I have learned a good deal here at Georgia Tech.
Prof. Leach is a great teacher and teaches analog circuits and audio engineering which offer a great start into the field of amplifier design. I have learned a lot from just going into his office hours and using his audio lab to build and test my amplifier.
--
Brian
Prof. Leach is a great teacher and teaches analog circuits and audio engineering which offer a great start into the field of amplifier design. I have learned a lot from just going into his office hours and using his audio lab to build and test my amplifier.
--
Brian
It is really off the track from the "new law" thread. 😉
I think a engineering degree only gives you a 'basic' understanding of the material and tools (math .. physics ..) to work with. How they behave.
The rest is up to the engineer to figure out, how to combine what he/she had learned with new stuff wich pop up all the time (This part is selfstudy)... He/she has to use his brain! (sorry.. but i could not leave it)..
The engineering degree also will give you some basic circuits to work with, so you do not have to start from the scratch..
If the teacher don't tell you this. Why do you people not complain to the boarder!? Just because he/she has a name or the university has a name does not mean that he/she is a good teacher.
Sonny
I think a engineering degree only gives you a 'basic' understanding of the material and tools (math .. physics ..) to work with. How they behave.
The rest is up to the engineer to figure out, how to combine what he/she had learned with new stuff wich pop up all the time (This part is selfstudy)... He/she has to use his brain! (sorry.. but i could not leave it)..
The engineering degree also will give you some basic circuits to work with, so you do not have to start from the scratch..
If the teacher don't tell you this. Why do you people not complain to the boarder!? Just because he/she has a name or the university has a name does not mean that he/she is a good teacher.
Sonny
Learnings...
Brian,
I'm always interested in learning as much as possible from others. I would be very keen to know what measurements Prof L recommends as most important to sound quality. I mean, once the circuit is basically operating correctly and does what it should on paper then what, say, top 3 measurements reveal most about the sound quality?
BAM
Brian,
I'm always interested in learning as much as possible from others. I would be very keen to know what measurements Prof L recommends as most important to sound quality. I mean, once the circuit is basically operating correctly and does what it should on paper then what, say, top 3 measurements reveal most about the sound quality?
BAM
jduncan and Harry,
cannot say i have a deep clue on how electrical CAD (Spice) works and what the designer is required to tell the system in order to get reasonable results. I worked with ECAD systems like Orcad, Eagle, PADS, Ultiboard and the like and tried out their layout autorouter, even poor non-knwong me was disapointed iwth the results, can it be that an 8-bit data bus marked as such has >700% route length variation and the several routes spread over the whole PCB?
So i presume it is like with mechanical CAD (a thing i am quite competent in): you have to tell the system EVERYTHING which means you have to have everything in your head before .. else the system fools you in style (and if you are a king-size fool, you may not even notice, observed this very often)
CAD is a tool, a brilliant tool, but nothing more, you have to know what to do with it.
One of my professors lectured design and CAD design and he was utterly competent and pedagogically gifted. He put a lot of effort in teaching students how to use CAD. I still remember his body language when he patiently/repeatedly tried to explain a student basics like how and where to select a geometry element in order to tell the system where you want the reference and failing because the student had no concept what a reference suiting his needs was. The student didn't have it in the brain!
I observed this quite often. THe best thing my professor did to me was to throw me in the cold water, saying, "you have six weeks to present your fancy turntable at the job fair with CATIA V3, six weeks to learn to master the system and to draw/model the TT and you ask me what you want and after six weeks i wanna be proud of you" and i had to prove i had what it took . He knew he could not teach it, i would have to have it, all he could do is providing answers and answers. He just knew he could not teach any of his students, the students had to find out themselves, he just was not permitted to admit it 🙂 . He was there to provide answers for the (sometimes rare) case there would be questions.
Today i work with "Pro/Engineer" a parametric system not forgetting anything, not any goof, forced history, building the model from dimensions and constraints and not forgiving any foggyness or contradiction. And and now and then i have to work on a CAD model i get from colleagues having fancy CAD diplomas, say, an assembly having symmetry in X and Y but not in Z direction. And guess where the fools put the reference planes??? Yes, crossing at the lowermost, leftmost point of the assembly. Killing any possible benefit from symmetry with their 1st step. 🙁
Don't blame the schools for that, blame the guys having left their brain with coat and hat at the entrance.
how many?! electronics engineer
who or what that have to blame ?
how many engineer can heard the different sound of an amplifier that used fast soft recovery diode and not use it?
how many engineer can heard the different sound of an amplifier that biased in class A and class AB?
how many engineer can heard the different sound of an amplifier that using tubes and non tubes?
how many engineer can heard the different sound of an amplifier using feedback and not using it?
how many engineer can heard an amplifier that have a lot of second harmonic distortion or not?
how many engineer can sing or play a musical instrument?
how many engineer really do their homework?
how many engineer that was born as an engineer and how many do not?
how many......................................
😕
who or what that have to blame ?
how many engineer can heard the different sound of an amplifier that used fast soft recovery diode and not use it?
how many engineer can heard the different sound of an amplifier that biased in class A and class AB?
how many engineer can heard the different sound of an amplifier that using tubes and non tubes?
how many engineer can heard the different sound of an amplifier using feedback and not using it?
how many engineer can heard an amplifier that have a lot of second harmonic distortion or not?
how many engineer can sing or play a musical instrument?
how many engineer really do their homework?
how many engineer that was born as an engineer and how many do not?
how many......................................
😕
Prof. Leach & AES......
I stopped sending money to the AES about 20 years ago. Everytime the good professor wrote something that made sense, a bunch of smarmy egghead PhD types (from some "bootcamp" somewhere) would rip him to shreds.
How did this thread get sidetracked his far? Hairy Holler.....it is all your fault. No, you can't have my autograph.
Someone asked that I "spread my wisdom". I'm open to suggestions on what topics, but let's move it somewhere, OK?
Jocko
I stopped sending money to the AES about 20 years ago. Everytime the good professor wrote something that made sense, a bunch of smarmy egghead PhD types (from some "bootcamp" somewhere) would rip him to shreds.
How did this thread get sidetracked his far? Hairy Holler.....it is all your fault. No, you can't have my autograph.
Someone asked that I "spread my wisdom". I'm open to suggestions on what topics, but let's move it somewhere, OK?
Jocko
How to succeed in audio with out really listening
Come guys lets step up the pace! Those guys in the " DIY Video Projector" thread are kicking our butts for shear volume of output.
Grey, it's time for your help. A guy with a thousand post surely can help us out. Jocko.....no autograph huh? I guess I will have to save the next schematic you draw. How about a trip to Tanners
this week. I saw a whole wall of surplus transistors and some oldy but goody small Polystyrene caps.
H.H.
Come guys lets step up the pace! Those guys in the " DIY Video Projector" thread are kicking our butts for shear volume of output.
Grey, it's time for your help. A guy with a thousand post surely can help us out. Jocko.....no autograph huh? I guess I will have to save the next schematic you draw. How about a trip to Tanners
this week. I saw a whole wall of surplus transistors and some oldy but goody small Polystyrene caps.
H.H.
P31R,
Thought I'd respond to your "How many engineer..." rant.
The art of engineering is to develop a product with the following characterists:
- Does something a customer will pay money for.
- Meets cost target, otherwise you don't make any profit.
- Meets manufacturability requirements, otherwise you can't make enough to meet demand (also plays into meeting cost targets).
- Meets durability requirements, otherwise nobody will buy from you again.
- Meets agency approval requirements like FCC and UL, otherwise you can't legally sell it.
- Meets development schedule requirements, otherwise someone else beats you to market and grabs market share from you.
There are other design objectives, but these are the truly important ones. Notice that only one of them is effected by how well the product sounds and even then only in varying degrees.
Sometimes the absolutely best raw performance (sound quality in this case) is what is most important, but this is very rare. You don't see that many people driving Formula 1 race cars to the grocery store do you? To get top performance in one area you usually end up having to sacrifice other desirable features, like convenient features, low price and ready availability. If the best sound quality is all that matters then why are MP3's the hottest selling sound product today.
The fact is that audiophiles are an insignificant portion of the audio electronics marketplace. For every audiophile dollar spent there is hundreds, maybe thousands of dollars spent on bottom feeder sound gear. Don't rant about the engineers. If the objective I have been given is to develop a $10 product then it doesn't make any difference how good my $100 solution sounds, I still get fired.
The most important thing a company has to do if it wants to succeed (and survive) is to figure out who is their customer and what will they pay money for. A big thing that some seem to miss is that it is just as important to decide who is not your customer. Nobody can make everything to satisfy everybody and you will doom yourself to failure if you even try.
Talk is cheap, but have you ever had to fire someone because there wasn't enough money coming in to make payroll? It happens, and if you don't make what people want to buy, it happens right away. The DotCom companies found that out last year.
Phil
Thought I'd respond to your "How many engineer..." rant.
The art of engineering is to develop a product with the following characterists:
- Does something a customer will pay money for.
- Meets cost target, otherwise you don't make any profit.
- Meets manufacturability requirements, otherwise you can't make enough to meet demand (also plays into meeting cost targets).
- Meets durability requirements, otherwise nobody will buy from you again.
- Meets agency approval requirements like FCC and UL, otherwise you can't legally sell it.
- Meets development schedule requirements, otherwise someone else beats you to market and grabs market share from you.
There are other design objectives, but these are the truly important ones. Notice that only one of them is effected by how well the product sounds and even then only in varying degrees.
Sometimes the absolutely best raw performance (sound quality in this case) is what is most important, but this is very rare. You don't see that many people driving Formula 1 race cars to the grocery store do you? To get top performance in one area you usually end up having to sacrifice other desirable features, like convenient features, low price and ready availability. If the best sound quality is all that matters then why are MP3's the hottest selling sound product today.
The fact is that audiophiles are an insignificant portion of the audio electronics marketplace. For every audiophile dollar spent there is hundreds, maybe thousands of dollars spent on bottom feeder sound gear. Don't rant about the engineers. If the objective I have been given is to develop a $10 product then it doesn't make any difference how good my $100 solution sounds, I still get fired.
The most important thing a company has to do if it wants to succeed (and survive) is to figure out who is their customer and what will they pay money for. A big thing that some seem to miss is that it is just as important to decide who is not your customer. Nobody can make everything to satisfy everybody and you will doom yourself to failure if you even try.
Talk is cheap, but have you ever had to fire someone because there wasn't enough money coming in to make payroll? It happens, and if you don't make what people want to buy, it happens right away. The DotCom companies found that out last year.
Phil
Haldor,
Agree whole-heartedly. You want audiophile, take a look at Pass' or Mark-Levinson's prices. I wouldn't pay that. I know it's better. I just won't be buying any.
And, of the ones that will pay it, many are taken in by pseudo-scientific gobbledy-gook and end up buying $1000 power cords.
So, folks, don't blame us engineers: we're just doing our job the best way we know how.
Agree whole-heartedly. You want audiophile, take a look at Pass' or Mark-Levinson's prices. I wouldn't pay that. I know it's better. I just won't be buying any.
And, of the ones that will pay it, many are taken in by pseudo-scientific gobbledy-gook and end up buying $1000 power cords.
So, folks, don't blame us engineers: we're just doing our job the best way we know how.
let's get back to the plot
Ok Jocko...some questions for the thread to nibble on. I have an endless number of questions, but here are three biggies to start with:
1) In a push-pull transistor output stage, say complementary darlingtons, what specific parameters need to match for it to sound good?
2) Why does the application of negative feedback to some circuits make the sound worse, such as harden, lose detail, become boomy, lose emotion or induce headaches after prolonged listening?
3) How should a LTP input stage be designed/laid out to make it as immune as possible from radiated noise - such as pick-up from high speed (high dV/dt) output circuits?
BAM
Ok Jocko...some questions for the thread to nibble on. I have an endless number of questions, but here are three biggies to start with:
1) In a push-pull transistor output stage, say complementary darlingtons, what specific parameters need to match for it to sound good?
2) Why does the application of negative feedback to some circuits make the sound worse, such as harden, lose detail, become boomy, lose emotion or induce headaches after prolonged listening?
3) How should a LTP input stage be designed/laid out to make it as immune as possible from radiated noise - such as pick-up from high speed (high dV/dt) output circuits?
BAM
3 "biggies"
1. I'm not so certain that they have to match, but I look for:
A.) Linearity, i.e., Ic vs Vce as a function of Ib. If the part is not linear to start with, all the feedback in the world will not really fix the problem. Remember, there is that very brief instant where the error voltage has to catch up with the input.
B.) Ft
C.) Cob.
Those 2 generally go hand-in-hand. The typical amp has 3 poles, with the dominat one set by the voltage gain stage. You want the pole of the output stage as far removed from it as possible to prevent instabilities. Especially if there is a capacitive load.
2. This could be an entire thread by itself, possibly generating more verbage than this rant. I believe (emphasis on believe) that anything you do to the audio signal must be applied equally across the audio band. (Anybody remember the TIM debate from the late 70's? Similar thought process here.)
Several years ago, I took an old GAS Grandson amp, and just to prove a point, I ran a series of listening tests as I lowered the open-loop gain. The more I lowered it, the more open the sound stage became, things sounded more natural. The bass wasn't quite as tight maybe........
I think a lot of factors enter in. TIM, for one. You have other IM products generated with RF getting in. A high feedack amp is not very linear to start with. As I pointed out in #1 above, the is that brief instant..........
Which reminds me........Prof. Leach wrote an article in the AES about 18-20 years ago explaining why you needed large degeneration resistors in the front-end of an amp. The point was that the input stage should NEVER be able to clip because there was that error signal thing. One of his usual detractors later opined pompously on his thoughts on amp design, and concluded that you shouldn't need anything larger that 50 ohms. Maybe 100 in extreme cases.
My amps use 2.5 K in each leg. Guess who I believed.
You are keen to pick up on the listener fatigue problem. I advise people that anytime they hear something that grabs their attention, don't buy it. It will always be boomy, bright and nasty, or both. Those sort of things grab peoples attention, and dealers love stuff like that. Which is probably why everyone here builds their own.
I know several designers, many of whom people here would recognise. We all say the same thing: when you build something good, you don't know it until several hours later. It sounds ok at first, but when you realise that you have spent hours listening to it (without getting a headache), then you know you have a winner. Which means the dealers will hate it..........(The average dealer does not want to spend more than 5-10 minutes screwing with a customer. Give them something that gets yuppies' attention, close the sale, rake in that 50 point margin and move on.)
Anyway back to feedback..........
The same thing goes for power supplies too. I prefer supplies that have work the same way as above.
A way to think of an amp is that it is basically a modulated power supply. Which is why lots of peole I design with feel that the power supply is roughly......say 75%.....of the sound of an amp. Most of the other stuff you do has a much lesser effect. Assuming we are right, then visualize you are basically listening to the filter caps going through a follower.
Not really on topic, but here are some tests to run on amps. And supplies too.
Hook a resistor, 8 ohm, 4 ohm, whatever, from one channel of a stereo amp to the other. Drive one channel with a sine wave, and measure the voltage at the other output terminal. Make a graph varying the frequency and drive voltage. Try it sometime......
Hook up a square wave generator to the output terminal. Drive it with various frequencies and levels. Look at how it behaves. Lots of ringing and stuff, and you have problems. Works on power supplies too. I believe Bob Pease suggests this also.
3. Not sure I understand your terminology. Please clarify the situation.
You always need to bandwidth limit the input signal. Just because you can build an amp flat to 2 MHz, doesn't mean that you need to amplify that crap. And if it isn't......well here comes the IM products. I believe HiFi News ran an article......late 80's I think....showing the effects of RF generated IM in a line stage. Forgot the details......that was a long time ago.......but the point was to keep out the stuff that doesn't belong there. Which is another reason I fell ICs sound really bad in general. They are much more susceptible to this sort of problem. (Someone who is routinely ripped apart here uses transformers for just this reason.)
In any case, good grounding is essential. For lots of reasons. Balanced inputs not only get rid of ground-loops, but help to remove stuff that you don't want. (Balanced inputs have other advantages, like getting rid of modualtion on power supply rails, and lowering gain requirements.)
Is this any help?
Jocko.
1. I'm not so certain that they have to match, but I look for:
A.) Linearity, i.e., Ic vs Vce as a function of Ib. If the part is not linear to start with, all the feedback in the world will not really fix the problem. Remember, there is that very brief instant where the error voltage has to catch up with the input.
B.) Ft
C.) Cob.
Those 2 generally go hand-in-hand. The typical amp has 3 poles, with the dominat one set by the voltage gain stage. You want the pole of the output stage as far removed from it as possible to prevent instabilities. Especially if there is a capacitive load.
2. This could be an entire thread by itself, possibly generating more verbage than this rant. I believe (emphasis on believe) that anything you do to the audio signal must be applied equally across the audio band. (Anybody remember the TIM debate from the late 70's? Similar thought process here.)
Several years ago, I took an old GAS Grandson amp, and just to prove a point, I ran a series of listening tests as I lowered the open-loop gain. The more I lowered it, the more open the sound stage became, things sounded more natural. The bass wasn't quite as tight maybe........
I think a lot of factors enter in. TIM, for one. You have other IM products generated with RF getting in. A high feedack amp is not very linear to start with. As I pointed out in #1 above, the is that brief instant..........
Which reminds me........Prof. Leach wrote an article in the AES about 18-20 years ago explaining why you needed large degeneration resistors in the front-end of an amp. The point was that the input stage should NEVER be able to clip because there was that error signal thing. One of his usual detractors later opined pompously on his thoughts on amp design, and concluded that you shouldn't need anything larger that 50 ohms. Maybe 100 in extreme cases.
My amps use 2.5 K in each leg. Guess who I believed.
You are keen to pick up on the listener fatigue problem. I advise people that anytime they hear something that grabs their attention, don't buy it. It will always be boomy, bright and nasty, or both. Those sort of things grab peoples attention, and dealers love stuff like that. Which is probably why everyone here builds their own.
I know several designers, many of whom people here would recognise. We all say the same thing: when you build something good, you don't know it until several hours later. It sounds ok at first, but when you realise that you have spent hours listening to it (without getting a headache), then you know you have a winner. Which means the dealers will hate it..........(The average dealer does not want to spend more than 5-10 minutes screwing with a customer. Give them something that gets yuppies' attention, close the sale, rake in that 50 point margin and move on.)
Anyway back to feedback..........
The same thing goes for power supplies too. I prefer supplies that have work the same way as above.
A way to think of an amp is that it is basically a modulated power supply. Which is why lots of peole I design with feel that the power supply is roughly......say 75%.....of the sound of an amp. Most of the other stuff you do has a much lesser effect. Assuming we are right, then visualize you are basically listening to the filter caps going through a follower.
Not really on topic, but here are some tests to run on amps. And supplies too.
Hook a resistor, 8 ohm, 4 ohm, whatever, from one channel of a stereo amp to the other. Drive one channel with a sine wave, and measure the voltage at the other output terminal. Make a graph varying the frequency and drive voltage. Try it sometime......
Hook up a square wave generator to the output terminal. Drive it with various frequencies and levels. Look at how it behaves. Lots of ringing and stuff, and you have problems. Works on power supplies too. I believe Bob Pease suggests this also.
3. Not sure I understand your terminology. Please clarify the situation.
You always need to bandwidth limit the input signal. Just because you can build an amp flat to 2 MHz, doesn't mean that you need to amplify that crap. And if it isn't......well here comes the IM products. I believe HiFi News ran an article......late 80's I think....showing the effects of RF generated IM in a line stage. Forgot the details......that was a long time ago.......but the point was to keep out the stuff that doesn't belong there. Which is another reason I fell ICs sound really bad in general. They are much more susceptible to this sort of problem. (Someone who is routinely ripped apart here uses transformers for just this reason.)
In any case, good grounding is essential. For lots of reasons. Balanced inputs not only get rid of ground-loops, but help to remove stuff that you don't want. (Balanced inputs have other advantages, like getting rid of modualtion on power supply rails, and lowering gain requirements.)
Is this any help?
Jocko.
Re: 3 "biggies"
Hello Jocko,
thank you for your hints! 🙂
Great!
Could you elaborate on this in detail and with examples? I got lost there, what do yo mean by "anything you do on the audio signal" ?
Second that from the vacuum tube world, had exactly the same tried out, exactly the smae experience. 140kHz was the upper BW limit we mostly settled on.
Fully agreed. One of my most important measurings is done with a stopwatch. When i play a new thing to buddies, i hiddenly measure (mischievous me 🙂 ) how long it takes until the first and the last starts to show escape body language. Goes from a few seconds to several hours. Treatment can be spiced with J.S.Bach violin sonatas or free jazz or 20th century music (believe me, i know 1st class 20th cent. performances), if then still noone tries to escape, then it is a real winner 🙂
yeah, and one reason for tube amps is that you have a chance to avoid elctrolytics and use oily foil instead 🙂 .... fully agreed!
Thanxalot! goes as well into my repertoire as Manfred's differential headphone amp and wil be combined with it (described on page 2 of this thread). All those stunts force the amp to show its dynamic stability in the real world of wide band mix of frequencies with constantly varying amplitudes (this mix being commonly known as "music").
Do you know more of those amplifier torturing methods? Please tell us!
shouting again from the tube world 🙂 dunno whom you mean, but i mean iron man Thomas Mayer's (Vinylsavor) incredible tube amp setup with 9 (in words: nine) signal transformers in the direct signal path and having one of the highest detail resolution and openness i ever had the chance to listen to.
Again, have to fully agree with what you say about proper grounding.
Hello Jocko,
thank you for your hints! 🙂
Great!
Could you elaborate on this in detail and with examples? I got lost there, what do yo mean by "anything you do on the audio signal" ?
Second that from the vacuum tube world, had exactly the same tried out, exactly the smae experience. 140kHz was the upper BW limit we mostly settled on.
Fully agreed. One of my most important measurings is done with a stopwatch. When i play a new thing to buddies, i hiddenly measure (mischievous me 🙂 ) how long it takes until the first and the last starts to show escape body language. Goes from a few seconds to several hours. Treatment can be spiced with J.S.Bach violin sonatas or free jazz or 20th century music (believe me, i know 1st class 20th cent. performances), if then still noone tries to escape, then it is a real winner 🙂
yeah, and one reason for tube amps is that you have a chance to avoid elctrolytics and use oily foil instead 🙂 .... fully agreed!
Thanxalot! goes as well into my repertoire as Manfred's differential headphone amp and wil be combined with it (described on page 2 of this thread). All those stunts force the amp to show its dynamic stability in the real world of wide band mix of frequencies with constantly varying amplitudes (this mix being commonly known as "music").
Do you know more of those amplifier torturing methods? Please tell us!
shouting again from the tube world 🙂 dunno whom you mean, but i mean iron man Thomas Mayer's (Vinylsavor) incredible tube amp setup with 9 (in words: nine) signal transformers in the direct signal path and having one of the highest detail resolution and openness i ever had the chance to listen to.
Again, have to fully agree with what you say about proper grounding.
OK........forgot you were a tube type.
Let's say you have an amplifier with a unity gain frequency of 2 MHz. At 20 kHz, you would have 40 dB of gain.
An amp with low open-loop gain might have its pole at say.......20 kHz. If you needed 26 dB of gain, you would have 14 dB of loop gain, which is the amount of feedback it would have. That 14 dB would be applied equally across the audio band.
The lots-of-feedback crowd hates this idea, but what happens in their world:
Take the same amp, except that its open-loop pole is at 20 Hz. You still have 40 dB at 20 kHZ, but 100 dB at 20 Hz. Then you would have 74 dB of feedback at 20 Hz, and the same old 14 dB at 20 Khz.
The T(ransient)I(nter)M(odualtion) argument goes something along the line that since you have a lot of feedback at some frequencies, and very litttle at others, then there are differing IM products generated with different (transient) signals. And since music is a bunch of sine waves arranged in some mish-mash, then you will get a mish-mash of IM products. The TIM crowd feels that this is one of the root causes of "solid-state sound". (My terminology)
One reason I prefer open-loop amps is that you have to make each stage linear all unto itself. You affect all parts of the audio band the same way, not with tons of feedback only at LF to just get earthquake bass and all the grit that seems to accompany it.
Easy way to make a low open-loop gain amp:
Take a resistor, say 200 K, and put it from the collector to the base of the gain stage of your typical SS amp. This is called Miller effect, and may be one of the few things I learned in "bootcamp". (The name, not the concept.........already figured that out, but admittedly I didn't know why it worked. Some may say I still don't.......[joke]. ) It effectively lowers the gain of the stage it is applied to, much the same way as if it were a cascode. Since you effectively lower the load resistor, you get lower gain, and higher bandwidth.
If you like it, try 150 K. And then 100 k. (You get the idea.)
Off to get my ears calibrated. Hope they play some decent Baroque music, and not any of that 20th century crap [joke].
Jocko
Let's say you have an amplifier with a unity gain frequency of 2 MHz. At 20 kHz, you would have 40 dB of gain.
An amp with low open-loop gain might have its pole at say.......20 kHz. If you needed 26 dB of gain, you would have 14 dB of loop gain, which is the amount of feedback it would have. That 14 dB would be applied equally across the audio band.
The lots-of-feedback crowd hates this idea, but what happens in their world:
Take the same amp, except that its open-loop pole is at 20 Hz. You still have 40 dB at 20 kHZ, but 100 dB at 20 Hz. Then you would have 74 dB of feedback at 20 Hz, and the same old 14 dB at 20 Khz.
The T(ransient)I(nter)M(odualtion) argument goes something along the line that since you have a lot of feedback at some frequencies, and very litttle at others, then there are differing IM products generated with different (transient) signals. And since music is a bunch of sine waves arranged in some mish-mash, then you will get a mish-mash of IM products. The TIM crowd feels that this is one of the root causes of "solid-state sound". (My terminology)
One reason I prefer open-loop amps is that you have to make each stage linear all unto itself. You affect all parts of the audio band the same way, not with tons of feedback only at LF to just get earthquake bass and all the grit that seems to accompany it.
Easy way to make a low open-loop gain amp:
Take a resistor, say 200 K, and put it from the collector to the base of the gain stage of your typical SS amp. This is called Miller effect, and may be one of the few things I learned in "bootcamp". (The name, not the concept.........already figured that out, but admittedly I didn't know why it worked. Some may say I still don't.......[joke]. ) It effectively lowers the gain of the stage it is applied to, much the same way as if it were a cascode. Since you effectively lower the load resistor, you get lower gain, and higher bandwidth.
If you like it, try 150 K. And then 100 k. (You get the idea.)
Off to get my ears calibrated. Hope they play some decent Baroque music, and not any of that 20th century crap [joke].
Jocko
Re: 3 "biggies"
Jocko is right about A.
You can modulate you currentgain by modulating Vce - Is this nice? I do not think so.
It could be the reason why parts like 2SC2240 sound better than 2N**** it has a ~ 3 times better linearity than 2N3904 but 2N3906 has a better linearity than 2SA970 at this point.
I have found devices with even better linearity than 2SC2240 - Again up 3 times... They are out there. You just have to search for them.
B.) Ft - Is a combination of base resistance and parasitic caps.
C.) Cob has to be low as possible. It is also modulated by the Vcb... Again!? - nice feature? .. NOT!!!!!!!! 😉
I think i will upload a page where i have shown Sims on different transistors of their linearity.
Sonny
Jocko Homo said:
Jocko is right about A.
You can modulate you currentgain by modulating Vce - Is this nice? I do not think so.
It could be the reason why parts like 2SC2240 sound better than 2N**** it has a ~ 3 times better linearity than 2N3904 but 2N3906 has a better linearity than 2SA970 at this point.
I have found devices with even better linearity than 2SC2240 - Again up 3 times... They are out there. You just have to search for them.
B.) Ft - Is a combination of base resistance and parasitic caps.
C.) Cob has to be low as possible. It is also modulated by the Vcb... Again!? - nice feature? .. NOT!!!!!!!! 😉
I think i will upload a page where i have shown Sims on different transistors of their linearity.
Sonny
Look at the linearity with constant base current.
At first a 2N3904 with a "va" of 100V:
Then a 2SC2240 with a "va" of 440V:
And then a other one with a "va" around 1000V:
I have tried to match Ic as good as i could.
As you can see 2SC2240 has a much lower currentgain modulation than 2N3904. But you can get BJT with even higher "va" in the area of 1000V. You just have to find them.
Sonny
At first a 2N3904 with a "va" of 100V:
An externally hosted image should be here but it was not working when we last tested it.
Then a 2SC2240 with a "va" of 440V:
And then a other one with a "va" around 1000V:
I have tried to match Ic as good as i could.
As you can see 2SC2240 has a much lower currentgain modulation than 2N3904. But you can get BJT with even higher "va" in the area of 1000V. You just have to find them.
Sonny
mmmh .. it did not work! Here are te links.
2N3904 : www.geocities.com/sonnya5000/2n3904ib10u.jpg
2sc2240 : www.geocities.com/sonnya5000/2sc2240ib10u.jpg
the bjt with a va of 1000V : www.geocities.com/sonnya5000/nonameib20u.jpg
Sonny
2N3904 : www.geocities.com/sonnya5000/2n3904ib10u.jpg
2sc2240 : www.geocities.com/sonnya5000/2sc2240ib10u.jpg
the bjt with a va of 1000V : www.geocities.com/sonnya5000/nonameib20u.jpg
Sonny
If you can't see the pictures go to my site and click on the first link called "Vce voltage modulated currentgain."
My page has the following address :
http://www.geocities.com/sonnya5000
It was not ment as any advertising at all!!
I am sorry for all this trouble.
Sonny
My page has the following address :
http://www.geocities.com/sonnya5000
It was not ment as any advertising at all!!
I am sorry for all this trouble.
Sonny
Hfe and linearity
High beta parts usually have less linearity than their lower beta counterparts. At least in terms of the slope, as in your measurements.
But you also have to look at the spread of the Ib curves. A good transistor will have them evenly spaced, whereas parts that might not have as much slope may tend to compress more as Ib is raised.
Two advantages of buying a part such as the 2SC2240:
1.) It is only made by one vendor. The 2N**** types are made by multiple vendors, and they don't all do things the same way. (Another famous audio designer buddy claims that the good old '5534 op-amp sounds good, but only if you buy the original vendor's version. His claim, not mine. But you get the point.)
2.) The part comes in different beta groups. You can chose to buy a high beta version, which will have more tilt in the Ic curve, or a lower beta version that is flatter.
Then you have to decide if you want to linearize the circuit with local feedback since you have lots of gain to play with, or just start with a more linear, but lower gain part to begin with.
Both approaches can be made to work. (I've done both.......!)
But only if you have a decent part to start with.
Which was the original point of all this.
Jocko.
High beta parts usually have less linearity than their lower beta counterparts. At least in terms of the slope, as in your measurements.
But you also have to look at the spread of the Ib curves. A good transistor will have them evenly spaced, whereas parts that might not have as much slope may tend to compress more as Ib is raised.
Two advantages of buying a part such as the 2SC2240:
1.) It is only made by one vendor. The 2N**** types are made by multiple vendors, and they don't all do things the same way. (Another famous audio designer buddy claims that the good old '5534 op-amp sounds good, but only if you buy the original vendor's version. His claim, not mine. But you get the point.)
2.) The part comes in different beta groups. You can chose to buy a high beta version, which will have more tilt in the Ic curve, or a lower beta version that is flatter.
Then you have to decide if you want to linearize the circuit with local feedback since you have lots of gain to play with, or just start with a more linear, but lower gain part to begin with.
Both approaches can be made to work. (I've done both.......!)
But only if you have a decent part to start with.
Which was the original point of all this.
Jocko.
- Status
- Not open for further replies.