Improving supplies
Harry,
>I've made a simple mod that offers stunning noise performance when compared with any monolithic regulator, whilst lowering output impedance and improving line rejection dramatically.
Again, if interested I'll post circuit / results.>
I'm certainly interested in knowing more on that mod. Can you tell us what you did and what you got?
Carlos
Harry,
>I've made a simple mod that offers stunning noise performance when compared with any monolithic regulator, whilst lowering output impedance and improving line rejection dramatically.
Again, if interested I'll post circuit / results.>
I'm certainly interested in knowing more on that mod. Can you tell us what you did and what you got?
Carlos
back to the plot
Iohk asks "What makes a seemingly simple and/or wellknown circuit sound good ( oh not just good, that is not so difficult, but VERY GOOD)?"
Sometimes it helps to look at things in a different way. I remember looking at circuits and thinking what else does this circuit need to make it sound good - naturally assuming the circuit is missing something crucial. Perhaps it needs more parts to make the output impedance lower, or it needs to run in class A so I need bigger transistors or more regulation of the psus or more complicated current sources or esoteric resistors or whatever. The tendency is to add things. But I have since found this isn't always right. Often the schematic is just fine as it is.
When one looks at a schematic there is a natural tendency to overlook the fact that any two components with the same part numbers (esp semiconductors) are not the same. This is particularly true of npn/pnp complements - they really aren't very complementary due to the physics of semiconductors. All these schematics we are discussing are extremely simplified and don't show what really ends up being implemented when the circuits are built. And each one you build is different because no two parts with the same part number are exactly the same. If the parts on a "simple" schematic were actually all the same when built and npn and pnp were perfectly matched then I suggest that the circuit would sound extremely good (I'm assuming here that basic things are ok like grounding and adequate current capabilities and voltage swings and reasonable stability and psus and all that).
So my advice is to look at the schematic and ask why the physical circuit may not behave the way the schematic would suggest. Try redrawing the schematic showing all the parasitic components and non-linear components that you can think of and assume that their values are different from one device to the next, even if the part numbers are the same. To make life easier my experience indicates that you can assume resistors are perfect and non-electrolytic capacitors are perfect for your first go at this. Focus on the semiconductors (therein lie the problems). Having got this more "realistic" schematic then think about how these new components might or might not affect the way the real circuit will do what the schematic intended.
I'm not a valve expert but I believe (please correct if wrong) that one of the virtues of valve amps is that valves characteristics depend on mechanical dimensions and they are large so managing to buy two that behave similarly is easier than two semiconductors. Also, valves are only npn so valve designers don't get caught out by uncomplementary pairs. The achiles heal of valve designs is, instead, the output transformer.
If you are a manufacturer then you have the horror of having to build hundreds of amps that all sound pretty much the same. So you have to work out which parasitic characteristics are most important to the circuit behaviour and then you have to find a way to select parts or combinations. This is expensive and tricky. For most DIY published circuits you would find that if you built a dozen of the same design that they would all sound different, some much worse than others. This is why the schematic is rarely enough on its own to make a great sounding amp and you have to be cautious of designs and suppliers who haven't solved this issue. A good design guide needs to give you both the schematic AND the means to select parts to go in it. The Naim circuits are especially sensitive to the latter and this is why I pick up on those who claim to have a superior circuit but neglect the parts selection aspect. This is one way I judge whether a designer actually understands how a circuit works.
There are other ways to get around this. One is to parallel up a bunch of nominally identical parts. Threshold do this and Krell does this and many others. We all recognize the power handling benefits but we should also note that it helps to "average out" some of the characteristics of the transistors - so you don't have to be so selective. The downside is that you end up with a harder load to drive (particularly due to capacitance) and that some differences like propagation delays cause a sort of smearing rather than simple averaging. This is why, I believe, Naim do not parallel up output trannies but instead spend a fortune on making single, super-spec'ed custom devices.
This may all sound complicated. Well it is
. But what can we expect? Getting from the ordinary to the extraordinary is always hard - or Kenwood and Krell would sound the same - and it isn't magic. I find it a source of endless fun and fascination. 
BAM
Iohk asks "What makes a seemingly simple and/or wellknown circuit sound good ( oh not just good, that is not so difficult, but VERY GOOD)?"
Sometimes it helps to look at things in a different way. I remember looking at circuits and thinking what else does this circuit need to make it sound good - naturally assuming the circuit is missing something crucial. Perhaps it needs more parts to make the output impedance lower, or it needs to run in class A so I need bigger transistors or more regulation of the psus or more complicated current sources or esoteric resistors or whatever. The tendency is to add things. But I have since found this isn't always right. Often the schematic is just fine as it is.
When one looks at a schematic there is a natural tendency to overlook the fact that any two components with the same part numbers (esp semiconductors) are not the same. This is particularly true of npn/pnp complements - they really aren't very complementary due to the physics of semiconductors. All these schematics we are discussing are extremely simplified and don't show what really ends up being implemented when the circuits are built. And each one you build is different because no two parts with the same part number are exactly the same. If the parts on a "simple" schematic were actually all the same when built and npn and pnp were perfectly matched then I suggest that the circuit would sound extremely good (I'm assuming here that basic things are ok like grounding and adequate current capabilities and voltage swings and reasonable stability and psus and all that).
So my advice is to look at the schematic and ask why the physical circuit may not behave the way the schematic would suggest. Try redrawing the schematic showing all the parasitic components and non-linear components that you can think of and assume that their values are different from one device to the next, even if the part numbers are the same. To make life easier my experience indicates that you can assume resistors are perfect and non-electrolytic capacitors are perfect for your first go at this. Focus on the semiconductors (therein lie the problems). Having got this more "realistic" schematic then think about how these new components might or might not affect the way the real circuit will do what the schematic intended.
I'm not a valve expert but I believe (please correct if wrong) that one of the virtues of valve amps is that valves characteristics depend on mechanical dimensions and they are large so managing to buy two that behave similarly is easier than two semiconductors. Also, valves are only npn so valve designers don't get caught out by uncomplementary pairs. The achiles heal of valve designs is, instead, the output transformer.
If you are a manufacturer then you have the horror of having to build hundreds of amps that all sound pretty much the same. So you have to work out which parasitic characteristics are most important to the circuit behaviour and then you have to find a way to select parts or combinations. This is expensive and tricky. For most DIY published circuits you would find that if you built a dozen of the same design that they would all sound different, some much worse than others. This is why the schematic is rarely enough on its own to make a great sounding amp and you have to be cautious of designs and suppliers who haven't solved this issue. A good design guide needs to give you both the schematic AND the means to select parts to go in it. The Naim circuits are especially sensitive to the latter and this is why I pick up on those who claim to have a superior circuit but neglect the parts selection aspect. This is one way I judge whether a designer actually understands how a circuit works.
There are other ways to get around this. One is to parallel up a bunch of nominally identical parts. Threshold do this and Krell does this and many others. We all recognize the power handling benefits but we should also note that it helps to "average out" some of the characteristics of the transistors - so you don't have to be so selective. The downside is that you end up with a harder load to drive (particularly due to capacitance) and that some differences like propagation delays cause a sort of smearing rather than simple averaging. This is why, I believe, Naim do not parallel up output trannies but instead spend a fortune on making single, super-spec'ed custom devices.
This may all sound complicated. Well it is
. But what can we expect? Getting from the ordinary to the extraordinary is always hard - or Kenwood and Krell would sound the same - and it isn't magic. I find it a source of endless fun and fascination. BAM
ALW,
So can you give us some more info of your experimentation into Naim clones and why you cannot get them to sound as good?
Is it just a matter that Naim has access to components of better quality or can match them better or what? It doesn't seem to me that Naim use components that are unaivalable to the public.
Your post doesn;t seem to explain your assertion very well.
So can you give us some more info of your experimentation into Naim clones and why you cannot get them to sound as good?
Is it just a matter that Naim has access to components of better quality or can match them better or what? It doesn't seem to me that Naim use components that are unaivalable to the public.
Your post doesn;t seem to explain your assertion very well.
Components
In the case of power supplies, Naim use LM3x7K's that are unavailable to the general public, and this is one of the reasons the performance cannot be emulated. The available off-the-shelf 317's, as far as I have been able to test and determine, do not perform as well, in certain key criteria.
The same is applicable to the amplifier output devices, although here there are likely to be solutions, once one knows what one is looking for.
I'll make it clear that my knowledge of the power amps is less than others here, but on PSU's I've done a LOT of work, and have a very deep understanding of what's important.
As I suspect is true for the amps, steady-state performance tells you absolutely nothing, sonically (that's a clue BTW!).
The performance gains from understanding the key criteria are universally applicable too, not something unique to Naim equipment.
A.
In the case of power supplies, Naim use LM3x7K's that are unavailable to the general public, and this is one of the reasons the performance cannot be emulated. The available off-the-shelf 317's, as far as I have been able to test and determine, do not perform as well, in certain key criteria.
The same is applicable to the amplifier output devices, although here there are likely to be solutions, once one knows what one is looking for.
I'll make it clear that my knowledge of the power amps is less than others here, but on PSU's I've done a LOT of work, and have a very deep understanding of what's important.
As I suspect is true for the amps, steady-state performance tells you absolutely nothing, sonically (that's a clue BTW!).
The performance gains from understanding the key criteria are universally applicable too, not something unique to Naim equipment.
A.
Bam
Thank you kindly for that cue
I've been formulating EXACTLY the same thoughts in my head recently, they're both high feedback designs, trying to correct for errors at the output. The bandwidth, stability, dynamic behaviour issues are largely the same it seems to me.
It's only the PP bit that complicates matters
A.
Thank you kindly for that cue
I've been formulating EXACTLY the same thoughts in my head recently, they're both high feedback designs, trying to correct for errors at the output. The bandwidth, stability, dynamic behaviour issues are largely the same it seems to me.
It's only the PP bit that complicates matters
A.
Thanks!!!.
I thank everyone for sharing your views in this forum. I've learnt a great deal about this amp...but no where near enough to produce a great performer.
I must admit I've seen very similar schematics over the years but It's what traderbam said...it's not just the schematic, but rather the component choice and PCB layout that really seperate Naim from the rest...and that's what I find hard to emulate, as Naim do use custom made/obselete trans.
For now Neil McBride's version of the amp is easier for me to tackle.
Minh
P.S stay cool can calm.
I thank everyone for sharing your views in this forum. I've learnt a great deal about this amp...but no where near enough to produce a great performer.
I must admit I've seen very similar schematics over the years but It's what traderbam said...it's not just the schematic, but rather the component choice and PCB layout that really seperate Naim from the rest...and that's what I find hard to emulate, as Naim do use custom made/obselete trans.
For now Neil McBride's version of the amp is easier for me to tackle.
Minh
P.S stay cool can calm.
hi capslock
.....would heartily recommend dennis feuchts' book on analogue design, only available on cd from..
http://www.innovatia.com/ACKTS_AD.htm
get a-hold of Selfs' book....third edition....
see attached for further info...
.....would heartily recommend dennis feuchts' book on analogue design, only available on cd from..
http://www.innovatia.com/ACKTS_AD.htm
get a-hold of Selfs' book....third edition....
see attached for further info...
...... But it also generates a nasty phaseshift when done wrong!help
Diyers help
I build an channel of naim clone this work well..but have an question…
When turn on/off the amp have an small boom in my diy probe speaker…I tester the speaker terminals an have tension in the moment of turn on/off…an second or less..after is ok 0V (normal ) at 40ma…
Is normal this problem….the sound is good this have best detail an more bass,,,, (p3a) is good but less fine
Help help Diego
Diyers help
I build an channel of naim clone this work well..but have an question…
When turn on/off the amp have an small boom in my diy probe speaker…I tester the speaker terminals an have tension in the moment of turn on/off…an second or less..after is ok 0V (normal ) at 40ma…
Is normal this problem….the sound is good this have best detail an more bass,,,, (p3a) is good but less fine
Help help Diego
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.