i confess ... i have big difficulties in accepting that kind of graph
Because what comes out is not only what comes in amplified ... the device adds "something" along with gain.
However thinking more about it i am quite convinced that measuring an amp on a pure resistive load, like it is normally done at the bench, can be deceiving.
The real loads (i.e. speakers) are very different from a pure resistive load
It would be better maybe to send a test signal in and capture it with a mic
from the speaker that the amp is driving and this nobody does.
I remember an extreme case of one speaker ... the Apogee Scintilla
With some amps (like Krell i think) they were sounding just wonderful
Other amps even caught fire ...
Gino, seems you are not as ignorant as I thought. Speakers have frequency dependent distortion due to non linear losses.
You may want the great SOA measurement microphone but that can always wait for later. No need to go crazy for your first step, measurement of acoustic FR and 0.5 - 1% distortion is not that difficult. You can start on the cheap and decide on the point of precision upgrade you want later. An "el cheapo" electret cell to a sound card is good enough to get a baseline, meaningful comparison and establish some "feel". Next step in precision I guess would be the Dayton iMM-6 measurement microphone, then available choices jump to $100+ range.
I was told good amp designers do measure acoustic output, they simply feel no need to show those.
You may want the great SOA measurement microphone but that can always wait for later. No need to go crazy for your first step, measurement of acoustic FR and 0.5 - 1% distortion is not that difficult. You can start on the cheap and decide on the point of precision upgrade you want later. An "el cheapo" electret cell to a sound card is good enough to get a baseline, meaningful comparison and establish some "feel". Next step in precision I guess would be the Dayton iMM-6 measurement microphone, then available choices jump to $100+ range.
I was told good amp designers do measure acoustic output, they simply feel no need to show those.
What do you expect to see? You can't send long signal because room reflections will interfere. But basically that is how impulse response measurement is done (and to generate frequency response when anechoic chamber is unavailable).
I think you can trust your ears. It's not that far away
Hi well i am trying to understand something. I have to say that i understand more measurements ... those are very obvious. Technicalities ...
But sometimes i read something that i do not understand completely but makes me think. Like an article on the single ended nature of sound by Nelson Pass sponsoring both the no-feedback approach and against push-pull stages. Very interesting.
i think i have understood the point ... the quality of an amp cannot be completely assessed with connecting it to a speaker and check the outcome in terms of sound ? if so good ! better ... i would have much preferred a lab testing maybe on a complex load ... to me objectivity is very comforting subjectivity is a nightmare.
very strange that a test so revelatory is not discussed and disclosed much more
The reason for doing tests that in the end cannot be correlated with the actual sound quality is a mistery to me.
What do you expect to see?
You can't send long signal because room reflections will interfere.
But basically that is how impulse response measurement is done (and to generate frequency response when anechoic chamber is unavailable).
I think you can trust your ears. It's not that far away
Hi and thanks for the helpful advice.
Sometimes i read story that make me thing a little. Like those about blind tests comparing very cheap and very expensive units and with similar resulting sounds.
I will try asking some manufacturer ... i am not very much into speakers actually. I am fascinated by electronics much more.
Compared to numbers from voltage output, tests on acoustic output shows embarrassingly bad numbers, breaks the high fidelity image. The crowd are just so, base their purchase decision on image rather than objective reality. Welcome to the dirty kitchen of audio amp designers.... very strange that a test so revelatory is not discussed and disclosed much more
The reason for doing tests that in the end cannot be correlated with the actual sound quality is a mistery to me.
Compared to numbers from voltage output, tests on acoustic output shows embarrassingly bad numbers, breaks the high fidelity image. The crowd are just so, base their purchase decision on image rather than objective reality. Welcome to the dirty kitchen of audio amp designers.
I don't understand the purpose of such test. I think it's silly. Probably the speaker is kind of speaker that needs typical amp to move them. So it is probably a match between ATC/Dynaudio and Behringer. 'Matching' means that a bad amp can make a bad speaker sound subjectively good.
No mystery. The standard tests are meaningful. They have correlation with sound quality. It is when your ears are very sensitive that you will hear something not covered by previous standard measurements. Sometimes, people make measurements even tho it has no meaning to them. Sometimes, the measurements have indirect meaning (like square wave test).very strange that a test so revelatory is not discussed and disclosed much more
The reason for doing tests that in the end cannot be correlated with the actual sound quality is a mistery to me.
Amps and speakers are designed/produced by separate entities. If you design/produce an amp, which speaker/driver design you will use to produce your acoustic measurement? You must agree with other amp producers on the standard speaker (to make the measurement meaningful to customers). BTW, when you produce active speakers like Bruno, you can make an amp that match with your speakers.Welcome to the dirty kitchen of audio amp designers.
Last edited:
Since I currently am not in the business here is a what if scenario. I am not one to go against the tide, it is a fruitless task. I'd have no hesitation to design a hi-fi illusion amp for the crowd buying on image, just like other amp producers. However, I'd have to custom fit the amp I design to those seriously concerned on sound quality, needs good full range driver or a multi amp system and a means to compensate driver irregularities and room mode the crowd will never swallow anyway. I have not come across a single solution to cure all ailments. At the end of the day, entertainment business engage in entertaining customers to satisfy their fancy.
With all the shortcoming and very limited experience that I have, I am happy to share my findings for all DIYers to see a bit more of the complete picture and arrive at what they truly want.
Hi i do not have an answer but i can notice one thing. I grew up reading with huge interest among other things the british HIFI magazines like HiFi Choice and What HiFi ...
The best buys were my objects of dream ... units that provided excellent performance at a nice price.
I guess that kind of test appeals to some budget minded people ...
i would not classify ATC speakers as bad ... they are commonly used as monitors in recording studios. Actually i am quite sure they are excellent.
Beyond suspicion.
well unfortunately this is not my case. Recently i had my hearing system checked ... i cannot listen above about 14kHz ... i am 58
And time will not help for sure ...
I can see already a comment .. i should not be worried about what i cannot listen anyway
Recently on Youtube i saw a very telling test. A SW sent through the cd input was distorted in normal conditions but much better preserved with the cd direct option engaged
If I want to sell an amp, I can put it side by side with expensive amps plus speaker of my choice such that my amp will put those expensive amps in shame... But it's cheating.
By 'sensitive ears' I didn't mean that it can hear 10Hz to 20kHz. 80-5kHz is fine but the key is hearing what's going on there.
Hi i see your point. Diabolic marketing strategy ...
ok ... you mean an educated ear. Sometimes happens that i hear good sound. I become aware of this because i stop worrying about the equipment and i captured completely by the sound. Very very nice feeling.
And sometimes this happens in the strangest conditions. I remember hearing good sound even from a micro system from Sony with a digital amp ?
As the power available was very limited i was thinking to use that amp as a front end putting a kind of power booster (like the ones once popular in car audio) after it to increase the power in order to be able to drive also more demanding speakers.
But, as usual, i did not developed the idea ...
Last edited:
Quality has many facets. You may hear a good sound when it is a rock music but then it may fall apart when it is classics. It may produce great sound with small speakers but fall apart with three-ways... You may have great vocals but less something else and so on... Technically, the amp may have good control of the speaker but THD+N is high, meaning that you have to choose which parameters most important to you or your ears.I remember hearing good sound from a micro system from Sony with a digital amp ?
What does it tell, BTW? Square wave does not happen with normal music. It shows high frequency and slew rate not needed in audio. In above case, the cd input may have the high frequency been filtered and so on. Does not really tell something related to audio quality I guess.
There is something about adding extra circuit (such as preamp or whatever) when you don't need the extra gain (or even the optimum gain structure). But most of the time it is a liability.
Last edited:
I will keep all these recommendations in mind from now on.
I understand the need to assess the sound quality of a chain as a whole with speakers included.
And in the end what sounds good is good. No doubt about that.
I just wonder if instead of fixing my attention on electronics i should start down the chain with listening and selecting speakers and then move upwards. In the view that speakers are the real limiting factor.
I just wonder if instead of fixing my attention on electronics i should start down the chain with listening and selecting speakers and then move upwards. In the view that speakers are the real limiting factor.
Yes, it's true. That's why it is okay imo to invest a lot of money on speakers. Problem is, expensive drivers mean nothing if the design is bad, and it is not easy to spot a good or bad ones. But once you have your speaker, then you can find amp that suit you or your speakers. If you keep the power supply the same, building amps are not expensive. Choosing speakers means you have to first decide between high sensitivity ones (incl. fullrange and horns) that you will match with flea amp (such as tube), or the 'big' insensitive three way matched with big class AB amp.
...how do you know if it will work/be stable or not?
I drew my first circuits when I was in the prime class trade school, sixteen years old in '79 and puzzling with transistors where my classroom mates were trying to distinguish the anode and cathode of a diode. Since then a lot has passed, but application notes and (good) service manuals from manufacturers are highly recommended to start with. But copying or rebuilding is not designing! Math & Calculating (without machines) are self-evident. Roots, logs, exponentials, sinus-cosinus, imaginary plane; ever heard of a Schmitt-chart? Now that's really horrible. But M&C are usefull for swift estimations before any simulation anyhow.
Designing circuits, and audio applications more specific, is fun in the first place. I'm always challanged to find an electronic solution for a sudden issue. During discription or showing the current state, I'm not thinking how to wrap around things but 'see' some 'connecting points or area's' which fall together in a joining design. The actual drawing is more work, the calculations converging by iteration to a 'proof of concept'. I've added a first idea of a HighLoadBuffer. Aim was simplicity, near zero dc output and capable of driving less then some 500 ohms. Black is original, blue results from first simulations.
So, in short these steps were taken:
Design (strategy, topology, aim, components, rail-definitions), simulations (I use AimSpice since very long; don't make errors!) of OP (operating point, very important first glance), DC (also, and showing 'modulation of settings'), AC (values, currents, dB, phase), and others if neccessary (transient, noise, pole-zero, transferfunction, thermal). Until a most promising candidate is defined.
Next is building a testboard. And more changes, tweakings and disappointments. Measuring OP, DC and thermal, and if ok next AC, noise, pulse and what not.
Back to the design, export to pcb design (I use Orcad because that's the only one I've learned to use) and make it impaccable. EMC is very important to understand here. Gerbers and production of test samples.
Then a proper build, verification measurements and if audio is involved, put it in the audiosystem available. Most often my home set.
What are we listenig to? How are we listening? What is our reference, what are our surroundings? What is our listening experience, the state of our ears?
A recording from what origin and how processed?
With our ears, a 'biological recording system' the size of two suger cubes decoding sound into chemical (!) signals to our processing human brain.
A classical concert, a jazz club, a dance cavern, a car?
At home (the dishwasher is still running), a theatre, on top of the mountain?
This is all so very personal. No equipment yet to replace our ego-perspective.
...the device adds "something" along with gain...
No matter what approach, it is indeed a dirty kitchen (!!!) which produce a device that 'change' the original. In a strict sense, amplification is distortion. And electronics are 'borrowing' energy to add to the signal. That is a imperfect process, alas. You might call this the 'Mysterious Audio Chemical Agent' (MACA).
Mainly there are two 'preferences': precise and pleasing. The first are like laboratry instruments, with high control (feedback) and a lot components. A lot of people experience these as 'hard'. The second are 'comforting', sounding more naturally to the human ear (and the rest beyond), such as low or non feedback and tube circuits. What is better? I don't know your personal preference!
What is my preference and reference? Mainly classical music live performed in various gigs. Ravel, Debussy, Bach, Scriabin to name a few.
But after many years, electrostatic loudspeakers prove (to me, again) to come most close to the original (in my home setting, again) due to a more linear conversion from Volts and Amperes to vibrating air. No filtering and hence phase shifts (also: distortion) in the frequency range from say 200 to 4000 Hz.
Around 250 Hz is the most energy processed, above 2kHz phase is directional.
The energizer (I like most, yes again) for this transducer is a Hiraga 2x8W class A amplifier. The pre is such an accurate TA-E86B with a respectfull mc-pre stage for the Thorens TD160 - SME3 - Empire combination.
Not to promote these, but it is my preference which is able to reproduce my reference most closely. For comfort an array of cd's is also available. I got the player when I worked for Matshusita around '88. Yes, it has an inverted Chebyshev output filter to supress the digital switching. I cannot hear that anymore either.
So, in brief, how do I know if something works or not?
What is your favourite colour?
Shown a single stage unity gain buffer with near zero-dc and capable of driving a 220 ohm load, a candidate for my running design of an intertwined Hiraga-bridge main amplfier with 10A bias current.
I drew my first circuits when I was in the prime class trade school, sixteen years old in '79 and puzzling with transistors where my classroom mates were trying to distinguish the anode and cathode of a diode. Since then a lot has passed, but application notes and (good) service manuals from manufacturers are highly recommended to start with. But copying or rebuilding is not designing! Math & Calculating (without machines) are self-evident. Roots, logs, exponentials, sinus-cosinus, imaginary plane; ever heard of a Schmitt-chart? Now that's really horrible. But M&C are usefull for swift estimations before any simulation anyhow.
Designing circuits, and audio applications more specific, is fun in the first place. I'm always challanged to find an electronic solution for a sudden issue. During discription or showing the current state, I'm not thinking how to wrap around things but 'see' some 'connecting points or area's' which fall together in a joining design. The actual drawing is more work, the calculations converging by iteration to a 'proof of concept'. I've added a first idea of a HighLoadBuffer. Aim was simplicity, near zero dc output and capable of driving less then some 500 ohms. Black is original, blue results from first simulations.
So, in short these steps were taken:
Design (strategy, topology, aim, components, rail-definitions), simulations (I use AimSpice since very long; don't make errors!) of OP (operating point, very important first glance), DC (also, and showing 'modulation of settings'), AC (values, currents, dB, phase), and others if neccessary (transient, noise, pole-zero, transferfunction, thermal). Until a most promising candidate is defined.
Next is building a testboard. And more changes, tweakings and disappointments. Measuring OP, DC and thermal, and if ok next AC, noise, pulse and what not.
Back to the design, export to pcb design (I use Orcad because that's the only one I've learned to use) and make it impaccable. EMC is very important to understand here. Gerbers and production of test samples.
Then a proper build, verification measurements and if audio is involved, put it in the audiosystem available. Most often my home set.
What are we listenig to? How are we listening? What is our reference, what are our surroundings? What is our listening experience, the state of our ears?
A recording from what origin and how processed?
With our ears, a 'biological recording system' the size of two suger cubes decoding sound into chemical (!) signals to our processing human brain.
A classical concert, a jazz club, a dance cavern, a car?
At home (the dishwasher is still running), a theatre, on top of the mountain?
This is all so very personal. No equipment yet to replace our ego-perspective.
...the device adds "something" along with gain...
No matter what approach, it is indeed a dirty kitchen (!!!) which produce a device that 'change' the original. In a strict sense, amplification is distortion. And electronics are 'borrowing' energy to add to the signal. That is a imperfect process, alas. You might call this the 'Mysterious Audio Chemical Agent' (MACA).
Mainly there are two 'preferences': precise and pleasing. The first are like laboratry instruments, with high control (feedback) and a lot components. A lot of people experience these as 'hard'. The second are 'comforting', sounding more naturally to the human ear (and the rest beyond), such as low or non feedback and tube circuits. What is better? I don't know your personal preference!
What is my preference and reference? Mainly classical music live performed in various gigs. Ravel, Debussy, Bach, Scriabin to name a few.
But after many years, electrostatic loudspeakers prove (to me, again) to come most close to the original (in my home setting, again) due to a more linear conversion from Volts and Amperes to vibrating air. No filtering and hence phase shifts (also: distortion) in the frequency range from say 200 to 4000 Hz.
Around 250 Hz is the most energy processed, above 2kHz phase is directional.
The energizer (I like most, yes again) for this transducer is a Hiraga 2x8W class A amplifier. The pre is such an accurate TA-E86B with a respectfull mc-pre stage for the Thorens TD160 - SME3 - Empire combination.
Not to promote these, but it is my preference which is able to reproduce my reference most closely. For comfort an array of cd's is also available. I got the player when I worked for Matshusita around '88. Yes, it has an inverted Chebyshev output filter to supress the digital switching. I cannot hear that anymore either.
So, in brief, how do I know if something works or not?
What is your favourite colour?
Shown a single stage unity gain buffer with near zero-dc and capable of driving a 220 ohm load, a candidate for my running design of an intertwined Hiraga-bridge main amplfier with 10A bias current.
Attachments
Hi ! i do not know if the question is for me ... i can only say that i have already destroyed a little Rotel amp in the attempt to reduce its gain decreasing the feedback resistor value (i wanted to make it a kind of power buffer for an headphone amp with already high voltage gain)
Stability is unknown territory for me
For sure it is not designing from scratch ... but a schematic is not the end of the game and by the way working points can be changed.
The design of lay-out is also important unless you mean copying also that.
Moreover some good schematics of the past were overdesigned ... cannibalize just a line amp or a buffer can be used for line stages for instance.
I am completely not interested in tone controls that were the norm in the past.
Better parts are available these days.
I was reading something lately on vintage units. If a restoration is done with knowledge the restored units end sounding quite better than the original (i.e. less noise and distortion)
For sure circuit design is very challenging. Not for everyone.
I am afraid not for me for instance. When i look at certain schematics i am both admired and lost ...
What about the verification measurements specifically ? could you elaborate on which kind of test do you perform to assess the performance of a prototype ?
i try to give an answer. I want to hear everything of the recording. I try to elaborate.
If we take a recording a lot of information are embedded in it. The most revealing systems expose the most of these information. The other systems are not able to do it and moreover add some information not present in the original recording.
I have little time but i am sure that what sound nice often is not representing the information in the recording. Nice ? for sure ... true ? i do not think so.
I want to hear the noise of the fingers sliding on strings ... the ambient noise ... Truth can hurt ... but i want to hear he truth, the whole truth and nothing but the truth
this is something i cannot easily accept ... if i send inside a signal test of 1 kHz i would like to see only 1kHz at the output ... more or less amplified.
If i see a comb of harmonics i am not comfortable.
I would like to add one feeling. If the source is analog the sound is always acceptable. Can be only decent or very very good but it is always musical.
Not so with digital ... that is a strange beast.
My feeling is that in digital embedded with the music signal there is some kind of annoying noise that is very evident with solid state and much less so with tubes for instance.
And actually digital sound almost always quite ok with tube amps.
Like tubes intercepting and neutralize this noise ...
Moreover digital with less resolution (like 14 bit) can sound more pleasant than digital with high resolution (24 bit)
More analog ...
For me personally it is like when i taste food ... maybe a new recipe.
I feel immediately if i like or not.
I have noticed that when the sound is good i stop worrying about the equipment and just enjoy the music (actually even bad music
)It is a mix of realism, space, tone, dynamics, naturalness, effortless power, ...
To try to understand something i would move upwards ... i would take something that sound good and try to understand why ... with instruments.
I am more interested in solid state ... but intrigued by hybrid solutions a lot. Tube front end and solid state output.
interesting ... why not a diamond buffer instead ?
Thank you very much indeed for all your valuable advice.
The only my certainty is that i know what i like also with music. If i had to mention just one characteristics is an emission of sounds without effort.
The ability to rende a very ample dynamic range ... and soundstage of course.
It was mostly rhetorical.
i can only say that i have already destroyed a little Rotel amp in the attempt to reduce its gain decreasing the feedback resistor value (i wanted to make it a kind of power buffer for an headphone amp with already high voltage gain). Stability is unknown territory for me
But you have learnt something
Still related with stability, if you have sensitive ears, like tweaking but don't understand electronics, here is a tweak you can do:Reduce the compensation cap (e.g. Miller or b-c bypass type, could be different with caps in complex compensation scheme), use styroflex, solder cleanly and shortly. To increase/reduce at small steps I use paralleling with smaller caps. Reducing the compensation cap will have the effect of reducing stability (but default cap is usually oversized to ensure stability and to use common size, such as 100p/82p), so you have to do it slowly while observing the stability issue. Assuming that you don't have stability problem, reducing the cap will increase the sound quality and up until certain point you will find an 'optimum' value (good trade-off between several quality aspects). With luck, this alone can turn a 'bad' sounding amp into a great one.
BTW, allow at least half day before trying to perceive 'difference'.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.