Was Alexander the one who introduced this?
I was pretty sure that I didnt want cdom and zobel. But for fun I tried to simulate to find out why they are audible. And yes, thats the answer.
But why a smaller cdom is needed at all? I rarely worked with LTP but I have never needed it.
As best I know the kit amps with bipolar- FET complmentary pairs.
What I think happened with " the book " is something taken to make a complex subject easier to follow became blameless.
The Bailey 30 watt amplifier published in Wireless World in May 1968 would pre-date the work of Mark Alexander (if he is the source referred to). The article is informative on aspects of complementary symmetry - scope shots and a reference to Early Effect.
http://www.keith-snook.info/wireles...orld-1968/30-Watt High Fidelity Amplifier.pdf
http://www.keith-snook.info/wireles...orld-1968/30-Watt High Fidelity Amplifier.pdf
Also if we are not careful we no longer have a Naim type sound. To shift or refine Cdom and to have a better VAS transistor should be inside what is possible. To question tantallum capacitors is valid as is using a couple of non polar ones arround the LTP. One could by a leap of faith do 2 x Exicon 10N20 to repalce BDY56 of the 1970's designs. I have a hunch that would be Naim plus more micro detail. It's not ideal like JLH version. It would be interesting. Bias would be different. I think to have some ripple filtering from VAS back to LTP should produce a sound not unlike NAP250 and help stop RF entering at that point. Think on this. If we build a class A amplifier hum is always a problem. All attempts to fix the problem have binary outcomes. Work or very often did not work. First lesson is getting the zero volts anchors right and then little tweaks. Often the 1000's uF come last or a capacitance multiplier is used. Along the way ripple current is looked at to keep the capacitors happy. It can take a month of trial an error to get it right. I never felt comfortable with using a ground plane when audio ( it must be better ).
The thing to say is this. That ripple that is so audible when class A is hidden when AB, however not hidden inside the music. It sounds mildly dirty for want of another word. Ironically the Quad 303 that could hide it so well has a regulated popwer supply and a very crafty one at that. I would guess the 303 draws as little at 10 mA at idle, that would hide ripple. One great advantage of the 303 and 1980's JLH is the regulator can be less expensive. If I understand correctely the Quad is a half way house to the virtues ( supposed ) of a shunt regulator. My doubts about shunt regulators is where needed most just like a series regulator it's the output capacitor doing the work. Lets say > 5 MHz. There is good evidence to suggest most op amps which a Naim amplifier is really don't like RF. The FET output stage with JFET input " might " be able to cope better. The speaker leads act as a aerial and the LTP as a detector which causes DC components that look and sound like crossover distortion. The output choke where used I suspect helps. No a big deal and maybe imagination on my part that it is a real problem.
Dave Mate of SSL said to me that a Bipolar- FET pair might offer better oppertunities to use negative feedback. Open loop the linearity looks worse. However in uV movements the increase in i for an increase in v has no step jumps. He went on to say that is the ideal requirement for a feedback amplifer. Otherwise we are talking class A with or without loop feedback. Linearity of output stages is less imprortant than the abrupt switch on. I should say FET's seem more willing to accept loop feedback. The exceptional 50 kHz distortion of the Hitachi design proves that as the open loop distortion of the FET alone is about 1%.
Last edited:
I'm not sure I get your point, but Latfet do need extra gain. You might be able to recall from your memory the gain of Hitachi amplifiers, I predict they are more than the "normal" 22x.
I have 2 designs for latfet amp. One is the Hitachi clone (unfortunately I have only the MPSAxx for the VAS), another is a CFA. The Hitachi clone is not as good as the original (in simulation) because I can't find better VAS transistor than the MPSA. But the CFA, may be this is the benefit of a CFA topology, really exceptional. THD/FFT 100Hz, 1kHz, 10kHz almost the same. THD 100kHz at full power is 0.1%.
The Hitachi clone, fed with 50kHz already collapse, while the CFA, fed with 15 MHz the output still perfect. May be this is normal for a CFA, something I just didn't notice with other designs before.
I forgot to say. Dave had a little trick when doing that which he wouldn't tell me which I infered from talking to his technician. I speculate that the the bipolar part was in class A and the FET picked up where it could. If the bipolar could supply some current to the load the exact bias point of the FET would be unimportant if in the less ideal source follower that my instict says it must be . That more suggests a bipolar- FET Darlington which is not a IGBT nor JLH pairs. It shouldn't be impossible to thing the Quasi version coul;d work as JLH had done the horrible bit.
Dave did hint that Quad couldn't hide the glitches when the 405 and this idea would be better. I always suspect a 405 made this way with higher current limits could sound very good. The Quad 303 and 405 invert. To my ears it matters. Their preamps also invert. If you look how a 303 works it's nothing like a blameless amp and it is blameless.
I would love to see the Blameless one and Dave shoot it out.
Dave did hint that Quad couldn't hide the glitches when the 405 and this idea would be better. I always suspect a 405 made this way with higher current limits could sound very good. The Quad 303 and 405 invert. To my ears it matters. Their preamps also invert. If you look how a 303 works it's nothing like a blameless amp and it is blameless.
I would love to see the Blameless one and Dave shoot it out.
What transistor do you use for the output? What is the fastest transistor(s) possible in this triple "topology"?
You seem to rate Naim amps very highly. Which clone circuit do you suggest I can build that will reflect that "quality"?
I'm familiar with all the mentioned amps, including Naim clone, but I rate the other amps higher. But I'm still searching for the best amp along with this BJT quasi topology.
Dave said if you look at the simplest bipolar class AB you have almost no conduction and then lots at the 0.6V point ( +/- 0.1 V or whatever ). The FET with it's crude switch on point gently rolls up to conduction. The conjecture is uV to uV the amplifer loop is happier when FET. Someone said to me that the amplifer when the 0.6 V or whatever point is reach is in " free fall ". The feedback loop jumps to restore it ( or if a LM324N doesn't ). I suspect all of us have experimented with primitive feed forward amps with 47R and dumper plus TL071 etc and at first thought how remarkable until the spectrum analyser was switched on. We then add bias and the quest begins. In the process we ask ourselves what if I could get more current and for example use a TDA2030 to repalce the TL071. The interesting thing is to replace the bipolar dumpers with FET's. Because the TDA2030 is oversized this could work and maybe with IRF9520 and 520. The TDA2030 will exceed itself as there is no low Z load at > 1 watt . Feedback works well because the amplifier never really jumps or when it does it is in the 3 Vrms area and has maybe 15R total load worst case ( could be >20R ).
That's how I learnt it. Hope it works for you.
ON SEMICONDUCTOR - MJ15015G - TRANSISTOR, NPN, TO-3 | CPC UK
I was in the process of building a better 303. That's as far as I got, I have some clone PCB's. I really must finish it. Every time I start a big project paid work comes in. To be honest money is what I should care about.
The owner of Crimson like me liked the Naim ideas and thought he could do better. He added a 303 type output stage and much more. I think someone should look at that as it's still all NPN. The Naim VAS could be turned down if so and slewing issues reduced. If we took the VAS down to 2 mA we change how difficult it is to drive the VAS base.
Look at the 303 current limiting. It looks far too simple to work. As my speaker gets above 115 dB @ 2 metre they start to work ( I should have the meter on all the time as it's too easy to go there ). They sound just like valves and say sorry chap we won't do that. No crunch crunch. I always thought the 405 was the Diesel version.
I was in the process of building a better 303. That's as far as I got, I have some clone PCB's. I really must finish it. Every time I start a big project paid work comes in. To be honest money is what I should care about.
The owner of Crimson like me liked the Naim ideas and thought he could do better. He added a 303 type output stage and much more. I think someone should look at that as it's still all NPN. The Naim VAS could be turned down if so and slewing issues reduced. If we took the VAS down to 2 mA we change how difficult it is to drive the VAS base.
Look at the 303 current limiting. It looks far too simple to work. As my speaker gets above 115 dB @ 2 metre they start to work ( I should have the meter on all the time as it's too easy to go there ). They sound just like valves and say sorry chap we won't do that. No crunch crunch. I always thought the 405 was the Diesel version.
As to clones. Ian and others know far more. I knew Julian and suspect I knew his brother. We are all from Oxford. I really liked Julian although likable was not how I would describe him. I speak up for Julian as I feel his mechancial engineers mind saw different pictures to most. Perhaps Autistic and had a bit of the Einstein about him. Michael Gerzon and Dave Mate also very similar types with very fast brains. Instead of saying hello to me Julian in a room of people said " I know you, I don't know them ". That knocked me sideways.
The Hitachi graphs show 100 kHz with no real distortion. I measured my ones and was doing all they claimed. I use 2SB872A 2SD756 ( gain 450 ) and 2SB716 ( gain 500 ) 2 x 220R gate stoppers and Exicon 10 N/P16. The power output was the surprise when that high. The Zobel didn't permit too much of that. I had the gain of the amp about like the Quad 405 at about 57. Bias is about 1.2V per pair and 6.5 mV per VAS.
Now to paint the house as I promised Colleen.
The Hitachi graphs show 100 kHz with no real distortion. I measured my ones and was doing all they claimed. I use 2SB872A 2SD756 ( gain 450 ) and 2SB716 ( gain 500 ) 2 x 220R gate stoppers and Exicon 10 N/P16. The power output was the surprise when that high. The Zobel didn't permit too much of that. I had the gain of the amp about like the Quad 405 at about 57. Bias is about 1.2V per pair and 6.5 mV per VAS.
Now to paint the house as I promised Colleen.
Last edited:
None of them.
I think the best Naim power amplifiers are audiophile quality. The manufacturing process is quite sophisticated and involves parts selection and test and measurement and lots of expensive equipment. The same goes for other audiophile grade manufacturers.
I very much doubt anything by the strict definition of "clone" exists at all. Because there would be only two ways to achieve this: know what Naim's process is or successfully reverse engineer the design. I imagine only a few people at Naim have that knowledge and they guard it carefully. JV gave a few things away but by no means everything. Reverse-engineering can be done but it is hard.
I have never bought or built a Naim "clone". Of the few I have seen, including Avondale, I don't see much evidence by examining their published schematics and parts choices that they will sound like a Naim. Not to mention basic mistakes sometimes like putting electrolytic capacitors the wrong way around in schematics. 😉
My advice regarding buying a "clone" kit is to buy the cheapest one you can and work on improving its sound yourself. Pay for the parts and PCB. I would not pay extra for the promise of Naim sound regardless of what marketing BS accompanies the thing. 😎
This should include the power supplies.The whole upgrade chain of Naim was power supplies and very little else. Hi Cap was more or less an NAP 250 transformer core for the pre amp alone.
The selection process was also to save money. Nothing was thrown away. They sent back low spec regulators to the supplier.
It is interesting to look at Exposure amplifers. Similar ideas of one time friends. In my opinion vastly different ideas about engineering and quality. Naim is in many ways like our most famous aero engine maker.
The selection process was also to save money. Nothing was thrown away. They sent back low spec regulators to the supplier.
It is interesting to look at Exposure amplifers. Similar ideas of one time friends. In my opinion vastly different ideas about engineering and quality. Naim is in many ways like our most famous aero engine maker.
Yes indeed. And the wiring layout and the box material - non-ferrous: aluminum. Screened toroidal transformers. Keeping the rectifier/capacitor charging currents away from the PCBs. The alignment of the components, the grounding scheme. Regulating the supplies (in the better models). It's all important.
Thanks, Nigel. It isn't just about the amplifier circuits, although I'd say that's where most of the secret sauce is.
Back to the painting now.
Most people were a bit frightened of Julian. I found he would answer every question and he gave very different insights. I suspect he saw the amplifier as an energy transfer machine.
One thing I read that makes sense is a class AB amplifier is a pendulum that swings between the speakers and the power station. Bad things happen in the AC to DC to musical AC. A class A amplifier is less like a pendulum which is both good and bad. Naim seem to have understood this better than many.
Most people were a bit frightened of Julian. I found he would answer every question and he gave very different insights. I suspect he saw the amplifier as an energy transfer machine.
One thing I read that makes sense is a class AB amplifier is a pendulum that swings between the speakers and the power station. Bad things happen in the AC to DC to musical AC. A class A amplifier is less like a pendulum which is both good and bad. Naim seem to have understood this better than many.
Thats funny 🙂
I think I know what makes an amp ticks. But you make it as if theres voodoo involved 🙂
I dont know what im trying to clone here. I heard Naim amps only from youtube, and even the "statements" sounded weird to my ears.
I have a clear "vision" (soundwise) of a perfect amp. It is not like food that is too subjective.
Here is an interesting hypothetical question: if I give you 5 amp soundclips and you have to differentiate which one is which (lets say one of them is Naim), but you have to do it blindfolded, and if you cannot do it, what is the value of your appraisal of the amps quality?
Ability to do the above is one of the criteria to be able to pick the best amp, and to be able to create one. JV seems to have good ears. And if he is smart enough he should know enough about physics and electronics. But there is no voodoo. It is highly possible to create better amps than Naim, at least better in a blind test. What we need is understanding of the list of variables of what human love in a music.
- Status
- Not open for further replies.