Hi Ian,
I understand. I have some experience of implementing this very type of mod.
I'm making a point about how to think about circuit changes.
If one makes a mod with the intention of breaking an unwanted signal path, and the mod results in a worse sound, it could be the broken signal path is the culprit. BUT it might not be...
When the circuit is changed other things happen other than what one intended. There can be side-effects AND/OR the original intention may not be fully realized because the mod is not perfect.
In this case, one can longer say for sure that the broken signal path caused the sound to deteriorate. There are other possibilities.
Brian
I understand. I have some experience of implementing this very type of mod.
I'm making a point about how to think about circuit changes.
If one makes a mod with the intention of breaking an unwanted signal path, and the mod results in a worse sound, it could be the broken signal path is the culprit. BUT it might not be...
When the circuit is changed other things happen other than what one intended. There can be side-effects AND/OR the original intention may not be fully realized because the mod is not perfect.
In this case, one can longer say for sure that the broken signal path caused the sound to deteriorate. There are other possibilities.
Brian
Example #1
The diode means that the impedance of the front end power rail will change depending on the sag in main supply voltage. Normally, with no signal, the diode will conduct the total current draw of the input stage...maybe 6mA. At this current the diode will have a certain resistance. Once a signal causes a significant drop in the main supply, the diode will turn off. So the resistance seen by the front-end will change from 330R + diode R to open-circuit (approximating that the main psu reactance is negligible). This resistance is in parallel with the 220uF cap.
This modulation of the impedance to ground is a side-effect of the mod.
There are more...do pitch in.
The diode means that the impedance of the front end power rail will change depending on the sag in main supply voltage. Normally, with no signal, the diode will conduct the total current draw of the input stage...maybe 6mA. At this current the diode will have a certain resistance. Once a signal causes a significant drop in the main supply, the diode will turn off. So the resistance seen by the front-end will change from 330R + diode R to open-circuit (approximating that the main psu reactance is negligible). This resistance is in parallel with the 220uF cap.
This modulation of the impedance to ground is a side-effect of the mod.
There are more...do pitch in.
Member
Joined 2009
Paid Member
I agree Ian, with the premise that the interaction between the front end and power stage via one or both power rails can be responsible for some of the ‘sound’ of an amplifier. An engineer will, rightly, say that is a parasitic issue, something wrong. However, in seeking to understand and reproduce the sound of a historical design it becomes part of the solution.
I remember reading that not everyone thought the supply-regulated version of the NAP sounded better.
The AKSA was based on a common topology and the interaction with the front end was exploited by Hugh, once he realized it’s importance, to get a sound he wanted. That design was far more amenable for such manipulation because the LTP was resistor loaded without CCS or current mirrors and it had a bootstrap VAS instead of a CCS.
I remember reading that not everyone thought the supply-regulated version of the NAP sounded better.
The AKSA was based on a common topology and the interaction with the front end was exploited by Hugh, once he realized it’s importance, to get a sound he wanted. That design was far more amenable for such manipulation because the LTP was resistor loaded without CCS or current mirrors and it had a bootstrap VAS instead of a CCS.
Thanks Gareth. I agree that the AKSA55 was more amenable to manipulation than Naim clones though some of the mods applied to the originals as well as kits in the past (see PFM forum - older threads and references) have been quite drastic and maybe unpredictable given that many specified types of components are now obsolete and/or unobtanium.
Traderbam (Brian is it?), I assume your example No #1 refers to Ultima Legione's schematic as copied here by c3kkos in #4023 and 4028. i.e. he adds the diode+resistor+capacitor block in the front end rails. A Naim front end should draw approximately a 12+2 mA current, which is somewhat significant but if we prefer to capitalise on sound effects rather than try to eliminate them, perhaps this still sounds good at modest output levels?
re: c3kkos' schematic, I'd like to point out that the VAS semis nominated on the asc. file and visible on the PCB, will not give you anything like a Naim sound. A lot of work is needed to fix the dismal mess of sound you get with most NAP clone kits as they come but substituting those special low capacitance audio transistors for the original types in there and adding other refinements of typical later Japanese designs, won't help with cloning at all.
Very few kit sellers seem to accept that unless you fit the correct, high capacitance VAS semis to Naim clones, they will sound very poor indeed. The compensation as well as sound quality is adversely affected in a single wrong assumption and repeated by others, from one kit producer to another. Of course, those Diodes Inc. semis are not cheap or easily faked, so you usually get common generic parts regardless. The Zero Zone kits have been good in that respect so don't lose those ZTX653,753 semis. The rest may be more easily replaced.
In summary, I don't think there will be any relationship between the sound quality of Ultima legione's or c3kko's designs and Naim products or clones. What they have done with just the PCB, output stage and some of the passives from them is another way to escape from a persistent cloning error. It's their business what their heavily modified amplifiers actually sound like to others but it must be rewarding and a lot of fun
Traderbam (Brian is it?), I assume your example No #1 refers to Ultima Legione's schematic as copied here by c3kkos in #4023 and 4028. i.e. he adds the diode+resistor+capacitor block in the front end rails. A Naim front end should draw approximately a 12+2 mA current, which is somewhat significant but if we prefer to capitalise on sound effects rather than try to eliminate them, perhaps this still sounds good at modest output levels?
re: c3kkos' schematic, I'd like to point out that the VAS semis nominated on the asc. file and visible on the PCB, will not give you anything like a Naim sound. A lot of work is needed to fix the dismal mess of sound you get with most NAP clone kits as they come but substituting those special low capacitance audio transistors for the original types in there and adding other refinements of typical later Japanese designs, won't help with cloning at all.
Very few kit sellers seem to accept that unless you fit the correct, high capacitance VAS semis to Naim clones, they will sound very poor indeed. The compensation as well as sound quality is adversely affected in a single wrong assumption and repeated by others, from one kit producer to another. Of course, those Diodes Inc. semis are not cheap or easily faked, so you usually get common generic parts regardless. The Zero Zone kits have been good in that respect so don't lose those ZTX653,753 semis. The rest may be more easily replaced.
In summary, I don't think there will be any relationship between the sound quality of Ultima legione's or c3kko's designs and Naim products or clones. What they have done with just the PCB, output stage and some of the passives from them is another way to escape from a persistent cloning error. It's their business what their heavily modified amplifiers actually sound like to others but it must be rewarding and a lot of fun
Yes Ian, the current is closer to 12mA. I threw 6mA in there without much thought because it doesn’t matter.
Example #2
The psu mod separates the front end from the back end with 330R when the diode conducts, and replaces the main psu caps with a much smaller cap. The front end now sees 220uF instead of, say, 22,000uF.
This is another side-effect of the original idea to isolate the front end that changes the circuit’s characteristics.
Example #2
The psu mod separates the front end from the back end with 330R when the diode conducts, and replaces the main psu caps with a much smaller cap. The front end now sees 220uF instead of, say, 22,000uF.
This is another side-effect of the original idea to isolate the front end that changes the circuit’s characteristics.
For the sake of the discussion I think that the 220uF discharge time constant will be much longer than the charge one. Theoretically, the front end shouldn't see anything but some mild saw-tooth shaped DC and shouldn't care very much about the diode on or off.
Having said that, I think that other factors are at play.
My personal opinion, though, is that this mod is pointless. It didn't for me for sure, but it could be a matter of system level design - existing grounding, main power supply quality etc.
Having said that, I think that other factors are at play.
My personal opinion, though, is that this mod is pointless. It didn't for me for sure, but it could be a matter of system level design - existing grounding, main power supply quality etc.
@mjf Yes.
@Ruwe Grounding is my next example.
Example #3
When the front end caps are implemented, they will usually be grounded to the input ground area on the pcb. This creates a new ground path back to the star point which modulates the input ground when the caps are charged up via the diodes. In other words, it creates 2 psu ground points.
@Ruwe Grounding is my next example.
Example #3
When the front end caps are implemented, they will usually be grounded to the input ground area on the pcb. This creates a new ground path back to the star point which modulates the input ground when the caps are charged up via the diodes. In other words, it creates 2 psu ground points.
Definitely, there will be another ground path. But from engineering point of view this may be a good thing - local small currents and noise find their low impedance and short return path coupled through the caps. Noise can sum or cancel and all depends on PCB topology and some other unpredictable conditions, which makes the whole game a guesswork.
From sound point of view, the modification will not necessarily sound better. In my case, long time ago, it didn't.
I think the best approach is to measure and to listen. Then a correlation can be established about what matters and what doesn't in a particular design. In the end we care about the sound. A circuit measuring well is a bonus
Unfortunately, most people who build amps from these kits do it without any equipment except maybe a DMM and a set of ears. Results are unpredictable. As far as the amps works it's a matter of luck if it actually sounds really well or just mediocre.
From sound point of view, the modification will not necessarily sound better. In my case, long time ago, it didn't.
I think the best approach is to measure and to listen. Then a correlation can be established about what matters and what doesn't in a particular design. In the end we care about the sound. A circuit measuring well is a bonus
Unfortunately, most people who build amps from these kits do it without any equipment except maybe a DMM and a set of ears. Results are unpredictable. As far as the amps works it's a matter of luck if it actually sounds really well or just mediocre.
Last edited:
i think the rail decoupling caps need an extra pcb line back to the star/psu ground........in my experience it sounds better (less harsh and annoying and so).
d. self wrote about this in his audio power amp book edition 5 ( distortion 5: rail decoupling distortion, chapter 3 and7)........
d. self wrote about this in his audio power amp book edition 5 ( distortion 5: rail decoupling distortion, chapter 3 and7)........
Naim chooses part types and values for good audio reasons. That can’t be said of clone kits. The design is like a carefully balanced clock mechanism. To choose parts requires understanding how the mechanism is intended to work.
Coming back to Ian’s point about ZTX transistors and their high Ccb.
Is it the case that the relatively high capacitance is truly the important factor or is it that this is just the most obvious distinguishing feature, and it has nothing to do with the sound? After all, I am sure many here would say it is counter-intuitive to choose a high C transistor for the VAS...especially since whatever value it has can be trimmed by the shunted, external Miller cap.
Speculation needs a good theory. If the theory is to find a distinguishing feature, why not choose pin-out or case shape or power dissipation? If there is no good theory, why should larger Ccb get picked as the reason for using Zetex parts?
Coming back to Ian’s point about ZTX transistors and their high Ccb.
Is it the case that the relatively high capacitance is truly the important factor or is it that this is just the most obvious distinguishing feature, and it has nothing to do with the sound? After all, I am sure many here would say it is counter-intuitive to choose a high C transistor for the VAS...especially since whatever value it has can be trimmed by the shunted, external Miller cap.
Speculation needs a good theory. If the theory is to find a distinguishing feature, why not choose pin-out or case shape or power dissipation? If there is no good theory, why should larger Ccb get picked as the reason for using Zetex parts?
... Perhaps Zetex, being a UK company at the time were able to source components to Naim faster, cheaper and who knows what else. There are too many things that are not that obvious had been in result of adjustments in design phase, parts availability, marketing, budget restrictions.
Without the little details the Naim circuit is just the ubiquitous 3-stage class-B amp.
I agree; you can dispute the reason why these and similar "wrong" semis have been there in NAP/NAIT designs from early times but the fact is they were retained over decades for a critical sound quality reason. Substitute a low-Cob semi and the sound simply falls flat, just as we hear from the majority of clone kits and assemblies - as they come. Speaking of capacitance, there is also the Miller cap to consider if changing the VAS transistor Cob. If you reduce it, you also need to increase the usual 39pF cap to something in the range of 47-68pF to compensate, using a 'scope or maybe simulation, if you know what you're doing.
I don't think there are other reasons why these types were settled on, other than their rugged industrial type specs. If you are able to experiment in the real world, try fitting a range of transistors of different capacitance. These might include BD139/140 and 2SD667or 669 plus 2SB647or 649 which were often supplied in kits or used in experiments because they're readily available, cheap and have intermediate Cob of about 15pF. Doubling them to make 30 pF isn't quite the same, I found.
I'll venture that JV had samples of many small signal transistors that could dissipate a watt or so and found that these and similar (originally Ferranti) types simply sounded better to him, regardless of optimum technical performance of the VAS. Remember, he's an amateur looking to make impressive sound quality - not an engineer looking principally for signal integrity, lowest possible noise and distortion etc.
If you read JV's interviews and other quotes, you soon discover that it was sound quality that drew attention at the shows and propelled his products to the top. Anyone can dig up a standard laboratory amplifier design and dress it up as high grade audio amplifier with pretty fascia and lots of gold plating but will it sound particularly good or would you feel an urge to buy? Methinks...not likely.
I don't think there are other reasons why these types were settled on, other than their rugged industrial type specs. If you are able to experiment in the real world, try fitting a range of transistors of different capacitance. These might include BD139/140 and 2SD667or 669 plus 2SB647or 649 which were often supplied in kits or used in experiments because they're readily available, cheap and have intermediate Cob of about 15pF. Doubling them to make 30 pF isn't quite the same, I found.
I'll venture that JV had samples of many small signal transistors that could dissipate a watt or so and found that these and similar (originally Ferranti) types simply sounded better to him, regardless of optimum technical performance of the VAS. Remember, he's an amateur looking to make impressive sound quality - not an engineer looking principally for signal integrity, lowest possible noise and distortion etc.
If you read JV's interviews and other quotes, you soon discover that it was sound quality that drew attention at the shows and propelled his products to the top. Anyone can dig up a standard laboratory amplifier design and dress it up as high grade audio amplifier with pretty fascia and lots of gold plating but will it sound particularly good or would you feel an urge to buy? Methinks...not likely.
More to the point.....
I don't think a theory is necessary to track down odd circuit behaviour but Ill give you a plausible one posted a few times here by Bigun if you search this monster thread and one or two others on the topic. I think it's the best I've heard and so far, nothing else rings true:
Low Cob transistors are preferred in the VAS because they have less effect on high frequency linearity. At higher frequencies, and voltages, the transistor's capacitance is more affected by the signal fluctuations, and compensation capacitance is modulated - an effect that gives rise to unique mix of harmonic distortion and maybe tiny bursts of oscillation, though I've not seen it. More capacitance, more effect here.
So, what do we get with high capacitance? Well, PR&T is one un-subtle effect I'm familiar with. Other people seem to find more subtle and nice things to say about the sound quality and considering Avondale Audio also adheres to this technique with their somewhat similar clone, I'd say the rest of the Britfi industry has known about this trick and probably others from the same stable, for a very long time.
I don't think a theory is necessary to track down odd circuit behaviour but Ill give you a plausible one posted a few times here by Bigun if you search this monster thread and one or two others on the topic. I think it's the best I've heard and so far, nothing else rings true:
Low Cob transistors are preferred in the VAS because they have less effect on high frequency linearity. At higher frequencies, and voltages, the transistor's capacitance is more affected by the signal fluctuations, and compensation capacitance is modulated - an effect that gives rise to unique mix of harmonic distortion and maybe tiny bursts of oscillation, though I've not seen it. More capacitance, more effect here.
So, what do we get with high capacitance? Well, PR&T is one un-subtle effect I'm familiar with. Other people seem to find more subtle and nice things to say about the sound quality and considering Avondale Audio also adheres to this technique with their somewhat similar clone, I'd say the rest of the Britfi industry has known about this trick and probably others from the same stable, for a very long time.
The Zetex parts are also very low noise -- you'll be hard pressed to find a better Vbe spreader than a ZTX851.
I also have the ZTX751's in a two-transistor CCS in a design I'm building -- simply because I was getting rather close to what a TO92 case could handle and the e-line cases can deal with quite a bit more power.
I also have the ZTX751's in a two-transistor CCS in a design I'm building -- simply because I was getting rather close to what a TO92 case could handle and the e-line cases can deal with quite a bit more power.
The relatively high Pd of ZTX is handy, Jeff.
JV raced minis, professionally. He knew a lot about cars. His wins earned him the cash to learn electronics and start a small business. He invented a folding bike. He was a music fan. Later he designed a racing yacht with novel engineering. He was skeptical of HP T&M; he rejected a reps multimeter because the mains transformer was incorrectly oriented and it was buggering up the low voltage accuracy.
To understand his amp design it helps to put yourself in his mindset.
JV raced minis, professionally. He knew a lot about cars. His wins earned him the cash to learn electronics and start a small business. He invented a folding bike. He was a music fan. Later he designed a racing yacht with novel engineering. He was skeptical of HP T&M; he rejected a reps multimeter because the mains transformer was incorrectly oriented and it was buggering up the low voltage accuracy.
To understand his amp design it helps to put yourself in his mindset.
Last edited:
Exactly, and a good reason to respect the earlier, somewhat crude looking products and designs that put Naim on the map.
What happened later with, for example, outboard power supplies for the preamps, was ill-conceived and insanely expensive for what it does. I never had a need for total backward compatibility but I guess some people like to buy more boxes and upgrade their systems at whatever it costs for the fun of it.
Member
Joined 2009
Paid Member
Interesting discovery.
This issue is the only real criticism I have of the sound I obtained from my clone - it can sound harsh on modern speakers. I suspect that the speakers available at the time this amplifier was developed would not have sounded as harsh as my PMC floor standers. My father uses a pair of speakers which we'd now call vintage and they sound very smooth (Celestion Ditton 22) on his Sansui stack. After hearing some harshness myself on my PMC speakers I did some searching around on the 'net and found others have made similar comments. My ears are particularly sensitive (hyperacusis) so I have an interest in this kind of thing. Having the option to add rail decoupling caps on the pcb would be a good thing for me but I have no plans to modify my pcb now. I need to get my clone up and running again to remind myself of the sound - I had assumed that the issue was my power supply - running at too high a voltage compared with the original. I have at least one other amplifier, of a different design, that clearly sounds nicer at lower rail voltages and in that case I tried 3 different voltages (AKSA amplifier).
Attachments
Ah those old Celestions! I always wanted a pair of Dittons when i was a kid; I thought they looked really cool. Times change.
Naim started out just making amps and preamps, of course, and Linn provided the turntable and speakers. That partnership was key to both their success.
Naim eventually designed its own speakers, which I have never heard so I cannot judge, but they never did that well. Now they have become a partner with Focal (common owner). I have heard this combo and it is lively but a tad edgy.
Naim started out just making amps and preamps, of course, and Linn provided the turntable and speakers. That partnership was key to both their success.
Naim eventually designed its own speakers, which I have never heard so I cannot judge, but they never did that well. Now they have become a partner with Focal (common owner). I have heard this combo and it is lively but a tad edgy.