I'm interested in your opinion on a) and d). For me, the better option for both was without.
I didn't hear difference with b).
With c) I haven't experimented. What would be your plan of action? I remember that once I tried to change the resistor values and the result was different output DC offset, which kinda puzzled me, but I didn't think much about it and returned to the original RC circuits.
Very nice built, by the way!
Why not 0R22 and a small choke. 6 turns 1mm wire 8 mm dia. Best of both worlds?
JV's familly was from Oxford. Hence Mr Horn got first look at reviewing it I was told. Naim were bemused at the review as Geoff loved the NAP. He then found by adding the 0R22 to the Quad 405 he got them to sound the same. Sometimes I think Mr Horn wasn't wrong. Critical damping is a question. If you think about it the amplifier being too well connected to the speaker may send time delayed harmonics into the long tail pair. Perhaps that small loss of 0R22 is enough to control the speaker and also improve the amplifiers marriage to the speaker.
When you fit your choke try undamped and square-waves. A damped choke, is it really the right choice?
JV's familly was from Oxford. Hence Mr Horn got first look at reviewing it I was told. Naim were bemused at the review as Geoff loved the NAP. He then found by adding the 0R22 to the Quad 405 he got them to sound the same. Sometimes I think Mr Horn wasn't wrong. Critical damping is a question. If you think about it the amplifier being too well connected to the speaker may send time delayed harmonics into the long tail pair. Perhaps that small loss of 0R22 is enough to control the speaker and also improve the amplifiers marriage to the speaker.
When you fit your choke try undamped and square-waves. A damped choke, is it really the right choice?
I got the same subjective results (although I didn't notice any change in dc offset with changes to the phase networks).
b) definitely changes the sound. I tried this in a slightly less Naim-like circuit I preferred 10R - however it needs tweaks to the LTP balance, so what is really making the difference??).
f) The power supply in general makes a large difference to the sound. No local decoupling in a NAP-like circuit seems to give good results. However I got different results with a preamp
The thing to say is note the current balance each time. The 22K helps as it gives voltages and from that current. In a D Self design we don't get that little bonus. With a little thought the adjustments are not difficult. My hunch is 810R 22K and 10R VAS emitter might do the job. That is guess work and bound to be adrift. I am tying to sweet spot the design. That's 4K3 across the 1K to make it easy to do.
If the DC offset is below 50 mV I would not give it a second thought. It can sound nicer with some.
I suspect a NAD 3020 would suit as stand in preamp.
If the DC offset is below 50 mV I would not give it a second thought. It can sound nicer with some.
I suspect a NAD 3020 would suit as stand in preamp.
Now I have a few more readings these are a few updated musings. Notice I am trying to have more current to the VAS input and swing the balance to be even more like a single transistor. A very small step so as to be safe.
The purple capacitor if my maths correct is starting to work at 10 kHz. If someone had the NAP140 simulated they could say if there is obvious phase shift problems. I don't really think this is the greatest idea I could show. However it restores the VAS gain where typically it needs it most. Ideally this helps keep 5th and 7th harmonics down. There is a school of thought that says > 10kHz distortion is not so important. I don't feel comfortable with that. This purple capacitor is the sort of thing manufacturers might try. Good sound and good specs. The latter a bit false really.
The purple capacitor if my maths correct is starting to work at 10 kHz. If someone had the NAP140 simulated they could say if there is obvious phase shift problems. I don't really think this is the greatest idea I could show. However it restores the VAS gain where typically it needs it most. Ideally this helps keep 5th and 7th harmonics down. There is a school of thought that says > 10kHz distortion is not so important. I don't feel comfortable with that. This purple capacitor is the sort of thing manufacturers might try. Good sound and good specs. The latter a bit false really.
Member
Joined 2009
Paid Member
I don't know if that purple cap ends up fighting with Cdom ? We have Cdom there to purposely roll-off the gain of the VAS in order to maintain overall stability. If it were important to preserve gain above 10kHz the more obvious approach would be to use two-pole compensation in place of Cdom but I don't think it's very effective without using a two-transistor compound VAS. That would make it a different amplifier.
From a sound perspective though, it is the treble that I will be interested to focus on first. I am sensitive to this region the most. My initial listening tests of this amp are that it can create fatigue from the treble. In a hard-surface room (my hobby room) on a full range single driver speaker I found it too harsh, especially at higher volumes. On the 2.5 way B&W speaker in a larger space it was much more balanced. However, I tended to gravitate to songs (on the tuner) that didn't emphasize the treble / presence region. The last amp which gave me this challenge was my TGM2 and in the end I gave up on that one for the reason of not being able to remove the fatigue despite it's other qualities. Given the popularity of the Naim amplifier I trust that it's capable of being set up so that this doesn't become a limitation. I also don't know how much the bench top power supply is impacting the treble performance so I need to get a PSU built. The power transformer I want to use won't be in my hands for awhile so I may have to use something less ideal to start with.
The other parameter that could be toyed with is the compensation, Cdom. I've found it to have a clear impact on the sound of a simple amplifier like this.
From a sound perspective though, it is the treble that I will be interested to focus on first. I am sensitive to this region the most. My initial listening tests of this amp are that it can create fatigue from the treble. In a hard-surface room (my hobby room) on a full range single driver speaker I found it too harsh, especially at higher volumes. On the 2.5 way B&W speaker in a larger space it was much more balanced. However, I tended to gravitate to songs (on the tuner) that didn't emphasize the treble / presence region. The last amp which gave me this challenge was my TGM2 and in the end I gave up on that one for the reason of not being able to remove the fatigue despite it's other qualities. Given the popularity of the Naim amplifier I trust that it's capable of being set up so that this doesn't become a limitation. I also don't know how much the bench top power supply is impacting the treble performance so I need to get a PSU built. The power transformer I want to use won't be in my hands for awhile so I may have to use something less ideal to start with.
The other parameter that could be toyed with is the compensation, Cdom. I've found it to have a clear impact on the sound of a simple amplifier like this.
Last edited:
Member
Joined 2009
Paid Member
What is TID ? how does it sound, is it a contributor to smoothness in the presence region/lower treble ?
I suspect the sound I'm after is all about the front end, especially the LTP because in a feedback amp the 'error amplifier' is key. And so is the compensation which affects the FR and phase behaviour of the feedback.
Tweaking Cdom is one thing to explore, did anybody play with the phase lead feedback values ?
I suspect the sound I'm after is all about the front end, especially the LTP because in a feedback amp the 'error amplifier' is key. And so is the compensation which affects the FR and phase behaviour of the feedback.
Tweaking Cdom is one thing to explore, did anybody play with the phase lead feedback values ?
Last edited:
Member
Joined 2009
Paid Member
I think some are expecting a fix by applying further modifcations and wandering off-track when we already know there are changes to the original. Some will be irrelevant but perhaps others have actually been the cause of the poor sound.
If the problem really exists, it's going to be somewhere in those variations. Trying to patch them over by making further changes to the base design as an afterthought, doesn't seem the right way to go about fixing it though.
FWIW, I often use a regulated dual tracking bench supply to power up new and repaired gear. I also hear sound quality differences to when the amplifier's normal supply is connected, even with similar supply voltages. I'm sure there are more reasons than I know of for why this happens but I usually find that due to lower subliminal noise, there is also greater contrast in low level sounds. In an otherwise quiet environment, the effect of quiet power can be quite arresting and it seems to demand more of your your attention in comparison. I doubt that alone will fit with your experience here though.
If the problem really exists, it's going to be somewhere in those variations. Trying to patch them over by making further changes to the base design as an afterthought, doesn't seem the right way to go about fixing it though.
FWIW, I often use a regulated dual tracking bench supply to power up new and repaired gear. I also hear sound quality differences to when the amplifier's normal supply is connected, even with similar supply voltages. I'm sure there are more reasons than I know of for why this happens but I usually find that due to lower subliminal noise, there is also greater contrast in low level sounds. In an otherwise quiet environment, the effect of quiet power can be quite arresting and it seems to demand more of your your attention in comparison. I doubt that alone will fit with your experience here though.
The 10R VAS resistor might do the magic. It could be 5R1 or 16 R. Make no initial attempt to change 1K 22K. For this test it is just sweetness of sound. If the 2nd/3rd balance needs changing that comes later.
TID is still a bit of a myth. However nearly every designer seems to respect that it could exist. Some call is slewing distortion. IM distortion is a well known type that can be measured. TID is a infered to be a similar distortion that only music finds rather than steady waves. This is highly possible because the steady wave charges Cdom and possibly hides the real situation.
The emitter resistor does two things. It reduces gain and it makes the VAS easier to drive. That is the current required for the Cdom is mostly for Cdom if so. It much more complex than that as there is sourcing and sinking of current.
The gamble is despite reduced VAS gain there is enough gain to overcome distortion. The big big deal is with VAS gain down we might reduce Cdom and have all the REAL HF gain we ever had with better TID.
The logic set up against this might be. We had distortion at -95 dB and degraded it to -85 dB at 1kHz. If that gives us better TID I would vote for that.
I tried to find a neat calcualtor for Re+re and input impedance. Anyone got one? The only thing I did find said external RE is vital for good stability. They didn't mean what we are talking about. Or did they? I can see a possible link.
The VAS illusion is it is purely a current fed input so nothing really matters. Others will say there is an optimum current. However they don't say how much although as much as possible is often said. The LTP is then degerated to bring things near to where it would be at perhaps 1mA tail current. Often 51R added or whatever. This is not a bad plan as then Cdom is made massive to bring it back to stability. Often a current mirror is fitted so as to have every last drop of current for the VAS ( tick ). Resistors could be added to make the NAP clone the same. They would be different values.
I did think last night we could fit a current mirror to the NAP clone LTP. Maybe a resistor to the usually 22K side to divert some current somehow. I didn't get further as I feel it's not where we want to go.
TID is still a bit of a myth. However nearly every designer seems to respect that it could exist. Some call is slewing distortion. IM distortion is a well known type that can be measured. TID is a infered to be a similar distortion that only music finds rather than steady waves. This is highly possible because the steady wave charges Cdom and possibly hides the real situation.
The emitter resistor does two things. It reduces gain and it makes the VAS easier to drive. That is the current required for the Cdom is mostly for Cdom if so. It much more complex than that as there is sourcing and sinking of current.
The gamble is despite reduced VAS gain there is enough gain to overcome distortion. The big big deal is with VAS gain down we might reduce Cdom and have all the REAL HF gain we ever had with better TID.
The logic set up against this might be. We had distortion at -95 dB and degraded it to -85 dB at 1kHz. If that gives us better TID I would vote for that.
I tried to find a neat calcualtor for Re+re and input impedance. Anyone got one? The only thing I did find said external RE is vital for good stability. They didn't mean what we are talking about. Or did they? I can see a possible link.
The VAS illusion is it is purely a current fed input so nothing really matters. Others will say there is an optimum current. However they don't say how much although as much as possible is often said. The LTP is then degerated to bring things near to where it would be at perhaps 1mA tail current. Often 51R added or whatever. This is not a bad plan as then Cdom is made massive to bring it back to stability. Often a current mirror is fitted so as to have every last drop of current for the VAS ( tick ). Resistors could be added to make the NAP clone the same. They would be different values.
I did think last night we could fit a current mirror to the NAP clone LTP. Maybe a resistor to the usually 22K side to divert some current somehow. I didn't get further as I feel it's not where we want to go.
Member
Joined 2009
Paid Member
Ian, I agree - but it's too early to call it a problem. It's just an area of sound quality that I'm sensitive to, some hyperacusis on my part.
fyi:
Naim: Sometimes Shrill, Sometimes Harsh and Sometimes oohhh so Sweet | Naim Audio Forums
Harsh sound | Naim Audio Forums
fyi:
Naim: Sometimes Shrill, Sometimes Harsh and Sometimes oohhh so Sweet | Naim Audio Forums
Harsh sound | Naim Audio Forums
I feel sometimes what we have here is a touchstone or even a religious relic. The true inventor seeks to know the most with least chance of damaging what they took so long to build. It's a bit like travel shows. In the end one has to get on the aircraft and give it a go.
I try very hard to say what might work within what I think people like. Mostly for fun and also to know more. The Venn diagram of what we might do is very narrow here( tick ). It's not without possibilities. This amplifer is not really a NAP140 so why not try to do what Julian did and play with the design. Remember your taste is the only question and not how near to what our hero JV did. That's a bit like painting the Mona Lisa again.
I try very hard to say what might work within what I think people like. Mostly for fun and also to know more. The Venn diagram of what we might do is very narrow here( tick ). It's not without possibilities. This amplifer is not really a NAP140 so why not try to do what Julian did and play with the design. Remember your taste is the only question and not how near to what our hero JV did. That's a bit like painting the Mona Lisa again.
Member
Joined 2009
Paid Member
I no longer have a CD player. I used to own a YBA and it was sweet sounding but I found that I was not using it so it was sold off. I'm not one for listening to a CD from start to finish so I figure some kind of computer will be more flexible for digital music.
Nigel - good advice for sure. What needs to happen still is to get the system closer to being in a proper state for listening. This includes a proper power supply.
And some more hours of burn-in (although I'm not sure I believe in burn-in for amps, it does no harm to offer up something to the audiophool gods from time to time).
I dont like the topology. But there are things unique to this topology. I think it is okay to forget what Naim did and just do the usual amplifier design. Make it as best as you can following your usual design wisdoms. It may be tempting to change the topology tho, but stick to it and try to design a good amp.
Some people like to have a few problems for the sake of having other goodies. Im not like that. For me it is crucial to pay attention to what the topology is lacking. Mid and low is surprisingly good but HF distortion is really terrible. Focus on that, set a design target for 10 kHz THD.
Member
Joined 2009
Paid Member
I don't think I have any usual design wisdoms
Anyhow, the point here is to get the best out of this topology using parts that are consistent with the Naim circuit but with some limited 'improvements' to suit the availability of modern parts. I've populated most of the 2nd channel now with the aim of duplicating the first channel so that I can compare them and satisfy myself they sound the same before implementing changes.
I've experienced a bit of harshness listening to live music, orchestra & symphony if close to the strings (my daughter plays violin at home and it's darn bloody loud if you ask me!). It's the hyperacusis. It raises the question, whether an amp that sounds good to me reproduces this harshness or 'fixes' it. Since it has been pointed out that I'm not building this amp for somebody else, it's my ears that have to be satisfied. How that's gong to happen is part of the journey.
Last edited:
I don't think I have any usual design wisdoms
If your ears are sensitive to fatigue i guess this amp topology has little potential to you. You will prefer a CFP output with its good HF performance. So why not focusing on lowering the HF distortion while keeping the topology intact?
The imbalance in LTP is double edged sword. I know you like low order distortion but NFB does not help the higher order ones caused by output stage and by careless input stage.
I will not focus on VAS as i found its performance is good already.
You can design the LTP using your usual wisdoms, but may be i will prefer a bit more focus on the output stage. Optimizing the bias current to ensure open loop linearity.
In your modificstion wish list you didnt mention about the above. Speedup capacitor etc. With the original driver, may be 100R across baxdiode is nice. But with BD139 i found that lower is actually more suitable. I will focus on perfecting this lower half of the output stage, with a carefully chosen driver first before anything else.
BTW, what is your 10kHz THD?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.