentioned. There also different manufactures to take into consideration

[Brassteacher]

Don, don't forget the other production parts like the 945 and the other one that has been mentioned. There also different manufactures to take into consideration

Oh, don't worry, I'm picking up some 945's tomorrow along with quite a few other goodies (hopefully). Managed to get enough 945's to be able to sort for Hfe. Now all I have to do is decide which of my meters that can test transistor beta is more accurate as far as the absolute number is concerned. I sorted my 733P's last night, one meter read the same transistor, at the same temperature, 288, while the other said 312.

Oddly enough, very few of my older transistors have maker's marks, other than a few K30A's made by mitsubishi.

I'm torn between building the x0x main board out with everything installed, as it would be in a kit, so that the current draw will be similar to what everyone else will experience, or building it out and omitting anything not necessary to run the thing off midi input, such as not plugging in the chips for the sequencer, not installing USB-specific components (probably will omit USB no matter what), not installing the headphone amp and mixer components, taking the output directly off the wiper of the volume pot via a decent coupling cap, at least temporarily. There are valid arguments either way. Of course, since every active component except for diodes are socketed, I could do it both ways, but that would double the amount of things to test.

[Brassteacher]

come on..
i'm trying to mimick the sound and behaviour of the 303 (in software) since almost 2 years or more (i don't know, it's all i'm doing)
i have a huge number of recordings from different sources (303s, x0xb0xes)
there are slight differences in lots of places
YOU CAN'T use your ears to tell the difference, no, you can't!
you can only tell "yeah in this recording the square is different than the square in that recording"
but that's if they are playing the same note (or same pattern)
..and if they are tuned identically (which is rather hard)
then, what's different in the two recordings? "they sound different" is not enough
there are quite a number of things that can differe in the square wave alone

wanna talk about the filter? it has a number of HP filters in which might vary a lot in frequency between different units (different capacitors and resistors)
how the hell can your ears tell that? - no way
what is needed is: scope/graph, sonogram, calculator, and someone well trained in using them, experience with comparing audio materials is a must too

blah
EDIT: try and tell me what do you "hear" in this recording
is it the same synth?
is it 2 synths?
is it 1 synth but recorded at different periods?
do you hear something different? - WHAT?
okay, i'm waiting

You know, I've been thinking for a few days just how respond to this. I've decided to just wait, and let you see what I have in mind. You will find that we really don't think that differently about things. However, I will ask you to not make judgments about what I can and cannot do until AFTER I have made a complete and total display of my ignorance in public, then go for it .

[Brassteacher]

i can try to upload the original .wav file (right before encoding to mp3)

but what i meant with this is a demonstration of how small the difference is
the fact that you can't compare _now_ is one aspect
you need specific speakers?
do you position your head in some special spot in order to be in the "center" of the speakers, or do you have some expensive room with _no_ reflections?

Ok, previously you lectured me on the need for quality test equipment, now you begrudge me the most important piece of gear. With the kind of work I do for a living, that stereo system in my living room is where everything has to pass the final quality test. Would you rather me judge your work on a 1.5 inch computer speaker that barely covers the range of the human voice, or on a decent rig? I don't feel like quoting specs at the moment, I'd rather go solder. Besides, specs of the playback equipment will be included later, since that is a very major factor in such types of evaluation.

in other words, if you try a blind test, and someone plays you some sounds and you have to tell if you heard anything different - you'll fail pretty bad
listening to the same sound twice - you can hear 2 different things, it's just like the eyes, your brain fools you, you see what you wanna see, you hear what you wanna hear, the missing parts are automagically filled with something "appropriate" by your brain
do you also try the temperature with your hand (or other part of the body) in the oven? 180°C when you cook a cake?
i hope i made myself clear
i got good ears too, maybe not as good as yours (i can hear up to 17.9KHz) but this has nothing to do with sound analisys and comparison or measurement
your ears are not spectrographs.. even if you hear a "perfect square" which is one of the easiest sounds - you can't draw it like an oscilloscope, because your ears can't catch the phase, thus a square wave sounds the same even when fed thru an allpass filter, and we all know how it looks after the allpass filter, while it sounds the same

AGAIN, please don't decide what I can and cannot do until I have had a chance to get this thing rolling, I haven't even finished building the sucker yet (about to head out to the shop though, insomnia combined with boredom often makes for good progress). Keep an open mind, and give me a fair chance to prove I'm an idiot! Hell, I may turn out to be even crazier and dumber than you think!

[antto]

i'm sorry
i'm just tired of people posting doubtful information about the 303 (and thus x0x)
when i started, a few years ago, i just wanted to get the sound of the 303, but as i don't have one - i started digging every piece of information i can find on the internet
articles like robin whittle's and other
my biggest mistake was i blindly believed all of the banned..

and as i began to advance a little i really started to get in the wrong direction by following such information sources

this is not my thread, but i would just like to prevent yet another piece of information to become doubtful

on the other hand, if the analysis you make out of this whole thing here, are layed out with good technical details, measurements, and so on.. it might become one (of the very few) useful pieces of information

don't get me wrong, i'm not trying to say you don't have good ears or equipment
but the results you post here - some one might try to derive algorithms out of it, and he'll either go into the right direction or into one of the many many many wrong ones..
eventually i'm one of those guys..

forget my audio recording.. don't waste your time, 2 is the softsynth, 1 is the x0xb0x, and there was no midi whatsoever

[stevenclements]

damn... sounds like a lot work for what?

write a song... if peeps dig it, cool

if not

write another song...

repeat, repeat

have a drink

[Brassteacher]

damn... sounds like a lot work for what?

write a song... if peeps dig it, cool

if not

write another song...

repeat, repeat

have a drink

You know, you're right. I guess I didn't make the main thought in my head known, or clear:

So many people want to build/have one of these neat little machines, but there is a finite supply of suitable parts to build them. The main goal is to establish a general idea of what the x0x sounds like with all "vintage", "authentic", or "correct" active parts, then experiment with easily available alternatives to see which have no measurable or audible effect on the sound.

The info may make it easier to build an acceptable machine without breaking into a museum to get parts! Also, since I'm in the analog synth repair/restore business, I need to know what types of things I can and can't sub out when working on things. 2SC1583, 2SK30A's of various flavors, 2SC945, BA662, 2SC733 appear in a lot more of the older Roland synths than just the TB-303, and with the exception of the BA662, a lot of other Japanese analog synths, especially Korg.

[Brassteacher]

Quick update: Build almost completed, delayed by the fact that I had forgotten to finish etching the I/O board. All parts in, including substitutes to test, except I don't have any alternative PNP transistors to play with, other than 2SA1015 and 2N3906, which weren't exactly first on my list to try. I'm going to dig through the "boneyard" and see if I can find enough compliments to the BC549B to play with.

Pics of work done so far coming soon. I need to at least partially fit the machine into its case, which is definitely different and far removed from the "standard" x0x case, or even the original TB-303. You guys will either love it, or hate it! Hint: it's recycled...

[bcbox]

I have also done a lot of this testing.
My number one tool is my ears. I use a single studio monitor and listen... in mono. If I can't hear a difference I don't care. If I do hear a difference I test on an Audio Precision system. I have done tons of tests and comparisons. I've also tested many transistors on a curve tracer. All of my results are in my engineering notebook and in test files... wav files and AP files. I'm content to wait and see what comes about here instead of tainting the results with any of my data.