redjr,
I am with you on that one, how in the world can you ever know what a synthesized sound was really supposed to sound like? When you hear an acoustical instrument at least you have a reference to go by. If someone thinks you need a massive sound field at high levels to do this test then just use a symphony orchestra, that should cover the entire range I would think as far as listening for distortions products.
I am with you on that one, how in the world can you ever know what a synthesized sound was really supposed to sound like? When you hear an acoustical instrument at least you have a reference to go by. If someone thinks you need a massive sound field at high levels to do this test then just use a symphony orchestra, that should cover the entire range I would think as far as listening for distortions products.
Hi Guys
Rick, I would have to say the music test samples you cited fall into the "simple" range to me. The all-time simplest music test - and thus the most meaningless - is guitar of any kind. Unless the circuit is defective in a "broken" manner rather than a nonideal manner, then you will hear guitar sound. There is little that any functional amplifier will do to make it not sound like a guitar. I play guitar and build amps yet would never use a guitar recording to test a hifi amp.
Of course, one first tests on the bench. basic DC tests, then sine waves without load, then with a light load, then a full load. If you can do more elaborate tests like THD or IM then those are good to do. First listening tests simply check noise and that signal passes. Then come critical evaluations.
Every article I've ever seen about some new "trick" amplifier circuit rests its case on listening tests with: voice and guitar, voice and piano, solo guitar or piano, small ensemble music of acoustic instruments, if 'pop' music then it is Sara Jones or equivalent - not dense music at all.
Yes, modern music is highly compressed - far too much so - and uses a lot of pre- and post-processing that makes it difficult to assess individual instrument characteristics, so recordings of this genre make poor choices as test music. Instead, there will be music you have listened to for years and are very familiar with. Rock, pop, some blues, movie sound tracks, some classical music - all are "busy" enough in the midrange to either be reproduced well or to be smeared by a poor performing amplifier.
Whatever makes you happy is the ultimate point when assessing equipment. Have we reproduced the original intended music tone? Is that even possible? We have no idea what the sound engineer's intention was other than to sell records and get at least the pay cheque for the gig.
We are attracted to a circuit or product for various aesthetic reasons. Our expectations may be met or not, as the builder of the preamp above posted. Trying to understand the difference between our expectations and our experience is a big part of what life is about.
Have fun
Kevin O'Connor
Rick, I would have to say the music test samples you cited fall into the "simple" range to me. The all-time simplest music test - and thus the most meaningless - is guitar of any kind. Unless the circuit is defective in a "broken" manner rather than a nonideal manner, then you will hear guitar sound. There is little that any functional amplifier will do to make it not sound like a guitar. I play guitar and build amps yet would never use a guitar recording to test a hifi amp.
Of course, one first tests on the bench. basic DC tests, then sine waves without load, then with a light load, then a full load. If you can do more elaborate tests like THD or IM then those are good to do. First listening tests simply check noise and that signal passes. Then come critical evaluations.
Every article I've ever seen about some new "trick" amplifier circuit rests its case on listening tests with: voice and guitar, voice and piano, solo guitar or piano, small ensemble music of acoustic instruments, if 'pop' music then it is Sara Jones or equivalent - not dense music at all.
Yes, modern music is highly compressed - far too much so - and uses a lot of pre- and post-processing that makes it difficult to assess individual instrument characteristics, so recordings of this genre make poor choices as test music. Instead, there will be music you have listened to for years and are very familiar with. Rock, pop, some blues, movie sound tracks, some classical music - all are "busy" enough in the midrange to either be reproduced well or to be smeared by a poor performing amplifier.
Whatever makes you happy is the ultimate point when assessing equipment. Have we reproduced the original intended music tone? Is that even possible? We have no idea what the sound engineer's intention was other than to sell records and get at least the pay cheque for the gig.
We are attracted to a circuit or product for various aesthetic reasons. Our expectations may be met or not, as the builder of the preamp above posted. Trying to understand the difference between our expectations and our experience is a big part of what life is about.
Have fun
Kevin O'Connor
Well-stated, Kevin. THD is a remarkably good evaluation tool (especially THD20), but IM, like 19kHz + 20kHz CCIF is where it is at in my view. I particularly thank John Atkinson for showing 19+20 kHz measurements. This is even more important for class D amplifiers. When we quote THD, we are often implicitly relying on the THD number being related to the IM number. And it is often true that a very low THD number suggests a very low IM number.
Cheers,
Bob
As I've said many times on this forum, I'm not a hard-core hardware nut that seeks to squeeze every ounce of design nuance out of what I build. Most of what I build nowdays is not component level. Mainly because of the time factor involved. I'm old enough to be retired, but my banker tells me otherwise. 
So I stick with a more modular approach, but do my home work with the chip sets, PSU, etc. It's just my opinion, but the quest of the audio holy grail just is not what I'm after. I know that sounds like anathema to most on here, but I just love taking an amp module and designing the overall amp, including selecting the enclosure, and dealing with the mechanical issues associated with bringing it altogether - successfully! Some more than others, but for the most part since I started in this DIY world I've got a pretty good track record.
I did build the DS 2012 incarnation of his famous Precision Pre. It was more of a challenge than it started out as. I guess I can thank my tenacity and attention to detail for sticking with and seeing the project through. Details can be found here.
I never thought it would be the project I'm most proudest of. And the most expensive. It the only 'high-end' piece of gear I own now. Not that I'm complaining. The problem is it now reveals the weakness in all my other mid-fi gear! But despite some of the grumblings about the design earlier in this thread, I've been very happy with the way it sounds and what it adds to my overall listening experience. It really has made my music jump to life so I have no complaints. But then again, when all you have to reference it against is of lower quality, you stand to be pleasantly surprised and amazed.
My quest now is going back and listening to older CDs that have been collecting dust on the rack for years and giving a listen with my newer (and hopefully better) gear to hear if there's a big difference, or more subtle to those old, lost tracks I enjoyed from yester-year. I've been surprised by many of them. Interesting note... along the current topic we've been discussing, some of the CD's that are older have a much better overall sound and dynamics to them. That sounds kinda of counter-intuitive if we believe we must be moving forward and upward with sound quality of commercial recordings. I'm not so sure when I listen to earlier tracks. They do sound less compressed. But maybe it's in part to a little better equipment than I had 20 yrs ago. Not that what I have today is by any stretch high-end, or close to it. Maybe I'm just enjoying the musical ride once again.
Rick
So I stick with a more modular approach, but do my home work with the chip sets, PSU, etc. It's just my opinion, but the quest of the audio holy grail just is not what I'm after. I know that sounds like anathema to most on here, but I just love taking an amp module and designing the overall amp, including selecting the enclosure, and dealing with the mechanical issues associated with bringing it altogether - successfully! Some more than others, but for the most part since I started in this DIY world I've got a pretty good track record. I did build the DS 2012 incarnation of his famous Precision Pre. It was more of a challenge than it started out as. I guess I can thank my tenacity and attention to detail for sticking with and seeing the project through. Details can be found here.
I never thought it would be the project I'm most proudest of. And the most expensive. It the only 'high-end' piece of gear I own now. Not that I'm complaining. The problem is it now reveals the weakness in all my other mid-fi gear! But despite some of the grumblings about the design earlier in this thread, I've been very happy with the way it sounds and what it adds to my overall listening experience. It really has made my music jump to life so I have no complaints. But then again, when all you have to reference it against is of lower quality, you stand to be pleasantly surprised and amazed.
My quest now is going back and listening to older CDs that have been collecting dust on the rack for years and giving a listen with my newer (and hopefully better) gear to hear if there's a big difference, or more subtle to those old, lost tracks I enjoyed from yester-year. I've been surprised by many of them. Interesting note... along the current topic we've been discussing, some of the CD's that are older have a much better overall sound and dynamics to them. That sounds kinda of counter-intuitive if we believe we must be moving forward and upward with sound quality of commercial recordings. I'm not so sure when I listen to earlier tracks. They do sound less compressed. But maybe it's in part to a little better equipment than I had 20 yrs ago. Not that what I have today is by any stretch high-end, or close to it. Maybe I'm just enjoying the musical ride once again.
Rick
There's been plenty of ideas put forward as to why it doesn't sound as good as it should, but I have to say they are all purely speculation. I suspect Mr Self would have checked IMD at some stage, but then what level is acceptable? And if the THD from several opamps adding in bad ways or whatever is to blame, how do we measure this? Check the FFT and look at the harmonics, but what are we looking for? I bet there are all very low and mostly only 2nd and or 3rd.
My favourite power amp, the Leach amp, has unimpressive IMD figures (no doubt as a result of high-ish THD at HF), yet sounds smoother in the top end than many others with better THD figures. Is this higher-fidelity? In my view yes, because eg violins sound more like the real thing without the added screach that you just don't get in a real concert hall. Maybe this is because the amp's distortion is masking some of the artifacts from the digital recording chain, but if the end result is more like how real instruments sound then perhaps this is a good example of why THD is not a worthwhile thing to be designing for in itself? Or rather that listening tests are more important?
Like Abraxalito's signature says
My favourite power amp, the Leach amp, has unimpressive IMD figures (no doubt as a result of high-ish THD at HF), yet sounds smoother in the top end than many others with better THD figures. Is this higher-fidelity? In my view yes, because eg violins sound more like the real thing without the added screach that you just don't get in a real concert hall. Maybe this is because the amp's distortion is masking some of the artifacts from the digital recording chain, but if the end result is more like how real instruments sound then perhaps this is a good example of why THD is not a worthwhile thing to be designing for in itself? Or rather that listening tests are more important?
Like Abraxalito's signature says
Hi Guys
Tests and listening will always be necessary as man-made devices are always imperfect. Were it possible to find or build perfect devices and perfect circuits, some would like the results and some would not. It comes back to our expectations and other personal factors. For everyone to like that perfect amp and perfect speaker, humans themselves would have to be transformed into a "perfect" form as well. Just not going to happen in this quantum universe.
One fact that does support the quest for ever improved specs is that human scale tests conducted in the most objective way possible demonstrated that the majority of listeners prefer flat response systems with low THD. These tests were conducted at the national Research Center in Canada back in the 1970s or '80s. Unfortunately I can't remeber the name of the famous fellow who oversaw it. Floyd Toole? My imperfect brain....
So, we each find our own path and explore the things we want to explore in the fashion that suits use. If we allow individualism to prosper, then we are all correct in our views and choices, and we all have the right to change our choices as we go along.
Have fun
Kevin O'Connor
Tests and listening will always be necessary as man-made devices are always imperfect. Were it possible to find or build perfect devices and perfect circuits, some would like the results and some would not. It comes back to our expectations and other personal factors. For everyone to like that perfect amp and perfect speaker, humans themselves would have to be transformed into a "perfect" form as well. Just not going to happen in this quantum universe.
One fact that does support the quest for ever improved specs is that human scale tests conducted in the most objective way possible demonstrated that the majority of listeners prefer flat response systems with low THD. These tests were conducted at the national Research Center in Canada back in the 1970s or '80s. Unfortunately I can't remeber the name of the famous fellow who oversaw it. Floyd Toole? My imperfect brain....
So, we each find our own path and explore the things we want to explore in the fashion that suits use. If we allow individualism to prosper, then we are all correct in our views and choices, and we all have the right to change our choices as we go along.
Have fun
Kevin O'Connor
Last edited:
As an addendum to my suggestion to try LM6172s in this design, its recently come to my attention that another opamp might well be a better fit. That's the LT1355 because its input bias currents are even lower typically than a 5532's. Its about 6dB noisier though. I've not listened to it, but judging by its topology (internally a CFB just like the 6172) I reckon it should sound better than the 5532 here.
Actually, CDs of this sort of material is exactly what I use now to stress test whether a system is truly in the ball park; because of all the processing, the compression, the complexity of the mixes it takes a truly clean playback setup to unravel it all, to allow one to separate all the sound threads and effects in a "transparent" soundstage.
Part of the value of them as test material is that the power amplifier is working hard, mightly hard, to supply a decently clean signal to the speaker - the power supplies have the accelerator pedal, so to speak, pushed close to the floor most of the time if the volume is running at solid SPLs -- and any deficiencies here are blindingly obvious. Plus, the savage current draws on the mains by the power amp keeping up, injects a very nice, concurrent round of interference into the other components.
No wonder many systems sound a disasterous mess when asked to play this material at solid volumes ...
I might try those LT's at some stage, thanks for the suggestion. But after reading the update in Linear Audio v5 I felt compelled to tweak the '96 preamp a little with the existing (NJM) 5532s and see if it improved things any.
I figured I'm stuck with the 10k pots for now and the basic topology has to stay the same. So I decided to:
1) Add an input LPF as per Small Signal Audio Design - this was achieved by replacing a wire link after the tape/source switch with a 100R resistor and adding a 100pF polyprop cap to gnd after it on each channel.
2) Adding a 10pF NPO cap across the 4k7 feedback resistor around the inverting tone stage opamp. This appears in the new design (presumably to improve stability with the LM4562) but it looks like a good idea so I thought I'd add it anyway. Figured it can't hurt and might do some good.
3) Replaced the existing 100pF ceramic (a high quality NPO) across the active gain stage output opamp's 4k7 feedback resistor with a polypropylene type. This was mainly out of curiosity as you hear many audiophiles saying that ceramics sound bad full stop, and even though I didn't really believe this could be true as long as the ceramic is a good quality NPO, I thought I'd see if it made any difference.
Well there was no measureable change in terms of looking at the waveforms with a scope, but it definitely does sound slightly better now. The slight edginess has been smoothed out and the top end seems slightly cleaner. Overall the sound is still somewhat "over controlled/compressed" for want of a better term and the soundstage remains reduced in size, but it is more listenable and I found myself enjoying the music through it on several albums without feeling compelled to flick the switch over to one of the other preamps I'm comparing it with. Ideally I would have made each change individually and listened in between but it's a bit of a pain to get to the board so I did all three at once.
No doubt some will dismiss these findings as delusional, psycho-expectational drivel, and for that reason I'm not going to speculate which change made the difference, but I'd thought I'd share my findings anyway
I figured I'm stuck with the 10k pots for now and the basic topology has to stay the same. So I decided to:
1) Add an input LPF as per Small Signal Audio Design - this was achieved by replacing a wire link after the tape/source switch with a 100R resistor and adding a 100pF polyprop cap to gnd after it on each channel.
2) Adding a 10pF NPO cap across the 4k7 feedback resistor around the inverting tone stage opamp. This appears in the new design (presumably to improve stability with the LM4562) but it looks like a good idea so I thought I'd add it anyway. Figured it can't hurt and might do some good.
3) Replaced the existing 100pF ceramic (a high quality NPO) across the active gain stage output opamp's 4k7 feedback resistor with a polypropylene type. This was mainly out of curiosity as you hear many audiophiles saying that ceramics sound bad full stop, and even though I didn't really believe this could be true as long as the ceramic is a good quality NPO, I thought I'd see if it made any difference.
Well there was no measureable change in terms of looking at the waveforms with a scope, but it definitely does sound slightly better now. The slight edginess has been smoothed out and the top end seems slightly cleaner. Overall the sound is still somewhat "over controlled/compressed" for want of a better term and the soundstage remains reduced in size, but it is more listenable and I found myself enjoying the music through it on several albums without feeling compelled to flick the switch over to one of the other preamps I'm comparing it with. Ideally I would have made each change individually and listened in between but it's a bit of a pain to get to the board so I did all three at once.
No doubt some will dismiss these findings as delusional, psycho-expectational drivel, and for that reason I'm not going to speculate which change made the difference, but I'd thought I'd share my findings anyway
Yes, but wouldn't you like to know?....I'm not going to speculate which change made the difference, but I'd thought I'd share my findings anyway
The LPF will have some attenuation effect on interference getting into the chips.
Reductions in the levels of VHF will reduce the the IMD. Some of that IMD is heard in the audio band. That reduction in IMD will not show in the scope of the audio signal nor show in the Vac noise measured at the signal.
Reductions in the levels of VHF will reduce the the IMD. Some of that IMD is heard in the audio band. That reduction in IMD will not show in the scope of the audio signal nor show in the Vac noise measured at the signal.
No doubt some will dismiss these findings as delusional, psycho-expectational drivel, and for that reason I'm not going to speculate which change made the difference, but I'd thought I'd share my findings anyway
In general, your perceptions about SQ are remarkably close to my, so I am inclined to think that there must be some truth to it. And since we never met there is no way of influencing each other in any way.
Recordings that Roy DuNunn did for Contemporary Records in the late 50's are considered by many, including me, as non plus ultra! Think for the moment, several condenser microphones from Germany, a few pots, direct to a valve tape recorder, and post production consisting of German built EMT echo plate, and that's it. That's as far away from linear as one can imagine and yet the sound of it is unsurpassed. Without any intention to make anything unsurpassed he achieved it. Just an improvisation.
The Search for Roy DuNann | Stereophile.com
The Search for Roy DuNann | Stereophile.com
Nice theory, sounds plausible. Any suggestions on how to test for this? A standard IMD two-tone test would not show it up presumably. With an f3 for this filter of about 16MHz, you wouldn't think it would be helping much unless I happened to live under a TV or FM broadcast tower (which I don't, I'm at least 20km from the nearest). I'm tempted to increase the cap value to say 1nF, since I won't ever be feeding the preamp from high Z source.
Nice to hear this, thanks. Have you also built the Precision Preamp, or are you speaking more generally?
I believe you're right in that a standard two-tone test won't show it. That's a high level test with large amplitude signals whereas the effects here are low-level. I reckon what you need as a stimulus is a high crest-factor multitone signal where no single tone supplies more than -80dB of the total energy. If such a broadband signal were created with spectral gaps in it, the amount of gap-filling probably will give some indication of the IMD.
No, I have not built it because I do not need "complete" preamp, but your comments about sound of opamp types are very accurate and consistent with my experiences. Therefore I think that your other comments are relevant too.
BTW, it seems interesting to me to build simplified (and less expensive) version of the new Precision, the one that Ian Finch hopes to see one day from Signal Transfer. I do not need balanced inputs, etc...
This is very important detail. I've noticed that whenever some specific form of distortion is highly reduced, some other form jumps out to replace it. So, when you lower THD distortion in preamp, some other distortion in source, or in power amp, or in loudspeaker jumps out to dominate. The best sounding equipment is the one that have a certain distortion composition in some lucky ratio, distributed in some specific way. It is my theory that some people prefer j-fet opamps because of higher noise that masks certain forms of distortion that is particularly annoying for them.
But if you ask me to say what is the lucky ratio, how much or how little of certain specific distortion gives subjectively satisfying results, I am unable to answer.
Hi ivanlukic,
I prefer JFET op amps in general, but especially at the front-end of equipment. I doubt it is because of noise masking of distortion, since they are quite adequately low in noise. Moreover, many circuits that people use may not be optimized for the very best noise, and even BJTs in those circuits with BJTs will have as much noise as a well-designed JFET-based circuit.
I believe that the higher RFI immunity of the JFET is partially responsible for the better sound. Bear in mind that the signal entering a BJT op amp goes into a forward-biased junction (base-emitter) eager to rectify. The signal going into a JFET op amp goes into a junction that is reverse biased (gate-source), sometimes by a volt or more, and much less close to the region where it might rectify.
Cheers,
Bob
I would go along with the "rf theory" too. In fact, this has been put forward as one reason why bjt's (such as 5532 etc) might have earned a poor reputation. This noise could apparently be "demodulated" in a non linear way (think of an old fashioned RF diode demodulator/detector) wheras FET input stages are essentially immune. It seems a very plausible theory given that the 5532 and similar was extensively used in DAC I/V conversion and filtering, a location that invariably does have a lot of hf hash present. And it was precisely these kind of applications that were seeing big sonic gains when switching to FET devices.