Member
Joined 2009
Paid Member
I don't think it's so simple. I have two amplifiers which do not fatigue but also do not have low HF distortion, one is SS and the other tube. The biggest factor for my ears is usually the source and the speaker, not the amp.
I think it's too early for me to be concerned about this amp until more progress has been made in setting up a proper listening environment and optimizing the chain. It may not be an exact clone but I think it should be capable of eliciting the same enjoyment as many Naim owners report. I'll ask my kids (one plays piano, the other violin) to assist with some listening.
Last edited:
Yes, I have been expecting this response, and I agree with that. First, the HF distortion could be a low order type from the unbalanced LTP, or it could be from the output crossover distortion, which is higher in order. And I just checked that your LTP is more balanced than the Naim.
Yes, I believe that some of the Naim sound is actually the sound of this topology plus the low order distortion. And there are more than one Naim amps. Some could be better than the others.
And like you, I'm also very sensitive to fatigue. I don't believe that the old Naims are good enough for my ears. I can say that even if I have never heard a Naim amp directly. Listening thru Youtube, and analyzing the circuit in Spice is sufficient for me.
But I can see that certain quality in my Naim clone circuit is actually better than my best amplifiers (BJT or Latfet), except for the amp I'm using for daily listening (10W, 0.07% THD!). May be this is worth fighting for (by improving what is lacking)
It's not the LTP. It is simply how much current you can pump into the VAS base. After that it is where that current goes. Ideally it joins the current in the VAS flow.
If the current gain of the transistor is high less current is needed to swing a voltage at the collector of the VAS ( effectively ).
High gain transistors are not available. If we add an emitter resistor to the VAS we make it look as if less VAS current is flowing as we have Re+re.
This idea is good to about 10% Re to Rc. If we say Rc could be a 4K ( 40 V 10 mA ) resistance that implies we wouldn't consider 400R. 10R seems mild. If we take 4000/13 = 308 or 4000/3 = 1300. It is vastly more as the CCS is perhaps 100K or more. Even with the 10R the loop gain will be in op amp country when the LTP is factored in.
Did you never think it is super lazy to make tons of loop gain then throw it away in Cdom? OK the second harmonic distortion will reduce. Surely reducing the VAS gain and reducing Cdom is better? As I said yesterday the mid band distortion might be -100 db and the emitter resistor will drop it to - 80dB ( I doubt it will ). Surely someone will see making the VAS easier to drive is more than worth a thought. It isn't strictly speaking local feedback as that disrespects how it work. It more a Robbing Peter to Pay Paul situation and you must work out who needs it most.
There is one distortion you must think of. Crossover distortion. Your only tool is to grab the output and send a horrible mirror image spike in to correct it. Some believe re-entrant distortion is made in AB amps. That is over correction makes new higher order harmonics that never were there before. Others say NO, it's the feedback loop running out of gain and pulling the stuff out of the noise floor ( I say that as they never do, where was it before ? ). Even siding with them in the noise floor is the better place for them. Like dither we can give a type of linearity by knowning when the job is done.
Now the crunch. When we run a repetitive waves we charge Cdom. Thus the THD improves. Some say the improvement is so dramatic as to invalidate the use of THD beliefs. If class A I would say less true as the loop only corrects device linearity. Thus until better proof is available this has a Quality value rather is a hard fact. Self says surely if these things are true then it should show up somewhere in general linearity. Not really. A pulse through the amp perhaps. A single one shot NE555 wave at 1 kHz perhaps ( ideal as it goes easilly to 19 kHz ). Making Cdom smaller with safety could make a better pulse. I will take Copyright on that idea as the method is cheap. You will need a storage scope, something cheap is OK as it is a trend rather than needing exactly shapes. Good, better, best. That surely is a test that can not lie? If not only play Kraftwerk or 8 Bit Beatles?
A VAS emitter resistor might cure fatigue at the risk of being a bit less dramatic. There might be all you want at 5R1 and 20pF. The NAP LTP makes nice harmonics already so leave it alone. That is unless you want a Denon PMA250.
If the current gain of the transistor is high less current is needed to swing a voltage at the collector of the VAS ( effectively ).
High gain transistors are not available. If we add an emitter resistor to the VAS we make it look as if less VAS current is flowing as we have Re+re.
This idea is good to about 10% Re to Rc. If we say Rc could be a 4K ( 40 V 10 mA ) resistance that implies we wouldn't consider 400R. 10R seems mild. If we take 4000/13 = 308 or 4000/3 = 1300. It is vastly more as the CCS is perhaps 100K or more. Even with the 10R the loop gain will be in op amp country when the LTP is factored in.
Did you never think it is super lazy to make tons of loop gain then throw it away in Cdom? OK the second harmonic distortion will reduce. Surely reducing the VAS gain and reducing Cdom is better? As I said yesterday the mid band distortion might be -100 db and the emitter resistor will drop it to - 80dB ( I doubt it will ). Surely someone will see making the VAS easier to drive is more than worth a thought. It isn't strictly speaking local feedback as that disrespects how it work. It more a Robbing Peter to Pay Paul situation and you must work out who needs it most.
There is one distortion you must think of. Crossover distortion. Your only tool is to grab the output and send a horrible mirror image spike in to correct it. Some believe re-entrant distortion is made in AB amps. That is over correction makes new higher order harmonics that never were there before. Others say NO, it's the feedback loop running out of gain and pulling the stuff out of the noise floor ( I say that as they never do, where was it before ? ). Even siding with them in the noise floor is the better place for them. Like dither we can give a type of linearity by knowning when the job is done.
Now the crunch. When we run a repetitive waves we charge Cdom. Thus the THD improves. Some say the improvement is so dramatic as to invalidate the use of THD beliefs. If class A I would say less true as the loop only corrects device linearity. Thus until better proof is available this has a Quality value rather is a hard fact. Self says surely if these things are true then it should show up somewhere in general linearity. Not really. A pulse through the amp perhaps. A single one shot NE555 wave at 1 kHz perhaps ( ideal as it goes easilly to 19 kHz ). Making Cdom smaller with safety could make a better pulse. I will take Copyright on that idea as the method is cheap. You will need a storage scope, something cheap is OK as it is a trend rather than needing exactly shapes. Good, better, best. That surely is a test that can not lie? If not only play Kraftwerk or 8 Bit Beatles?
A VAS emitter resistor might cure fatigue at the risk of being a bit less dramatic. There might be all you want at 5R1 and 20pF. The NAP LTP makes nice harmonics already so leave it alone. That is unless you want a Denon PMA250.
I tried to find a book that covered the sweetspot of Re+re. Self's VAS buffer does seems well thought out and sort of looks at the problem.
Op amps sometimes have external Cdom pins. This has the vital Cdom inside and your Cdom outside. NE5534 being one. This is highly valued as seldom do we need a gain of 1 ( a buffer ). I think a typical example is > gain of 5 none needed.
The NAP is an op amp. We could view the capacitance of ZTX753 as the internal Cdom and the 39pF at the external Cdom.
One test I did on a similar style amp was giving it a gain of one and then worse. I made it into a filter as you might an op amp. I was very surprised how little extra Cdom was requied.
Anyone worked out the slew rate of this amp? Not very high I guess? At least 10V/uS if considering a 50 kHz potenial. That's assuming 30 V peak and 50kHz. It is said why we really need it is reactive speaker loads. Could be.
Op amps sometimes have external Cdom pins. This has the vital Cdom inside and your Cdom outside. NE5534 being one. This is highly valued as seldom do we need a gain of 1 ( a buffer ). I think a typical example is > gain of 5 none needed.
The NAP is an op amp. We could view the capacitance of ZTX753 as the internal Cdom and the 39pF at the external Cdom.
One test I did on a similar style amp was giving it a gain of one and then worse. I made it into a filter as you might an op amp. I was very surprised how little extra Cdom was requied.
Anyone worked out the slew rate of this amp? Not very high I guess? At least 10V/uS if considering a 50 kHz potenial. That's assuming 30 V peak and 50kHz. It is said why we really need it is reactive speaker loads. Could be.
My design process is such that I see reducing or increasing VAS gain and reducing Cdom are not design objectives. It is part of the many tools that can be used with other tools to achieve certain objectives.
To me it doesn't matter how small the Cdom is because what matters is something else that is achieved when Cdom is of certain value.
I think you have learnt from experience that in many situations, the best trade-off solution is to lower the VAS gain a little and at the end you come up with lower Cdom. But to me this is not a rule. In some cases, I think in all cases actually, collector resistor by itself is not wanted.
But in this topology lowering the VAS gain has little consequences, because you might get almost "nothing" with high VAS gain anyway. In other topology it can be different.
BTW, I have several candidate Naim clones. The one with low supply rail has no collector resistor. The higher power has 33R.
Do you know why a series resistor with Cdom can be a good thing?
The "B" part in class-AB is usually worse than Class-B. Why build class-AB if we have interest only with the "A" part of it. Why not build full class-A. I do have a candidate clone for a class-A version. But then I like JLH69 current-split topology better.
Spot on.
The carpet fitter making my dog worry was from Hayes London. I told him of his local hero Alan Blumlein. Like myself and the carpet fitter Alan was Dyslexic. I did this for him. Alan was said to be a very friendly man who in my opinion was the most remarkable scientist we never remember ( H2s etc ). Alan was said to look at the student engineers and if he saw one in trouble he said " I am a little bit bored, would you mind if we did that together" ?
The carpet fitter making my dog worry was from Hayes London. I told him of his local hero Alan Blumlein. Like myself and the carpet fitter Alan was Dyslexic. I did this for him. Alan was said to be a very friendly man who in my opinion was the most remarkable scientist we never remember ( H2s etc ). Alan was said to look at the student engineers and if he saw one in trouble he said " I am a little bit bored, would you mind if we did that together" ?
If you listen to an uncompressed "raw" recording of live music you will soon realise the that the sound most of us get from our systems is nothing like the real thing - and most of the time, thank goodness for that!
Recordings need to be cleaned up and compressed (not the nasty peak compression used on modern pop) to be palatable for home consumption.
That seems like a better way to start investigating what is going on than restarting with something fundamentally different. I have had to "borrow" the kids' listening abilities several times and its not just their better HF range that's helpful - their ability to quickly learn to resolve complex sounds, assigning it as music, discordant, harmonic and similar, different etc. that surprises me still.
Even so, I guess there will be a few nuances in the layout and I would be interested to see what effect, if any, bridging out the SSR has.
If there are differences worth worrying about, it's in the upgrade choices you have already considered - no real topology issues that I can see there - just some parts choices I have my 'druthers about.
Ian. The older ones seem the critical ones. Not because we have better hearing. It's that we don't want to add to similar problems our own ears have.
It's a bit like a piano. Same piano with a different sound. How come. It's the little things. Steinway had ironed the felt above the place where the keys rest. The BBC didn't like it. So 6 more were made to the older ways. The BBC used students to run them in and felt they had the better Steinways as in days of old.
It's a bit like a piano. Same piano with a different sound. How come. It's the little things. Steinway had ironed the felt above the place where the keys rest. The BBC didn't like it. So 6 more were made to the older ways. The BBC used students to run them in and felt they had the better Steinways as in days of old.
In my work with poweramps the by far biggest difference lies at n Th output stage and especially the A to B transition, FB correction of device switching is not really working well with nasty spikes as a result, I have solved the issue by simply using some of the output current to aid the bias-spread, this way I can prevent the devices from ever switching. With that I'm place I find most topologies remakebly similar sounding, regardless of how they measure
Member
Joined 2009
Paid Member
The interesting thing is that I once built two identical amplifiers except for the LTP. One was a regular LTP running at around 1mA (TGM1) and the other was a CFP LTP running at around 4mA (TGM2). The regular LTP sounded great. The CFP LTP had much better bass, generally cleaner overall thanks to the improved linearity of the error amplifier. But it had a harshness that I couldn't tame. I tried using more Cdom but that didn't address the problem - it wasn't overall loop stability. The VAS was identical in both cases, the drivers, and the power devices. Same power supply. So the VAS and the output stage are important but in the example I'm describing here it was the LTP that made all the difference in the world.
That makes Absolutely sense as it's the feedback path during switching that is most affected. There the conventional LTP does a much better job as large signals literally can pull the CFP out of region where it stuggles to correct the switching artifacts. Solve the switching and you'll hear what the amp, or you'll not hear the amp. (Yet you'll still hear the amp)
I think you got it right MiiB, I do not believe the NFB could fix the transition delay between the output devices switching. There would not be a problem if you never turned them off. You are trying to correct something that was not there in the first place.
It is easy to imagine cross-over with a single sinusoid, try and imagine the mess with a complex musical signal, none of which are harmonically related and probably resulting in harsh the sound.
It is easy to imagine cross-over with a single sinusoid, try and imagine the mess with a complex musical signal, none of which are harmonically related and probably resulting in harsh the sound.
Member
Joined 2009
Paid Member
MiiB, I'm not sure why the CFP would fall-down on large signals, it's generally a far more linear LTP than the regular version - which is why the bass is stronger and the sound cleaner overall ?
The more I read about the Naim amps on various forums (PFM for example), there is a sense that this amplifier has a characteristic sound that includes being somewhat brash in the treble.
The quasi-comp output is not known to have a particularly brash sound, I read that there are many other amps with LTP input and quasi-comp outputs that are well regarded. So perhaps there is something about the Naim implementation. And I'd wager that it might be difficult to remove the brash treble for sensitive ears without destroying the essential character of the Naim sound.
The more I read about the Naim amps on various forums (PFM for example), there is a sense that this amplifier has a characteristic sound that includes being somewhat brash in the treble.
The quasi-comp output is not known to have a particularly brash sound, I read that there are many other amps with LTP input and quasi-comp outputs that are well regarded. So perhaps there is something about the Naim implementation. And I'd wager that it might be difficult to remove the brash treble for sensitive ears without destroying the essential character of the Naim sound.
I am speculating now and I may be completely wrong, but what is the possibility that during the cross-over transition, the output is zero volts for some time thus you are feeding back nothing, while an input signal remains present. The fact that the gain for the long-tailed pair is not balanced may amplify the effect and could make it sound worse/better depending which half has the higher gain.
Last edited:
I remember the early to mid seventies when I often went to our local hi-fi shop to sit and listen to their demo system consisting of Lin LP12, Itok, Asak, Naim, Lin Sara or Isobarik because that was the naim of the game, unfortunately way above my pay scale.
I read the British hi-fi mags and thought it to be the epitome of reproduction systems. Can I really remember what it sounded like for comparison with a clone , I doubt it.
Bigun, you have a tough task ahead of you to call back the past. I don't think anything will ever be the same. Its like wishing you can be young again....
I read the British hi-fi mags and thought it to be the epitome of reproduction systems. Can I really remember what it sounded like for comparison with a clone , I doubt it.
Bigun, you have a tough task ahead of you to call back the past. I don't think anything will ever be the same. Its like wishing you can be young again....
Member
Joined 2009
Paid Member
The way I see this one is: We know that the phase difference between input and output will not be zero across the entire frequency range so the instantaneous voltage at the output might be zero when the input is not zero, but the feedback loop is 'always on', the error amplifier is always comparing input and output - zero is just one voltage of an infinite number of possible voltages, there is nothing at all special about this voltage as far as the error amplifier is concerned which of itself doesn't 'know' where the ground reference is. However, the feedback loop can't fully correct the distortion at the output. If the distortion is worse around cross-over it isn't because the error amplifier went to sleep, it's because there is more distortion at cross-over. I don't see an escape from that in this project since it isn't designed to operate in Class A, or with sliding-bias and there is no provision to increase the feedback factor by adding more OLG.
Still, as there are many examples of quasi-comp amplifiers around which people like, the cross-over distortion of this type of output stage is not an obstacle to enjoying the sound. Unless it's biassed wrong perhaps.
The VAS is a solid design, used in good sounding amplifiers. The LTP is a solid design too, it's just that '22k' resistor that presents something new here. All the elements by themselves are 'proven' somewhere it seems. My spider sense therefore, is that the key element remains the compensation, along with the interaction between amplifier and load.
Last edited:
Member
Joined 2009
Paid Member
Good day. NAIM equipment is, in a sense, a "thing-in-itself", like its fans, which in many forums have their own separate communities, as they do not meet the understanding of supporters of the modern engineering approach.
Bigun, you have completely and accurately reproduced the power amplifier circuit, and it is hardly possible to change anything in the basic characteristics of sound, since the possibilities of this scheme are not unlimited.
You still do not like the transfer of high frequencies, which on some compositions may appear to be "underlined", unduly allocated. That is, there is a feeling of disturbed tonal balance, although this sensation can only occur on certain fragments of music.
But, as I continue to make sure, based on my experience, your descriptions of sound and other comrades who are currently making their clone-NAIM - it can be argued that the power amplifier, without pre-amplifier, has such a feature of sound.
And here it is worth remembering that most fans of this equipment prefer to use both signal sources and preamps of the same brand. And it's worth remembering that specific DIN connectors are used between them. And this in many ways facilitates the construction of a holistic NAIM system. And, as I wrote before, the situation with the balance of the sound varies radically when using their preamplifier: from fixing attention on the upper range there is no trace! Even the recordings made with the original emphasizing of high-frequency sounds, both in the arrangement and in the voices - are brought to a natural balance. This is a very noticeable and significant effect, and when this is observed, the thought comes that: it is in such a combination of components and sound - that there is a "NAIM mystery" ....
Clearly outlined bass, dynamics and a well-transmitted attack of stringed instruments, which was appreciated by a man with experience playing the bass guitar.
Therefore, do not rest solely on improving the NAP component, but find the opportunity to build a more complete NAIM system - "thing-in-itself".
Nikolai.
Bigun, you have completely and accurately reproduced the power amplifier circuit, and it is hardly possible to change anything in the basic characteristics of sound, since the possibilities of this scheme are not unlimited.
You still do not like the transfer of high frequencies, which on some compositions may appear to be "underlined", unduly allocated. That is, there is a feeling of disturbed tonal balance, although this sensation can only occur on certain fragments of music.
But, as I continue to make sure, based on my experience, your descriptions of sound and other comrades who are currently making their clone-NAIM - it can be argued that the power amplifier, without pre-amplifier, has such a feature of sound.
And here it is worth remembering that most fans of this equipment prefer to use both signal sources and preamps of the same brand. And it's worth remembering that specific DIN connectors are used between them. And this in many ways facilitates the construction of a holistic NAIM system. And, as I wrote before, the situation with the balance of the sound varies radically when using their preamplifier: from fixing attention on the upper range there is no trace! Even the recordings made with the original emphasizing of high-frequency sounds, both in the arrangement and in the voices - are brought to a natural balance. This is a very noticeable and significant effect, and when this is observed, the thought comes that: it is in such a combination of components and sound - that there is a "NAIM mystery" ....
Clearly outlined bass, dynamics and a well-transmitted attack of stringed instruments, which was appreciated by a man with experience playing the bass guitar.
Therefore, do not rest solely on improving the NAP component, but find the opportunity to build a more complete NAIM system - "thing-in-itself".
Nikolai.
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.