Occam's Soldering Iron
"Pluralitas non est ponenda sine neccesitate" or "plurality should not be posited without necessity."
http://www.weburbia.com/physics/occam.html
Maybe with need a DIY audio version of Occam's Razor, Occam's Soldering Iron might be a good name.
The simplest audio design that satisfies the listener....
1. is most likely to be built.
2. requires the fewest number of solder joints.
3. requires the least expenditure and hassel for buying parts.
4. is easiest to build and troubleshoot.
5. is easiest to repair.
6. is easiest to layout a pcb or hardwire with the shortest signal path.
7. has the least parasitic coupling due to stray capacitance.
8. encourages attention to detail of the simple gain stages.
9. encourages the use of the best parts affordable since there are less of them.
10. can be increased in complexity more easily than a complex cicuit can be simplefied.
Other suggestions?
H.H.
"Pluralitas non est ponenda sine neccesitate" or "plurality should not be posited without necessity."
http://www.weburbia.com/physics/occam.html
Maybe with need a DIY audio version of Occam's Razor, Occam's Soldering Iron might be a good name.
The simplest audio design that satisfies the listener....
1. is most likely to be built.
2. requires the fewest number of solder joints.
3. requires the least expenditure and hassel for buying parts.
4. is easiest to build and troubleshoot.
5. is easiest to repair.
6. is easiest to layout a pcb or hardwire with the shortest signal path.
7. has the least parasitic coupling due to stray capacitance.
8. encourages attention to detail of the simple gain stages.
9. encourages the use of the best parts affordable since there are less of them.
10. can be increased in complexity more easily than a complex cicuit can be simplefied.
Other suggestions?
H.H.
Geoff,
thanks for sending me the circuit. I will try to track down Self's books in a university library the next time I spend time in an appropriate city. They are pretty expensive and from other web comments I have read do not contain too much new information.
He makes many good points on his web pages, but in my eyes is also sloppy on a couple of points:
- what he calls optimum class B is class AB in classic terminology
- he does not give a definition of what the optimum point is
- he seems to endorse CFB output pairs in spite of the specific problems he points out very clearly
- he skipped a couple of important points in his darlington vs. cascode VAS analysis
- his op amp disto measurements give much worse figures than the data sheets, but he does not state the actual source and feedback resistance values which may be causing the problem
I like your web pages very much. I started to read the Hiraga articles & was surprised how easy to read and yet beautiful technical French can be (my last attempt was reading a mystery story by Izzo that contained so much specific slang that I bought the translation as well -- which did contain quite a few errors). If you need a hand with some part of translation, just ask. Ftorres would be even more competent, being French and writing impeccable English.
Michael,
I have some thoughts on the Self circuit. I think this would be the appropriate starting point for a new thread in the solid state forum to be called "non-switching output stage bias circuits". I will probably start it this weekend, unless there is already a similar thread that has eluded my attempts at finding it. In addition to the Self circuit (mind if I post your scan, Geoff?), I have collected some more references and experiments on this that I will undig, though not this weekend. Could be enough for a whole new forum...
Greetings,
Eric
thanks for sending me the circuit. I will try to track down Self's books in a university library the next time I spend time in an appropriate city. They are pretty expensive and from other web comments I have read do not contain too much new information.
He makes many good points on his web pages, but in my eyes is also sloppy on a couple of points:
- what he calls optimum class B is class AB in classic terminology
- he does not give a definition of what the optimum point is
- he seems to endorse CFB output pairs in spite of the specific problems he points out very clearly
- he skipped a couple of important points in his darlington vs. cascode VAS analysis
- his op amp disto measurements give much worse figures than the data sheets, but he does not state the actual source and feedback resistance values which may be causing the problem
I like your web pages very much. I started to read the Hiraga articles & was surprised how easy to read and yet beautiful technical French can be (my last attempt was reading a mystery story by Izzo that contained so much specific slang that I bought the translation as well -- which did contain quite a few errors). If you need a hand with some part of translation, just ask. Ftorres would be even more competent, being French and writing impeccable English.
Michael,
I have some thoughts on the Self circuit. I think this would be the appropriate starting point for a new thread in the solid state forum to be called "non-switching output stage bias circuits". I will probably start it this weekend, unless there is already a similar thread that has eluded my attempts at finding it. In addition to the Self circuit (mind if I post your scan, Geoff?), I have collected some more references and experiments on this that I will undig, though not this weekend. Could be enough for a whole new forum...
Greetings,
Eric
damping factor
Alaskan:
There are easier ways to lower the damping factor than to play with the feedback:
- insert series resistor
- use cheap cables
- look at the cheapo output coil that many amps have (high series resistance, a very high inductance that will insure operation in all load conditions, not even glue to keep the windings together...)
- look at many crossovers, even in expensive speakers, they have so much resistance that the amp won't matter in many cases
Eric
Alaskan:
There are easier ways to lower the damping factor than to play with the feedback:
- insert series resistor
- use cheap cables
- look at the cheapo output coil that many amps have (high series resistance, a very high inductance that will insure operation in all load conditions, not even glue to keep the windings together...)
- look at many crossovers, even in expensive speakers, they have so much resistance that the amp won't matter in many cases
Eric
Since there are many posts on DiyAudio about simple circuits, I would like to ask the opposite, namely, what is the most number of parts that you have used in any audio component?
I finished designing a new phono amplifier a few days ago, and despite that the amplification is opamp-based, the parts count is surprisingly hefty.
Just wondering about other DIYers and designers.
regards, jonathan carr
I finished designing a new phono amplifier a few days ago, and despite that the amplification is opamp-based, the parts count is surprisingly hefty.
Just wondering about other DIYers and designers.
regards, jonathan carr
John,
I have tried taking GNFB in a 50:50 ratio from driver and output on a SS PP amp. Damping factor is much lower, of course, but interestingly the sound becomes mellower, much like a tube amp, and imaging is improved, along with much greater tolerance to reactive loads. The proportion can be changed with a pot, in fact, and it is instructive to listen as the ratio is changed.
This is pretty obvious stuff, of course, but I certainly agree with you that most amps have far too high a damping factor and it adversely affects the sound by seriously shortening decay. I suspect this is produced by modifications to the operational mode of the driver in terms of the feedback signal. Shame drivers aren't designed exclusively for current feedback.
Eric,
Doug Self does indeed have blind spots. I agree with you emphatically about CFP; he does push them, despite being well aware of their dynamic instability, particularly on the negative rail array. But his work is still seminal because it lays down a credible approach for researching audio amplifiers by defining the problem areas. This adds huge focus to any R&D, and has been extremely helpful to me.
Harry,
I'm particularly taken with your list! Wonderful! It shows almost philosophical zeal about the pragmatic, implementation, and ownership issues. To me, the philosophy and the implementation are the crux of it, and the best amplifiers are created by a ruthless appreciation of topology issues and a relentless attention to detail - all taken with an eye on cost. Interestingly, the math and PSpice models play second fiddle; the issues are largely practical, like component selection, short feedback paths with vanishingly low parasitics, and exquisite care with operating points. I suspect there are lots of designs where form has been more important than function, but this is perhaps not fair since only a few very good amp schematics/layouts are actually in the public domain.
A very Merry Christmas and Happy New Year to all present - with due regard to those not celebrating Christmas and those who recognise other New Years, of course...........
Cheers,
Hugh
I have tried taking GNFB in a 50:50 ratio from driver and output on a SS PP amp. Damping factor is much lower, of course, but interestingly the sound becomes mellower, much like a tube amp, and imaging is improved, along with much greater tolerance to reactive loads. The proportion can be changed with a pot, in fact, and it is instructive to listen as the ratio is changed.
This is pretty obvious stuff, of course, but I certainly agree with you that most amps have far too high a damping factor and it adversely affects the sound by seriously shortening decay. I suspect this is produced by modifications to the operational mode of the driver in terms of the feedback signal. Shame drivers aren't designed exclusively for current feedback.
Eric,
Doug Self does indeed have blind spots. I agree with you emphatically about CFP; he does push them, despite being well aware of their dynamic instability, particularly on the negative rail array. But his work is still seminal because it lays down a credible approach for researching audio amplifiers by defining the problem areas. This adds huge focus to any R&D, and has been extremely helpful to me.
Harry,
I'm particularly taken with your list! Wonderful! It shows almost philosophical zeal about the pragmatic, implementation, and ownership issues. To me, the philosophy and the implementation are the crux of it, and the best amplifiers are created by a ruthless appreciation of topology issues and a relentless attention to detail - all taken with an eye on cost. Interestingly, the math and PSpice models play second fiddle; the issues are largely practical, like component selection, short feedback paths with vanishingly low parasitics, and exquisite care with operating points. I suspect there are lots of designs where form has been more important than function, but this is perhaps not fair since only a few very good amp schematics/layouts are actually in the public domain.
A very Merry Christmas and Happy New Year to all present - with due regard to those not celebrating Christmas and those who recognise other New Years, of course...........
Cheers,
Hugh
This is an interesting thread. Most of the simpler designs have a tendency to sound far superior in my experience. The shortest signal path usually does pay off. The amps listed here are all amps that I've owned.
1. Measures good and sounds good
A few older Krell Models, KSA 50, KSA 80. Some of the better Audio Research Tube Amps but even these require mods to achieve the best they can offer... Dual 75, Dual 76, Dual 115.
2. Measures good but sounds bad
Almost all Mcintosh stuff, tube or solid state, Most Audio Research Solid State stuff along with the majority of their later tube stuff. Almost any Krell amp AFTER the KSA 100, All Mark Levinson stuff I've been in contact with.
3. Measures bad but sounds good
Pass Labs, BAT VK60 ( absolutely stunning amp and the finest sounding commercially built amp I've owned!), Atmosphere M60-MK2....decent amp but way overated and WAY overpriced. Even the kit forms of these I built were overpriced..one can build a pair of them for about 30% of the kit price.
Mark
1. Measures good and sounds good
A few older Krell Models, KSA 50, KSA 80. Some of the better Audio Research Tube Amps but even these require mods to achieve the best they can offer... Dual 75, Dual 76, Dual 115.
2. Measures good but sounds bad
Almost all Mcintosh stuff, tube or solid state, Most Audio Research Solid State stuff along with the majority of their later tube stuff. Almost any Krell amp AFTER the KSA 100, All Mark Levinson stuff I've been in contact with.
3. Measures bad but sounds good
Pass Labs, BAT VK60 ( absolutely stunning amp and the finest sounding commercially built amp I've owned!), Atmosphere M60-MK2....decent amp but way overated and WAY overpriced. Even the kit forms of these I built were overpriced..one can build a pair of them for about 30% of the kit price.
Mark
walker said:
I am similar. For most decent amps (anything mid-hi and up), I cannot tell them apart if I listen to them blindly, not knowing what / which is on.
However, i can improve my chance if I know what amps are included in the comparison (so I know where / what to look for clues).
If I know what is playing and what others to compare against, I usually can tell differences between those amps.
Re: Rube Goldberg anyone?
BTW: I have no fear for using many parts....
BTW2: Anyone who knows what happened to the thread starter John Fassotte Alaskan Audio? ABout complex solutions, anyone who have seen his vinyl amp and power amp? ... not few parts....
When you really think about it, how many parts are involved from the CD out to the speakers including those in each IC? How many parts has a LM3875/3886 for instance?Fred Dieckmann said:
BTW: I have no fear for using many parts....
BTW2: Anyone who knows what happened to the thread starter John Fassotte Alaskan Audio? ABout complex solutions, anyone who have seen his vinyl amp and power amp? ... not few parts....
An interesting statement that also points in this direction was once made by Nelson Pass when he mentioned that they tried both, feedback-takeoff from the drivers and from the output, when he was developing the Stasis (back when he was with Threshold). Both measured amost the same but feedback from the drivers sounded better !
I assume that the sound improvement is not primarily because of less damping factor. I assume it is somehow vice-versa (don't know how to express it in this language), I.e. the feature that actually improves the sound also lowers the damping factor. Two possible "suspects" (IMHO):
- better phase-marging
- less load-dependancy of the feedback signal
Regards
Charles
Like Peranders, I'm not afraid of using many components when there is some performance advantage to be gained by doing so. One exception to this rule is a high-voltage amplifier I recently made; being very afraid of its 2660V supply voltage, I wanted to keep everything as simple as possible, hoping to minimise the required amount of debugging...
Show and tell
>I have no fear for using many parts....<
OK. The parts count for my new opamp-based phono stage is 477, on two 4-layer circuit board, one stacked on top of the other.
Incidentally, the parts count does not include other necessities like power transformers, fuses, LEDs, and so on.
Although the amplification is opamp-based, I decided to design individual voltage regulators for each opamp and circuit block. I ended up with 20 discrete-device voltage regulators (mostly shunt regulators).
I should have realized that things were getting a bit out of hand was when I discovered that I needed to print the schematic on an A2 sheet to make it legible (it was bloody hard to read on A3).
jonathan carr
>I have no fear for using many parts....<
OK. The parts count for my new opamp-based phono stage is 477, on two 4-layer circuit board, one stacked on top of the other.
Incidentally, the parts count does not include other necessities like power transformers, fuses, LEDs, and so on.
Although the amplification is opamp-based, I decided to design individual voltage regulators for each opamp and circuit block. I ended up with 20 discrete-device voltage regulators (mostly shunt regulators).
I should have realized that things were getting a bit out of hand was when I discovered that I needed to print the schematic on an A2 sheet to make it legible (it was bloody hard to read on A3).
jonathan carr
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.