That means that I trow another theory in the mix, or just repeat what we believe is the most likely reason - the harmonic structure...
I'm sure that in the 70's the value of this resistor was experimentally determined with measurements.
I doubt that JV and whoever designed the circuit sat on a table and decided "Let's put 22k here, it will sound magical. It will be our trade secret from now on."
As someone who's paid to test, verify the validity and ring the alarm if the theory and the reality disagree, my suggestion would be something in that direction:
- borrow a NAP-250 and perfectly balance the LTP
- attach a load (several resistive/capacitive combinations) will reveal more for sure.
- Put across the load a spectrum analyzer. Measure and record.
- Put back the 22k and repeat the same measurements.
Publish the snapshots here so we all can see and finally put to rest this thing. Another option is Nigel to take to the pub a Naim engineer, to get him very drunk and ask him for the data
Otherwise we can circle in the same circle forever and you know what is the Einstein definition of insanity, don't you?
The problem is, human ears are so bad. Given 2 different amps they are usually not capable to tell which one is better by ears alone. Objectively, they don't even know what metrics to use to judge if an amp is as good or worse as it sounds.
And people think it is a "mature" technology?
You're really funny and the perfect hi-end audio customer!!!
The fact that you don't have equipment available or you don't know what to measure and where doesn't mean measurements are wrong.
What is the right way to determine this resistor. Listening at full moon? Youtube?
Or JV had such magical ears that of all resistors they tried he said that's the one. 22.1k was wrong, 21.9k was wrong, but 22k was exactly what he wanted.
My version is that they just created decent harmonic structure with 22k, meaning 2nd more than 3rd, 3rd more than 4th etc. Then it happened that he liked it. He was in good mood that day and it looked right on paper, so they kept it. I don't know if you're aware, but in the 70's and 80's Naim didn't even publish power ratings of their amps. Customers shouldn't care of that. Just connect and enjoy. No distortion specs, no power, no optimal speaker impedance info. Nothing. Only improtant was to buy a whole Naim system.
JV was saying so many things in his threads and interviews:
- Naim will never go digital, because digital sounds like crap only vinyl was right. Nevertheless I somewhat agree with that, check their current digital line!
- Naim will never use RCA, because they sound like crap. At some point they started using RCA, now I don't know. Back to DIN I heard.
- Naim can sound right only with their own speaker wire and with the SNAIC. Not true anymore in their current line of amps.
- When they started with CD players: no DAC boxes - it's bad for digital signal to put the transport far from the DAC. How about their DAC line today?
- And last my personal experience: Loose AC socket on the back on my player. According to Naim it happens that if it's tight my CD player wouldn't have sounded right.
My quick inspection revealed that the snap-on AC socket was just made for thicker back panel. So, it sits loose. When I sold this player I bought an Oppo 103. I find it better in any respect and about twice cheaper than what I got for the Naim... My ears, my system.
I don't blame Naim, though. Compared to other companies they are top notch. I don't know if any other company would allow their customers to see the serial number of their units online and to determine the production year. If you buy second hand that might be important.
Also they have full in-house production - cast iron enclosures, boards production, CAM machine, assembly. That's respectable for a relatively small company in a niche market.
What I mean is that this voodoo is going in this business forever. Why worry who put what resistor where? You don't see it through the cover, right?
You understand that I write all this because, just like you, I'm not that busy right know?
I'll leave someone else to pick up the torch. It's late.
Last edited:
No. It is obvious that I don't want to share everything I know. But still, I want people to know that there is something there to find out. You don't feel challenged?
I'm not a customer. I collect and sort the best sounding DIY amplifiers for almost 30 years. Since several years ago, my best amps are my own creations.
The fact is I have the equipment. But that procedure you mention is not how you can decide the values, or how you can understand the reason behind it. It is called a secret because smart people like you don't have no idea
I doubt that he used his ears for that one. Did he has SPICE back then? Unlike oscilloscope, everything can be seen in Spice. That's the beauty of simulator.
Why worry who put what resistor where? You don't see it through the cover, right?
Many cloners use exactly the same compensation as NAP250. Do they know exactly how these components work? Check other Naim amps with different transistors, I'll bet the compensation is different.
I heard they also use Sanken output? Can you show me the schematic?
Finally some similarity between us
Just that I don't collect them and don't call them "best"... I believe that you have secret... So far we have heard only various claims pointing in all possible directions. So, rest assured, you are keeping your secret very well. I'am not interested in your secret. What I try to understand is if you use some difficult to figure out logic in your thinking.
I don't believe you about the equipment, though. You would know better if that was the case.
And, please, don't take too seriously my "test plan". I was teasing you, because I knew you didn't have original Naim around. I also do not have one. Although, the output is what you want to see first, because that is what the speaker will see. The speaker won't see the VAS for sure, but the VAS operation will affect the output. Do you agree at least on that??
SPICE in the 70s is something not to fully trust. Even the models today are not to trust 100%. Don't expect too much from it. What you see is not necessarily what you'll see in reality. I know it because I witness it every day. However, the spectrum analyzers where almost as good as today's. Bulky, heavy, with small ugly CRT screens, but very accurate.
Finally some similarity between us
What I try to understand is if you use some difficult to figure out logic in your thinking.
Although, the output is what you want to see first, because that is what the speaker will see. The speaker won't see the VAS for sure, but the VAS operation will affect the output. Do you agree at least on that??
I collect/sort them, meaning that i keep the best and throw away the rest. I have built your creation too.
I dont understand the second paragraph about logic.
Yes, you can measure thru output. My mistake. But you have to plot it.
Last edited:
I meant that I try to understand what is the point that you try to make in your posts by following your thinking. You claim things that are invalidating one another in many of your posts, so they make no sense. Could be the language.
By two halves, do you mean the slew-rate? The clones have non-symetrical rise times, which I consider to be not good. And the load is a load, it's a given. Of course, it can't be affected easily. But the load affects how the amp behaves under it...
Unless, we are talking about power electronics, where inductive loads are compensated with capacitive reactive component from the driving device and basically become resistive in nature.
I don't design anything, I mostly copycat for the fun of it. If you found a circuit posted by me on diyaudio, it's most likely some variation. I don't care for them. However, the tube amp that I use now I won't post. I have no plans to sell it, but it's my baby. I don't want to read how bad it is
Just like Gareth, I guess, I relax when I solder. Sometimes I believe that a change in given schematic will be beneficial, but usually I don't hear a difference. I take all basic measurements, but am too lazy to go really systematic and to refine a circuit. I don't put much effort in it. And I don't have that good equipment at home anyway to do it correctly.
Last edited:
I think I use things like rhetorical questions and you think I don't know the answer and that confused you.
I edited my post. I was thinking the other compensation components.
Two halves (positive and negative rails). On the VAS, PNP is at the top, NPN at the bottom, and so on.
It took years until I can build my own amplifier from scratch. Prior to that, I modified other people amps (for better result of course).
Of course. That's your secret machine.
My objective is not set by ears, but to prove/validate the concept I use ears. So my amps are proven both by numbers and ears.
There are many questions unanswered. Recently there is a post regarding negative feedback. The OP disagree with the expert (I think Bruno?) and I agree with the OP. You need another expert to disagree with Bruno and we have sufficient proof that this field is far from mature.
Here is a simple question: how do you put loop gain in your design objective? (Do you maximize it, do you minimize it, or what). I'll bet there will be no answer.
One afternoon I was bored and took some RS tag strip. It was useless trying to build a stable complex amp on tag strip. So I built this. Everything about it is wrong except the measurements.
1/ CCS has too low a voltage gain to work. Opps it does.
2/ There should be a CCS to LTP. Try if you can to measure a difference. You will be chalenged.
3/ Surely 100R to VAS is dangerous ( 2K2 at first ). No it isn't, I think I got down to 47+47 pf. Try Self version of 2 pole first ( his to ground or rail ).
4/ The VAS can't drive the MOS FET's. Not so as the Cgs is bootstrapped to the the speaker load. Add MOSFET's if wanting more PRaT. 3 sets is about the limit. Cgd is important.
If you are foolish enough to build this amp you might find even 10 uF will serve as Bootstrap CCS. At 1 uF the distortion rises at 1 kHz. It is still very low.
Do use Panasonic Non polar caps where shown. They have distortion at about -122 dB I am told ( better than Black Gate Capacitors I was also told ). They cost pennies. In the ideal world keep them below 250 mV offset. Here is was far less.
A friend of mine unknown to me took something I made like this to a guy who has an Audio Precission test set. He said " What Planet in Nigel from" ? I was deeply hurt by that. Years later I got the courage up to ask what he meant. What he meant was the AP was hard pushed to find anything wrong with it. The 2 pole VAS clean up removes my last doubts about MOS FET's of this type ( BUZ900/905 also ). I wonder would it help the LTP also to make the load slightly more invisible at the VAS when doing this? I would have big doubts doing this with the NAP140 clone if the phase shifting networks retained ( VAS to outputs ).
I like the idea of a resistor into the VAS base using it's capacitance as the pole . Perhaps it has down sides? I never have tried it. Forgive the drawing and doubtless and error will be spotted. I did this a long time ago when I had just started using a computer.
1/ CCS has too low a voltage gain to work. Opps it does.
2/ There should be a CCS to LTP. Try if you can to measure a difference. You will be chalenged.
3/ Surely 100R to VAS is dangerous ( 2K2 at first ). No it isn't, I think I got down to 47+47 pf. Try Self version of 2 pole first ( his to ground or rail ).
4/ The VAS can't drive the MOS FET's. Not so as the Cgs is bootstrapped to the the speaker load. Add MOSFET's if wanting more PRaT. 3 sets is about the limit. Cgd is important.
If you are foolish enough to build this amp you might find even 10 uF will serve as Bootstrap CCS. At 1 uF the distortion rises at 1 kHz. It is still very low.
Do use Panasonic Non polar caps where shown. They have distortion at about -122 dB I am told ( better than Black Gate Capacitors I was also told ). They cost pennies. In the ideal world keep them below 250 mV offset. Here is was far less.
A friend of mine unknown to me took something I made like this to a guy who has an Audio Precission test set. He said " What Planet in Nigel from" ? I was deeply hurt by that. Years later I got the courage up to ask what he meant. What he meant was the AP was hard pushed to find anything wrong with it. The 2 pole VAS clean up removes my last doubts about MOS FET's of this type ( BUZ900/905 also ). I wonder would it help the LTP also to make the load slightly more invisible at the VAS when doing this? I would have big doubts doing this with the NAP140 clone if the phase shifting networks retained ( VAS to outputs ).
I like the idea of a resistor into the VAS base using it's capacitance as the pole . Perhaps it has down sides? I never have tried it. Forgive the drawing and doubtless and error will be spotted. I did this a long time ago when I had just started using a computer.
BTW. The Bootstrap can be done differently to encourage maximum voltage swing. This 3K3 + 3K3 worked well enough. I didn't ever try to work out the open loop gain of this amp. It must be rather good. It loves negative feedback. If you wanted more second harmonic do the 22K 1K trick. It has some already. I always specualtaed that FET amps of the Hitachi type didn't have the same Slewing problems for a given Slew rate. The way I do it is look for crossover distortion at high frequencies. Mostly where it is on the wave rather than being on the centre line. Use a LM324 giving 10 mA as a teaching aid. Often oscillation is seen when the problem arrives. I have no idea if that is a test others use. Asymetrical square waves at 30 kHz I have learnt to disregard as long as stable and not awful, not so if 20 kHz.
Within this month I read a good explanation from a guru about the plus and minus of several compensation schemes. Electronic jargons confuse me. But I know why I don't want to use it, from simulator.
Most people use compensation to stabilize things. That's funny. My process ensures that I will yield a stable circuit, even tho I don't focus on the stability.
Looking from your circuit, the VAS transistor is the one used by Hitachi, right? I think it will not have difficulty driving LATFET input capacitance. I have used similar circuit but driving Naim's output stage, and without any compensation cap so I guess it is not a difficult topology. Why you use such compensation scheme is only you know the answer.
No. It got it past the AP test set looking to be a vastly more complex amplifier, nothing more than that. DVV on another thread said " Nige takes things out until they stop working ". That's absolutely right. Thanks Devan, I dare say I will email you this Christmas.
Devan likes no more than 20 dB loop feedback and as much local feedback as he can. He says and I think he is right some problems of crossover distortion become less when done this way which is against the doctrine of Self or my nearly broken Hitachi design. That never occured to me as right for medium speed devices. His amplifer is highly complex to the eye, not in reality. It uses PNP and NPN LTP at the input and other curve cancellations. My goodness he gets told off and treated like a fool ( sorry if a bit strong ). I think he has sold them for many years in Beograd. He likes medium speed ultra tough Motorola transistors. If I were willing to sell things I would almost try to sell his products even if not my exact cup of tea. I have a hunch I would like them more than Krell.
Devan likes no more than 20 dB loop feedback and as much local feedback as he can. He says and I think he is right some problems of crossover distortion become less when done this way which is against the doctrine of Self or my nearly broken Hitachi design. That never occured to me as right for medium speed devices. His amplifer is highly complex to the eye, not in reality. It uses PNP and NPN LTP at the input and other curve cancellations. My goodness he gets told off and treated like a fool ( sorry if a bit strong ). I think he has sold them for many years in Beograd. He likes medium speed ultra tough Motorola transistors. If I were willing to sell things I would almost try to sell his products even if not my exact cup of tea. I have a hunch I would like them more than Krell.
Its in the bruno putzeys on negative feedback thread. Been reading about nfb, might mixed things up.
Good time. I read with interest this thread of the forum. One acquaintance took the spectra of distortions from his clone NAP140(1К+22К) + NAC, and compared it with the spectra of the Yamaha amplifier P2100. White color indicates measurements with the equivalent load on the amplifier output. In my opinion - the spectra are similar, according to the distribution of harmonics, in the main points, except for a slight difference in the levels...
It was very interesting to see the spectrum of this complete set from NAIM, so I myself own such a system. I apologize for the inaccuracy of the Google translation.
Nicholas.
It was very interesting to see the spectrum of this complete set from NAIM, so I myself own such a system. I apologize for the inaccuracy of the Google translation.
Nicholas.
Attachments
Member
Joined 2009
Paid Member
That's my point about the technology being mature - for the person at the end of the chain, the consumer, the person who pays for the kit and hence who's judgment about the value of the kit is what pays for the industry - Class AB amplifiers are good enough, no further gains are really required. The consumer is looking for other things. One of them is the usual 'feel good' factor that comes from buying into hype.
I love this kind of challenge but I simply don't have the time for it and there are so many other things that interest me. I'll never spend 30yrs collecting the knowledge you have collected - I'll gain some through my own work but the rest of it I will learn from the internet and books. If that means some secrets are not available then so be it, that I can live with. In 100 years from now mostly everyone alive today is replaced with new people - will they care what we share today ?
+1
Member
Joined 2009
Paid Member
I don't know which post either but what I remember disagreeing about was the idea that global feedback is 100% equivalent to (and preferable to) local feedback. One reason is that global feedback samples the output at the load, where the interaction of amplifier and load occurs. Local feedback may not be as sensitive to the load. Another is that phase shifts can be less with local feedback, allowing much higher feedback factors at high frequencies. Maybe I didn't understand Bruno's paper.
Nice data Nikola. I think there are some differences between the green and the white traces but I don't know how important they are. Do these amplifiers sound similar ?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.