Really. Human bodies have all the same topology, but anyway doctors use lots of different diagnostic methods to check if the person is in good health or not. Why do you expect the single set of criteria to check which amp will sound better in all conditions?
Edit:
Speaking of health... The whole week I had been weak... Influenza. What kind of measurements would you propose in this case? The answer is simple: specific measurements, according to symptoms. The same with amps: symptoms can be very different, and no measurements can be used to tell which one is better. Especially, when you hear some symptoms, but have no experience to tell what can cause them. Like in my case: absence of appetite can be caused by common gastritis, while I know it had been caused by virus. Experienced doctor can tell. Experienced designer can tell. But you can't choose the single set of measurements to judge doctors' and desighers' results. Except subjective measurements: feeling of health, feeling of reality of sounds.
Last edited:
Perhaps that reducing the slew rate , if it s high , would be worth it,
by increasing the relevant cap from 39pF to 47/68pF.
This would increase THD and IMD slightly but would provide robust stability,
and thus would prevent any risk of oscillation.
In an experiementation , better to be cautious....
I highly recommend 2 oz of 150 overproof rum with a full lime squeeze, and a spoon of honey, warmed , take it twice a day, the virus or you will be dead in 2 days, usually the virus ....
Thanks Wayne. However, I don't use alcohol, but lemon and honey use regularly the whole week. Can you recommend this cure for all illnesses, including allergy to honey?
The same about amps. They are more different than our bodies. Try to give 2 oz of 150 overproof rum to chicken and see what happens...
The same about amps. They are more different than our bodies. Try to give 2 oz of 150 overproof rum to chicken and see what happens...
pooge-ing the Hafler may be a fine pursuit but I don't see how it is useful to the thread topic - particularly if you literally rebuild the existing amp there is far too much room for skepticism over "results" reported from you/your wife's memory of the baseline performance
given the poor quality of the Hafler pcb layout and number of proposed changes you may as well build a new amp entirely so you can compare side-by-side
given the poor quality of the Hafler pcb layout and number of proposed changes you may as well build a new amp entirely so you can compare side-by-side
The problem with measurements is that they have polarized people to the extreme.
We cannot measure everything that is relevant, but not everything we can measure is relevant.
To wit: say you have an amp with a THD spec of say 0.001%. Say you have another amp with the same spec of of say 0.05%, or fifty times more (sic!).
On the face of it, one would be tempted to say the first one will sound better. I would venture a guess that the second has a better chance of sounding better.
I would do so based on my own experience of designing and building audio gear. I don't think, I *KNOW* what are the reasonable limits of distortion you can have from a decently designed piece of gear without using tremendous amounts of negative feedback. To be specifi, for me, "low feedback" assumes no more than 20 dB of NFB, "medium" raises this to say 28 dB, and"high" is above that. Obviously a very personal qualification.
Over the years, I have consistently discovered and rediscovered that the best overall performance, at least for mw, is achieved by the "medium" class, followed by the "low" class. Some peopkle are willing to sacrfice stability for low NFB figures, while I am not.
THIS is the additional piece of data we all need to be able to judge those THD figures properly. Tell me 0.05% THD with 26 dB of NFB versus 0.001% THD with say 40 dB of NFB, and my money will be on the first one.
So, the problem is not so much a matter of measurements as such, but of their completness, of what has been published and what has not.
Measurements are an indespensible tool while desigining and prototyping because they are the inital set of parameters which tell you how you did the job in rough terms, and perhaps push you to improve on it. Once you have many years of measuring under your belt, believe it or not, but based on a set of well taken measurements you can actually have a rough idea of what to expect from an amp. Not precise, true, and exceptions are always possible, but in general, it is so.
When you see the sweep signal on your spectrum analyzer, and see that the decay of the even and odd harmonics is as it should be, you know that amp is likely to sound clean and clear.
When you measure your open loop full power response and get something like 60 or 80 kHz, you know that amp will not sound congested, lazy and slow. Of course, getting such figures is the trick.
If your open loop full power measurements show that your harmonic distortion is 0.3% 20...20,000 Hz, then you don't NEED much feedback in general.
But, this is designer/engineer stuff and means little to the typical customer. In that sense, to him, practically all measurements are meaningless simply because he doesn't know how to interpret them properly. Nor is that his job. He is paying for somebody else to take care of such matters.
Having no better choice, the audio industry tries to convince us that theirs is a great product because it has practically nonexistent distortion. They carefully opmit telling us HOW they achieved those figures.
This is why specifications cannot faithfully describe gear: 1) we don't know HOW they did it, by super careful design, or by adding tons of feedback, and 2) the average customer has no idea of how to interpet those fugures, all he knows is that they should be as low as possible.
By the same token, calling measurements irrelevant is overdoing it just the same as calling them everything, just in the opposite direction.
@Wavebourn
For a natural riddance of flue and influenca, try heating up a glass of red wine as high as you can to be able to swallow it, then add a soup spoon of natural honey. Sip it at once, make sure it rolls down your throat constantly but slowly. Drink it in the evening, just before going to bed. Be aware you will sweat a LOT that night, but in the process, you will sweat out the culprit as well. Look mama, no aspirin! Worked for my ancvestors and me every time.
We cannot measure everything that is relevant, but not everything we can measure is relevant.
To wit: say you have an amp with a THD spec of say 0.001%. Say you have another amp with the same spec of of say 0.05%, or fifty times more (sic!).
On the face of it, one would be tempted to say the first one will sound better. I would venture a guess that the second has a better chance of sounding better.
I would do so based on my own experience of designing and building audio gear. I don't think, I *KNOW* what are the reasonable limits of distortion you can have from a decently designed piece of gear without using tremendous amounts of negative feedback. To be specifi, for me, "low feedback" assumes no more than 20 dB of NFB, "medium" raises this to say 28 dB, and"high" is above that. Obviously a very personal qualification.
Over the years, I have consistently discovered and rediscovered that the best overall performance, at least for mw, is achieved by the "medium" class, followed by the "low" class. Some peopkle are willing to sacrfice stability for low NFB figures, while I am not.
THIS is the additional piece of data we all need to be able to judge those THD figures properly. Tell me 0.05% THD with 26 dB of NFB versus 0.001% THD with say 40 dB of NFB, and my money will be on the first one.
So, the problem is not so much a matter of measurements as such, but of their completness, of what has been published and what has not.
Measurements are an indespensible tool while desigining and prototyping because they are the inital set of parameters which tell you how you did the job in rough terms, and perhaps push you to improve on it. Once you have many years of measuring under your belt, believe it or not, but based on a set of well taken measurements you can actually have a rough idea of what to expect from an amp. Not precise, true, and exceptions are always possible, but in general, it is so.
When you see the sweep signal on your spectrum analyzer, and see that the decay of the even and odd harmonics is as it should be, you know that amp is likely to sound clean and clear.
When you measure your open loop full power response and get something like 60 or 80 kHz, you know that amp will not sound congested, lazy and slow. Of course, getting such figures is the trick.
If your open loop full power measurements show that your harmonic distortion is 0.3% 20...20,000 Hz, then you don't NEED much feedback in general.
But, this is designer/engineer stuff and means little to the typical customer. In that sense, to him, practically all measurements are meaningless simply because he doesn't know how to interpret them properly. Nor is that his job. He is paying for somebody else to take care of such matters.
Having no better choice, the audio industry tries to convince us that theirs is a great product because it has practically nonexistent distortion. They carefully opmit telling us HOW they achieved those figures.
This is why specifications cannot faithfully describe gear: 1) we don't know HOW they did it, by super careful design, or by adding tons of feedback, and 2) the average customer has no idea of how to interpet those fugures, all he knows is that they should be as low as possible.
By the same token, calling measurements irrelevant is overdoing it just the same as calling them everything, just in the opposite direction.
@Wavebourn
For a natural riddance of flue and influenca, try heating up a glass of red wine as high as you can to be able to swallow it, then add a soup spoon of natural honey. Sip it at once, make sure it rolls down your throat constantly but slowly. Drink it in the evening, just before going to bed. Be aware you will sweat a LOT that night, but in the process, you will sweat out the culprit as well. Look mama, no aspirin! Worked for my ancvestors and me every time.
Name a high feedback back amplifier considered to be good sounding , contrary name a low or zero feedback amp considered bad sounding . The Audio Note UK amp discussed previously , failed every test known to man , yet it got a pass for sound ...
PS: Please quantify high feedback and low feedback ..........
PS: Please quantify high feedback and low feedback ..........
PS: Please quantify high feedback and low feedback ..........
I weight effects that are results of feedback. Some are desirable, some are not. As in dvv example, compensation required for stability can cause more harm than decrease THD below already small enough number on 100 Hz frequency where speaker adds own 30% of THD.
Thank you dvv for such good analysis!
Also, thank you for the recipe. But I am afraid for me a glass of vine would mean a death, like 20 dB of feedback to the conditionally stable amp that will inevitably oscillate on frequency lethal for output transistors.
Ohhh, Ok , so you are not full Ruskian ......
OK, but quantify , what is considered high feedback 10 db, 20 db , 30 db, 120 db ..?
Anyone , but you're as good as any , so we will start with you ......
Anyone , but you're as good as any , so we will start with you ......
I'm afraid that my commercial amp experience is severely limited, I build, not buy. And I'm interested in amps that merely take an electrical signal and make it bigger, not the ever-popular effects boxes. For power amps, John Curl's Parasound seemed to work well when I was over at his apartment. I have an old Adcom (first series 555) which I use from time to time that seems to not tear my ears off. For preamps, the Blowtorch is sort of legendary, sounded fine when I heard it (briefly), nothing obviously wrong.
My phono and line preamps use cathode followers, call that 30-35dB of feedback. My mike preamps use high feedback circuits, probably 50dB or more. Not sure what's inside my A/D and D/A cards, but presumably very high.
But for better woo-woo cred than some dumb guy in Austin who would rather have a musician playing in his living room than run his stereo, look at the Class A and B rankings in Stereophile over the decades- lots of high feedback amps in there.
It is all from where we started. My wife's hearing is completely repeatable. The question is why some amps she tolerates, others not. That led to what is different, are we measuring it, are we interpreting it? What could one change to effect it? If identified, can I measure it with my simple tools or is it above my measurement capabilities? Real numbers objectively measured and repeatable are the end goals. The numbers published in slick sheets do not provide a consumer of the indication of performance. It is not as easy as THD, DF, BW. I agree completely, "My wife likes it" may be important to me, but not to anyone who designs amps. WHY is important.
This got me fascinated with amplifier design and the different approaches, biases, and outcomes. Pooging an old Hafler is an exercise to see if we do know what is causing the difference. What can be made better or worse. I am taking best advantage I can to understand amplifiers. It is not just do this to fix it, I am trying to understand why and be able to measure each change. If it was just about a better amp, I would just make another trip to e-bay.
So we have identified grossly asymmetric slew from at least two sources, heavy loading of the VAS, and limited overall slew rate. If these are addressed incrementally, does one need to even switch 30 years in transistor technology, or is the topology still the limiting factor? Views seem to vary. Was the answer this was a lousy amp with higher than published distortion and that was all there is to it? Is this high order harmonics bothering metal dome tweeter breakup ranges causing IM? (good possibility) Just plain second order distortion from not matching the inputs? The Rotel 840 is a very basic topology. What is it not doing that makes her prefer it? Can I make it to it and sound bad?
Back to study. Found a good basic primer on our Safari books, but it is still too expensive to buy off Amazon, so I have to work my way through it on lunch break. Found an article on "Duncan's WEB pages" on loads to simulate speakers. Time to walk through the Simetrix tutorial.
Hi a.wayne,
Actually, the people who confused the entire measurements thing were the sales departments, not the engineers. If you have an inferior product, what do you do? The only thing you can. Misdirect and all-out lie. As was observed earlier, point out the biggest weakness and present it as a positive feature.
Example: Ford has active sway control on their current trucks, no one else does. Well, that's because everyone else builds real trucks and Ford makes mush buckets. Every watch the rear end of a late model Ford pickup (150s are the worst) going down the road? Exactly. No one else had a problem to begin with.
A midst all this confusion, customers haven't a clue what to believe, and I think they gave up on it. Witness the complete lack of excitement over new audio products compared to what we saw in the 70's and 80's. It is important to note that for the vast bulk of products in the market back then, they are considered to be higher quality than what is available today (the high end market excepted of course). Do you remember what was going on back then? The numbers war were power went up and distortion down. Not surprisingly, some of the better basic circuit configurations (as opposed to conflagrations seen in new products) were designed in the 70's. The diamond structure, John Curl's input stage, better complimentary output designs, triple EF output stages, Blomely (sp?), J-Fets in diff stages, high feedback due to better devices, capacitor defects quantified, resistor defects quantified and even bias circuitry and Thiel-Small design methods for loudspeakers. Some of this stuff probably developed in the 60s, but they saw common use throughout the 1970's. We were searching for what types of measurements would correlate with perceived sound quality. We found some, but not all that mattered. Most tests and readings are tools that explain just what an amplifier design is doing. We are still missing some. But you can't throw out all measurements just because we need a couple extra tools in order to describe audio performance completely.
You know what is surprising? From a schematic, knowing the parts used and some measurements, you can make an educated guess at how good an amplifier might sound. Of course, the build design also matters. You can look into some audio devices and see the problems. That is a far cry from where the industry was in 1960. It seems that general understanding as devolved over the last 20 years or so. But the one thing that is true is that a measurement doesn't lie. They are often misinterpreted though.
As for making meaningful measurements, it takes both skill with the equipment, and good equipment up to the task. The dude (or dudette) on the bench has to know how to proceed or the results will be as useless as the person measuring the stuff. Proper numbers and screens of response and phase do give a measure that can be compared between products. People sitting in a room can tell you they do or do not like what they hear at that time, but they are not very good at placing equipment on a scale in a repeatable fashion. The test uncertainty is large and undefined. Ever wonder why there is confusion in the market place? That's why.
-Chris
Actually, the people who confused the entire measurements thing were the sales departments, not the engineers. If you have an inferior product, what do you do? The only thing you can. Misdirect and all-out lie. As was observed earlier, point out the biggest weakness and present it as a positive feature.
Example: Ford has active sway control on their current trucks, no one else does. Well, that's because everyone else builds real trucks and Ford makes mush buckets. Every watch the rear end of a late model Ford pickup (150s are the worst) going down the road? Exactly. No one else had a problem to begin with.
A midst all this confusion, customers haven't a clue what to believe, and I think they gave up on it. Witness the complete lack of excitement over new audio products compared to what we saw in the 70's and 80's. It is important to note that for the vast bulk of products in the market back then, they are considered to be higher quality than what is available today (the high end market excepted of course). Do you remember what was going on back then? The numbers war were power went up and distortion down. Not surprisingly, some of the better basic circuit configurations (as opposed to conflagrations seen in new products) were designed in the 70's. The diamond structure, John Curl's input stage, better complimentary output designs, triple EF output stages, Blomely (sp?), J-Fets in diff stages, high feedback due to better devices, capacitor defects quantified, resistor defects quantified and even bias circuitry and Thiel-Small design methods for loudspeakers. Some of this stuff probably developed in the 60s, but they saw common use throughout the 1970's. We were searching for what types of measurements would correlate with perceived sound quality. We found some, but not all that mattered. Most tests and readings are tools that explain just what an amplifier design is doing. We are still missing some. But you can't throw out all measurements just because we need a couple extra tools in order to describe audio performance completely.
You know what is surprising? From a schematic, knowing the parts used and some measurements, you can make an educated guess at how good an amplifier might sound. Of course, the build design also matters. You can look into some audio devices and see the problems. That is a far cry from where the industry was in 1960. It seems that general understanding as devolved over the last 20 years or so. But the one thing that is true is that a measurement doesn't lie. They are often misinterpreted though.
As for making meaningful measurements, it takes both skill with the equipment, and good equipment up to the task. The dude (or dudette) on the bench has to know how to proceed or the results will be as useless as the person measuring the stuff. Proper numbers and screens of response and phase do give a measure that can be compared between products. People sitting in a room can tell you they do or do not like what they hear at that time, but they are not very good at placing equipment on a scale in a repeatable fashion. The test uncertainty is large and undefined. Ever wonder why there is confusion in the market place? That's why.
-Chris
Thanks Sy,
I have used many , many commercial ( including the ones you listed ) and a few custom built amplifiers over the years, I'm sure all if not most would be considered high feedback amps. What is really considered high feedback, is 50 db, high, is 30 db low, what about Class-d with 120 db feedback ( is this correct). As to zero feedback, that's not possible is it ..
PS: I have owned and heard JC stuff, including the HALO brand , never heard the blowtorch Pre-amp, same with NP products, unfortunately none of his most recent stuff...
I have used many , many commercial ( including the ones you listed ) and a few custom built amplifiers over the years, I'm sure all if not most would be considered high feedback amps. What is really considered high feedback, is 50 db, high, is 30 db low, what about Class-d with 120 db feedback ( is this correct). As to zero feedback, that's not possible is it ..
PS: I have owned and heard JC stuff, including the HALO brand , never heard the blowtorch Pre-amp, same with NP products, unfortunately none of his most recent stuff...
The biggest factor by far has got to be psychology - easy to understand if you find that an unchanged system sounds different every time you listen to it. Sometimes it's like the best thing you ever heard, and the next night just flat and uninspiring. If that happens with a system that hasn't changed at all, then how on earth can you discern the differences between capacitors and DACs when you want/expect to hear differences between them? It defies belief. Ah, but you know they're different, not because of meaningless 'engineering' measurements, but because of blind testing (certain people say here quite a lot). In my experience, a single real blind test would take so much organisation that no person could hope to audition all the individual capacitors, cables, DACs, BJTs, MOSFETs, valves, power supplies, feedback topologies they spout on about, in a lifetime. I don't believe that anyone has ever done it. I do believe that people hear differences that don't exist when they have the slightest inkling of what it is they are supposed to be listening to. I also believe that few people here would ever want to find the ultimate amplifier design even if it could be decided upon, because then there'd be nothing quasi-religious to talk about any more.
Thanks Chris,
Agree with most of that , hence why i posted the graphs previously from what was considered the best , ( soulution 700) to avg ( ayre) to what i considered the worst (audio note ) yet the perceived sound quality was high amongst all of them ...
Maybe JA might chime ................
Agree with most of that , hence why i posted the graphs previously from what was considered the best , ( soulution 700) to avg ( ayre) to what i considered the worst (audio note ) yet the perceived sound quality was high amongst all of them ...
Maybe JA might chime ................
Hi tvrgeek,
Your wife, like mine, seems to be able to pick out what sounds good and what doesn't. That is perfectly reasonable and believable. However, all you can say is that she likes this and doesn't like that. This is valid to a degree, and I have found that by measuring the equipment, I do find where things misbehave. Often times though, the difference in measurements are small. You just have to learn to recognize it and maybe figure out a test that amplifies the differences. A spectrum plot often sheds light on things if the bandwidth is wide enough. Sometimes you must subtract one result from another to clearly show a problem. Then you also have to do control experiments to prove what is a change and what is uncertainty with the method.
-Chris
Yes, this is a great way to learn what works and what doesn't.
Does an asymmetric slew rate really matter at all? Presumably, we avoid getting anywhere near the slew rate limit of any signal electronics. I don't think that a difference in slew rate is going to matter in that case. If we are near the slew rate limits, we have other problems as well. Making the slew rates the same is not going to help much.
Your wife, like mine, seems to be able to pick out what sounds good and what doesn't. That is perfectly reasonable and believable. However, all you can say is that she likes this and doesn't like that. This is valid to a degree, and I have found that by measuring the equipment, I do find where things misbehave. Often times though, the difference in measurements are small. You just have to learn to recognize it and maybe figure out a test that amplifies the differences. A spectrum plot often sheds light on things if the bandwidth is wide enough. Sometimes you must subtract one result from another to clearly show a problem. Then you also have to do control experiments to prove what is a change and what is uncertainty with the method.
-Chris
- Status
- Not open for further replies.