Member
Joined 2009
Paid Member
If we believe what's written, the designer did in fact utilize many aspects of the topology to get the sound he liked so I'd ready to believe he was happy with whatever H2 (and other harmonics) that the LTP generates. The lack of degeneration is another 'feature'.
The degeneration resistors are evil and so the resistor (22k) at the second leg of the ltp. But he has used both to achieve his goal in his different amplifiers.
I don't use H2 as my objective. I tend to do the opposite ways of increasing H2. I don't even bother to check to see how it is like. I don't do what many people do which i called 'trivial tests'. They are just 'side effects'. But if i had to check, the result is usually good. As for the distortion, i found that mine have low H2 but monotonic decrease of H2 to at least H4.
I think you are right.
From the technical perspective the 22K resistor doesn't make much sense. It slows down the f/b side of the LTP (by increasing the local gain). This has a marginal effect on the loop stability, but in the opposite direction to what HD suggests i.e. it reduces the phase margin very slightly.
From the subjective perspecive it has quite a marked effect. It provides a "density" to the sound which is appealing in the short term and quite possibly compensates for the thinness to the sound of the NAC circuit. It's a very easy experiment to simply short out the 22K (7K5 in the NAP200) and hear the amp transformed into something that sounds a lot more like most amps.
I have also seen what I call "density" to the sound being described as the Naim "wall of sound".
The old NAP amps are not neutral but do make for nice listening to unfamiliar or background music, maybe because they somehow add some tunefulness, or perhaps simplify the music. I call this the nursery rhyme effect.
However on longer, louder more involved listening sessions I do prefer a more neutral amplifier.
You seem to have very good ears, Dave. After all this time how do you like Mooly's amp? Or do you have another favorite, TGM8 may be? I'm studying CFP right now, to see how much benefit can be gained from the topology (not quite sure yet, but the distortion level is much lower than the quasi like the Naim).
Since you can design your own amp, why not try to design one using same topology but without the 'things' used by Naim. The compensation in series with the output base can be removed, the 22k can be removed, the Vbe 29R can be removed. It is a very challenging topology.
Member
Joined 2009
Paid Member
When I get time (full time job and single parent...) there are a few things like that which would indeed interest me. The project was started because I wanted to hear for myself 'what all the fuss was about' or put it another way - it's industrial archeology!
Many years ago ( 40 ? ) I took a trip to Salt Lane Salisbury. It was a beautiful day and the A338 runs almost to my house. My Honda 550F2 DBW 222R was almost new. Naim Audio was my quest. Julian turned out to be an educator. He could see I was listening so told me more than anyone had before or since about what really matters. His first arguement was that few speakers get close to the performance of a modest amplifier as THD. Quad ESL57 being one he could recomend. Sugden amplifiers also. He gave me to understand the NAP200(?) had been super tuned to the Quads. Naim would make a special NAP250 for Quad owners. The Naim Nait One is very good if on a budget with them.
Julian hinted he had help with the amplifier. The tuning was him. I speculate the 22K 1K is simply to move the balance closer to older designs ( Sugden? ). The Quad 303 which is said to be linked with the NAP200 would have this balance. As would tape recorders. Julian built a British Revox type. If the 22K 1K was replaced with a suitable current mirror I doubt it would be greatly different to a Denon PMA250, a very nice amp. Take the tuning out also.
It could be argued the tuning if unique for Quad ESL 57 would be wrong for other speakers. I suspect Julian would have asked you to prove you could hear it.
Many people say the Harley Davidson like Naim is a flawed design. Ducati is better. Funny how Harley do OK isn't it.
If I find the distortion spectrum of the NAP250 which is very like all Naim amps of the period it will show the second harmonic is very low and is the larger harmonic. It's not quite a perfect ski slope curve. Very close. If I still have it I will post it.
Julian hinted he had help with the amplifier. The tuning was him. I speculate the 22K 1K is simply to move the balance closer to older designs ( Sugden? ). The Quad 303 which is said to be linked with the NAP200 would have this balance. As would tape recorders. Julian built a British Revox type. If the 22K 1K was replaced with a suitable current mirror I doubt it would be greatly different to a Denon PMA250, a very nice amp. Take the tuning out also.
It could be argued the tuning if unique for Quad ESL 57 would be wrong for other speakers. I suspect Julian would have asked you to prove you could hear it.
Many people say the Harley Davidson like Naim is a flawed design. Ducati is better. Funny how Harley do OK isn't it.
If I find the distortion spectrum of the NAP250 which is very like all Naim amps of the period it will show the second harmonic is very low and is the larger harmonic. It's not quite a perfect ski slope curve. Very close. If I still have it I will post it.
Last edited:
One thing I just read is bias settings were very poor on Naim amps. Replacing the bias pots with fixed resistors recomended. Distortion taken at low level ( below 1 Vrms ) was 1.4% as was and 0.02% fixed. Some say this Naim liked you to never switch off the amplifiers. What I could see when fixed was mostly second harmonic.
The Mooly is very good. maybe just a touch dark sounding. Probably more neutral than my NAP200 clone, but they use the same PS so I cannot do a back to back comparison.
I have one amp that always seems to rise to the top of my rankings: Leach low TIM.
Thanks Dave. Unfortunately i don't know what a dark sounding is. I think the crossover distortion (cross conduction) that makes the pronunciation unclear (especially with high bias) or low damping that makes dynamics not so exciting (especially with low gain).
Ah, the feedback at hf is cut off from output and taken from driver stage. More or less similar in principle with Apex' RETRO amp and SANSUI AU-117. I have always wanted to build the Sansui.
My son wanted some active speakers. I suggested to use some of my nice speakers. I have some MS20 which are related to Naim SBL. For fun we bought some PAM8406 boards. Although unlikely to top 1 watt 8R they go loud enough. The graphs if you read them carefully are not unlike a 1960s amp. After an unpromising start they sound very ok. I decoupled them with 10uf. I am using USB from the TV and Teac LSX-8 speakers. Ihave to say I have heard a few mismatched Naim systems sound worse! A £1 amplifier the size of a £1coin! The maths says even old style USB is enough if eight ohms. I seem to remember they can be classs d or ab! The distortion curves are slightly better in ab. PAM8403 is another option. The amp is bridge mode and has no output filtering. I suspect the 240 kHz is in-phase and music out of phase. If so a genius device.
Member
Joined 2009
Paid Member
I would expect those tiny amps are mostly Class D
What criteria should one follow to set the bias current for the lowest distortion - can Spice tell us or do we need a distortion meter?
YouTube
Last edited:
The guy who did the Naim testing used an Audio Pression analyser costing silly money. It is very easy to get identical results using oscillator and. oscilloscope. From memory set the speaker output at 700 mV rms 1 kHz. A real speaker could be the load. After a day of warm up set the waveform for nicest shape. Reduce output into noise and seek best wave shape as it descends. In the ideal would the distortion is filtered to just show harmonics. I dare say Spice can show the low level readings.
Any old style oscilloscope could be used. Elektor did a low cost oscillator at about £2 using a lamp as limiter and TL074 giving -108dB. It is a State Variable Filter type. A sound card scope can be used.
Any old style oscilloscope could be used. Elektor did a low cost oscillator at about £2 using a lamp as limiter and TL074 giving -108dB. It is a State Variable Filter type. A sound card scope can be used.
Member
Joined 2009
Paid Member
Total Harmonious Distortion
Nevertheless, with Class AB output stages, you are looking for close to ideal bias because the distortion products produced there are of the non-harmonious kind.
Nevertheless, with Class AB output stages, you are looking for close to ideal bias because the distortion products produced there are of the non-harmonious kind.
So I don't think Naim ever publishes THD in their specs and JV was always in battle with review magazines to dissuade them from going on about it.
Popular power amp measures include:
THD
Output resistance
bandwidth
But Naim does not excel in any of these. I don't know what the THD of a NAP200 is but the Naims have a highish resistance >0.22 ohms and a lowish bandwidth of 100kHz or less.
Popular power amp measures include:
THD
Output resistance
bandwidth
But Naim does not excel in any of these. I don't know what the THD of a NAP200 is but the Naims have a highish resistance >0.22 ohms and a lowish bandwidth of 100kHz or less.
Douglas Self has written about this many times. He has under biasing as better than over biasing. I don't really think he means that. It's just to make us think. Whilst Self is not critical of the Naim design many of his pet hates are to be found in the Naim range of the 1970's. Most importantly bias drift. His biasing designs are better. I would imagine a copy of his bias system would be better.
My little class D amp I just tried will not have problems of this sort. Whilst THD is modest it should look like class A. If the NAP140 clone is badly set up my little class D will sound better. You can set the NAP 140 clone by ear. Be sure it keeps warm rather than hot. Ears are surprising good for doing this!!!!!!!
THD was said to be <0.1%
My little class D amp I just tried will not have problems of this sort. Whilst THD is modest it should look like class A. If the NAP140 clone is badly set up my little class D will sound better. You can set the NAP 140 clone by ear. Be sure it keeps warm rather than hot. Ears are surprising good for doing this!!!!!!!
THD was said to be <0.1%
BTW. Geoff Horn of Oxford was friendly with Julian. Geoff was much impressed by the sound of NAP250 with Quad ESL57. Mr Horn was curious as to why it was better than the Quad 405. He inserted a 0R25 ( 0R22 ) resistor in the Quad 405 output and noted it was greatly improved. Mr Horn was a trained telecoms engineer so very good at measuring things. He was very certain the Naim idea technically good.
My speculation is the 0R22 divorces the speaker back EMF or whatever from the feedback loop a little. Maybe a little better than a choke at LF. It has no phase shift if not too inductive being a wire wound type, a little is likely to be helpful at RF. Even into 4 ohms damping factor is quite OK at just under 20 ( simple 1+ Rl/Ramp calculation ). 10 is considered where one might notice. Damping of 3 starts to control a cone. If critically done " poor " damping is much better. Each speaker has ideal damping. Some of the valve sound is " poor " damping.
Julain asked me if I ever heard hiss when a record was playing using his products, I said no. He smiled as said " That's what I thought ". It is the right way of looking at things.
My speculation is the 0R22 divorces the speaker back EMF or whatever from the feedback loop a little. Maybe a little better than a choke at LF. It has no phase shift if not too inductive being a wire wound type, a little is likely to be helpful at RF. Even into 4 ohms damping factor is quite OK at just under 20 ( simple 1+ Rl/Ramp calculation ). 10 is considered where one might notice. Damping of 3 starts to control a cone. If critically done " poor " damping is much better. Each speaker has ideal damping. Some of the valve sound is " poor " damping.
Julain asked me if I ever heard hiss when a record was playing using his products, I said no. He smiled as said " That's what I thought ". It is the right way of looking at things.
Very true.2% on a good day. KEF LS50 0.8% Quad ESL 0.1%. The fact the coil moves out of the magnetic field causes second harmonic.If a magnet was placed before and after the cone that might reduce that
- Status
- Not open for further replies.