Hi i see. This is fine. If a system sounds good in a controlled situation the problem will not be the system but the listening room maybe.
Still acoustic engineers rely on instruments to check the impact and performance of acoustic treatment of a room.
i do not why given that the same systems will be listened to in rooms very very different from their listening labs. Anyway they can do what they want of course ...
Yes. As i said i am quite willing to accept the idea that a good sound can be obtained also from units performing so so at the lab bench.
But not the viceversa. If so we have focused the wrong parameters in the measurements.
Instruments if correctly used tell so many things that is unbelievable.
Hi ! just to give a little of my background i have always wanted to work in the audio sector. When i was young (13 years old) i had the opportunity to visit an electronic lab and fell in love with scopes/instruments ...
Problem is that between me and electronics there is math ... i am not good at math. At all.
This is one of my biggest regret ... i feel sincerely stupid for this inability.
For me the ability at math is the measure of intelligence.
So i had to give up with the idea of studying electronics and revert to other things. With no passion at all. Just to survive let's say.
So i became a chemical engineer ... still i was decent in the labs (actually more than decent with instruments) but very very bad at engineering.
I passed exams copy here and past there without fully understand what i was really doing.
So in the end i have no a real area of expertize. Maybe if i had studied pure chemistry i could have ended being a decent lab analyst
But i did not ...
Last edited:
Measure the acoustic we hear at listening position, should show better correlation but many found to be very difficult to interpret using traditional methods.
Give voltage measurements on output of amps less focus, see the bigger picture, we listen to the sound, not voltage.
Don't worry, the strong math guys would have not much of an advantage in audio. We have tools to help us with regards to math such as simulators. But repeatable measurement setups and clear understanding of the confounders are very important.
Wine taste is a poor analogy for an audio system. Wine may be an analogy for a musical performance. The issue then is how to convey the wine/performance to the taster/listener without damaging it too much. A lab can tell you if wine has been contaminated, and people may seek correlations between perceived taste and contaminants. Similarly, audio measurements tell us what has happened to the signal, and careful listening tests can seek correlations between perceived reproduction quality and signal degradation.ginetto61 said:
.........
But i did not ...
same in every area - learning and mileage
say that you're schooling for Carpenter (who is working with plenty of things , including Hammer)
say that you're using some simulation proggie while schooling ; you can work in sims all day long with Hammer , with not much difference are you 're beating properly on wood nail , or your own thumb ; only when you start real in vivo practice , first time when you hit your thumb with Hammer - you'll bloody know what difference is that
so , sims are same as Hammer - wonderful tool with which one can do glorious thing , while someone else will do nothing ;
learning and mileage are necessary for efficient use of sims ; they're here to replace old ways , where prototypes are only way of developing and finishing product
now one can start with sims and do majority of work fast , quick and efficient , switching in late phase to real physical prototyping and finishing the product
so , general speaking about actual things is useful , but only as short intro; as ppl are saying from Old Rome to these days - one would learn swimming only when thrown in water**
conclusion - if you want to learn about audio electronics - read and think ; in first time you will and should use sims only to understand how things are working , and only later you'll be able to use it to construct something
and remember - nothing beats whack of Hammer in your thumb
**this is post #26 or so , in thread which is complete hogwash , and it could and needed to end after 3 or so posts , with question asked and response understood ; learning is one thing , while discussions about learning are complete another thing - not good and useful in any way for Students , they're Teachers thing
i read of a quite good test. To record very well a speech. Then play it back with a system and record the sound from the speakers.Measure the acoustic we hear at listening position, should show better correlation but many found to be very difficult to interpret using traditional methods.
The better the system the more intelligible will be the speech after some playback/record passages.
With a bad system the signal degrades very quickly. Because distortions will sum up at every passage.
Yes i took out the speakers from the equation and i understand they will have the biggest impact at least in terms of distortion.Give voltage measurements on output of amps less focus, see the bigger picture, we listen to the sound, not voltage.
Yes and those tools are a god gift ... really nice tools indeed.
Really a good starting point.
I see. One thing is to have a hammer another one to be able to use properly. Nobody said it is easy. But i understand it is a very fundamental design tool and i guess all designers use it commonly. Of course after that there will be a prototype ... that is clear. And after that the lab bench for measurements ...
if i understand well sim sw have inside all the equations to calculate the circuit. A very handy tool that saves a lot of time to designers in selecting different values and parts.
i think you have perfectly answered to my original question.
Sim is the first step in the design process after learning some of the very basic circuits.
Then protype construction > measurements > listening ...
But if the measurements are ok and the listening experience is bad ... then i would be lost. Completely.
The problem is that perception is a tricky, tricky thing to measure, and in fact we don't even have a good language to discuss what we hear (Which is a real headache).....
There are a whole load of words you hear around listening tests that tend to be at best poorly defined.
Also, setting up a test that tells you things about the audio, and ONLY about the audio is actually stunningly difficult, turns out that the look of the box or knowing the price has a surprisingly big impact on perceived audio quality!
Now personally, I like sims for little blobs of a dozen parts that do some particular thing, say a filter or a control loop or something, and sometimes for noise analysis, but I very seldom sim a complete system because setting up the parasitics takes ages and you are never sure you got all the ones that matter.
When you get a repeatable listening test that indicates a problem (and they DO happen, which is why they are a thing), first thing to do is to fire up the Prism or AP and check that the thing is really working correctly (Also, LOOK to check that something is not wired out of phase or with an L/R Reverse).
It is rare to get a repeatable statistically significant listening test result that you cannot then measure if you try hard enough (Usually, such things come down to overload recovery or similar, which you can measure it just takes work).
Note also that there is a massive difference between 'I like the sound of that one' and 'that one is accurate', you need to be VERY, VERY clear on what your objectives are. A SE Triode might sound nice on some material, but few would say it was anything like a 'wire with gain'.
Finally, to a certain extent, forget the electronics, transducers are where the action is if you actually want better sound, the speaker is almost always worse then even a cheap amp by a stupid amount.
Regards, Dan.
There are a whole load of words you hear around listening tests that tend to be at best poorly defined.
Also, setting up a test that tells you things about the audio, and ONLY about the audio is actually stunningly difficult, turns out that the look of the box or knowing the price has a surprisingly big impact on perceived audio quality!
Now personally, I like sims for little blobs of a dozen parts that do some particular thing, say a filter or a control loop or something, and sometimes for noise analysis, but I very seldom sim a complete system because setting up the parasitics takes ages and you are never sure you got all the ones that matter.
When you get a repeatable listening test that indicates a problem (and they DO happen, which is why they are a thing), first thing to do is to fire up the Prism or AP and check that the thing is really working correctly (Also, LOOK to check that something is not wired out of phase or with an L/R Reverse).
It is rare to get a repeatable statistically significant listening test result that you cannot then measure if you try hard enough (Usually, such things come down to overload recovery or similar, which you can measure it just takes work).
Note also that there is a massive difference between 'I like the sound of that one' and 'that one is accurate', you need to be VERY, VERY clear on what your objectives are. A SE Triode might sound nice on some material, but few would say it was anything like a 'wire with gain'.
Finally, to a certain extent, forget the electronics, transducers are where the action is if you actually want better sound, the speaker is almost always worse then even a cheap amp by a stupid amount.
Regards, Dan.
Everything relevant has been typed, I'll still give my own take: you need to know what you want to figure out when simulating a circuit. If you want results that mirror real life behaviour as close as possible you need to enter as much data as possible, which includes all parasitics. When dealing with signal transformers especially parasitics.
And you need to learn how to measure and you need to listen. If things develop nicely patterns will show correlating simulated behaviour, measurements and subjective reception of the built appliance.
Sidenote 1: I developed certain biases that way and can make choices that limit the range of variants to try out. Still I get baffled regularily by experiences that just shouldn't be...
Sidenote 2: you need to know your parasitics and which matter and which don't in a given situation. In my job most of what I do is FEM simulations. To get useful results experience was needed to know what to enter and what to ignore. Also we do batch simulations, dealings with tolerance ranges of parts to be delivered by our suppliers. That's the funky part where having powerful calculators come in handy and I can spend the morning setting up a simulation, let it run and go to lunchbreak and have data when I'm back in the office with a coffee. I still remember booking slots and hoping I didn't waste 78h of mainframe time
And you need to learn how to measure and you need to listen. If things develop nicely patterns will show correlating simulated behaviour, measurements and subjective reception of the built appliance.
Sidenote 1: I developed certain biases that way and can make choices that limit the range of variants to try out. Still I get baffled regularily by experiences that just shouldn't be...
Sidenote 2: you need to know your parasitics and which matter and which don't in a given situation. In my job most of what I do is FEM simulations. To get useful results experience was needed to know what to enter and what to ignore. Also we do batch simulations, dealings with tolerance ranges of parts to be delivered by our suppliers. That's the funky part where having powerful calculators come in handy and I can spend the morning setting up a simulation, let it run and go to lunchbreak and have data when I'm back in the office with a coffee. I still remember booking slots and hoping I didn't waste 78h of mainframe time
Last edited:
Those who insists on wire with gain has a certain attachment to keep the high distortion of their speakers as pristine as possible. Member Circlomanen made rudimentary measurement on reduction of distortion on SE topology in "Triodity" to cancel loudspeaker distortion. Works with single driver and multi amp systems, difficult to impossible to apply to multiways.
Walk before you run. Crawl before you walk.
As Osvaldo says, electronics guys built a LOT of stuff before simulators, even before useful analytical tools. Set up one device (horn, carbon-amp, tube, transistor, etc) and try different power, bias, and input/output coupling networks.
The crosstalk of a poor Balance control can be approximated on your thumbs. Say there is 10k resistance in each side path and 10r resistance in the common ground path. That's 1,000:1 or 60dB crosstalk. That's simplified (source and load resistances must be accounted; also stray capacitance) but we rarely need an exact number. How much crosstalk is acceptable?? Objective studies say 30dB is inaudible. You feel better with Balance network removed. Sometimes setting "goals" is harder than doing the math.
> an expert jumps in saying that it will not work ... how can it be sure about that ?
You know chemistry (I do NOT). Radon
is a problem here. Say I propose to react Radon with Helium to make the compound HeRn, which might be harmless, or might be easier to separate and remove from air or water. You are the expert. Will it work? How can you be sure?
I had read the big old authors, like Gray, Seely, Seeley, Langford-Smith, Wallman, Millman, etc, and still in the waiting state is Terman. I love this old text books, because they are well explained and lots of maths and examples. Fink explains in great detail the design concerning old radars. And they did it without simulation(ors) of any kind. Only a big big big mathematical background and lots of money to spend in time, journals and time.
The key is to design the most important stage in the set, usually the last (Power), analyze the requirements needed to extract the most of it can give, and going back to the input to satisfy the input requirement (Load, source, noise, etc) . And in between them, you need to study interactions, and finally, make a prototype. The pic below is my prototype of a full compactron tube no SS, DC coupled audio amp. More than 2 years working ion it. I found some problems derived from the capacitive coupling between last stage plates to input stage. How could I imagine it in a simulation?
Last pic is how I started the project: the two 6JN6's alone.
The key is to design the most important stage in the set, usually the last (Power), analyze the requirements needed to extract the most of it can give, and going back to the input to satisfy the input requirement (Load, source, noise, etc) . And in between them, you need to study interactions, and finally, make a prototype. The pic below is my prototype of a full compactron tube no SS, DC coupled audio amp. More than 2 years working ion it. I found some problems derived from the capacitive coupling between last stage plates to input stage. How could I imagine it in a simulation?
Last pic is how I started the project: the two 6JN6's alone.
Attachments
Hi i beg to differ.
I can use a calculator to calculate the square root but if i had to calculate it myself i would not be able. I do not remember the method
The very and historical fundamental issue is that it seems very difficult (almost impossible ?) to correlate measurements with the actual sound quality. (Assuming that there is an ideal good sound quality)
The fact that units excellent at the lab bench sound bad it is the sad evidence.
I say sad because i like objectivity always ...
Gino, you show confusion on a very simple issue, we listen to acoustic sound, not electric voltages nor current. Measure acoustic output to arrive at some correlation in a simpler manner, end of story. Correlating measurement of voltage and current at amp output to perceived acoustic output is actually more complex and could easily take a lot of money for many years.
Not yet. Well, okay, everything relevant has been typed, and so everything irrelevant. Now the problem is to know which one is which.
Hi ! actually my idea was to start by simulating very basic circuits like for instance a diamond buffer and try different bjts/fets, voltages, resistors ... in order to try to understand ho circuits behave at least on paper.
I would have no chance to calculate a diamond buffer on my own
Then when i find a schematic that "simulates" well i will try to build a prototype.
I have the feeling that many commercial schematics could be improved only by changing parts and working conditions ...
Thank you very much for the kind explanation but my "ideal" preamp has only a power switch, a volume control in the middle and a input selector on the right ... and the same for an integrated.
My goal is to hear the more of a recording ... the good and the bad ... without make-up ... without "soundshopping" (ref. Photoshop sw)
Actually in chemistry chemists study substances and develop processes ... chemical engineers usually have not a very deep knowledge of pure chemistry. After that a process is available engineers should take care of scaling-up the lab reactors set-up by chemists to an industrial size.You know chemistry (I do NOT) Radon
Say I propose to react Radon with Helium to make the compound HeRn, which might be harmless, or might be easier to separate and remove from air or water. You are the expert. Will it work? How can you be sure?
I mean the good ones ... the bad ones just wait for retirement.
To be honest the only thing i have learned very well about chemistry is to stay away from it ... the first time i visited a chemical plant for a job interview i asked myself ... did i kill anyone to work here ?
Seriously it is a very dangerous job for health in the long term
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.