Pretty much the opposite for me. Self is obsessed with distortion and even then only the quantity, not the "quality". JLH's articles provide much greater insights, and a vastly more rounded view of what actually matters in amp design.
Member
Joined 2009
Paid Member
From DaveS:
Absolutely the same for me! I particularly noticed JLH's preference for phase lead too, it does not change the sound like lag compensation.
HD
Absolutely the same for me! I particularly noticed JLH's preference for phase lead too, it does not change the sound like lag compensation.
HD
Last edited:
Funny you should say that as I found a NAP140 circuit which was offered on a previous thread, very little can be changed if wanting a Naim type of sound. Here are my musings on the diagram. One idea I had recently was to make a preamp where if I choose the output can drive speakers and certainly 32 ohms headphones. It would very easilly drive the 1K inverting input. The bigger problem I see with a regulated pre Darlington stage is to be sure the better 0 volt reference is taken, having the pre Darlington clip first is the nicer clipping and that's ignoring the colouration ripple causes ( NAP 250 has less ). From visiting these threads I get the impression many do not own even a basic oscilloscope, the cheapest 10 MHz types tell you plenty. If so the bigger upgrade one can make is to own one. Has anyone looked at the protecton circuits?
The capacitors Naim used are very much part of the Naim recipe. It's good to find out. Some years ago non polar caps were found to outperform so called Audiophile grade polar caps in terms of distortion, Panasonic are good ones. They even outperformed 63V polyester. The important thing being to keep any offset below 250 mV or at least 400 mV to gain this advantage. The long tail pair 68uF could usefully be improved and even 220 uF might be worth trying. Ironically the bass might sound less impressive at first, let it settle in before saying it's worse.
Using a simple RC filter for the 40 V should be ultra safe.
The capacitors Naim used are very much part of the Naim recipe. It's good to find out. Some years ago non polar caps were found to outperform so called Audiophile grade polar caps in terms of distortion, Panasonic are good ones. They even outperformed 63V polyester. The important thing being to keep any offset below 250 mV or at least 400 mV to gain this advantage. The long tail pair 68uF could usefully be improved and even 220 uF might be worth trying. Ironically the bass might sound less impressive at first, let it settle in before saying it's worse.
Using a simple RC filter for the 40 V should be ultra safe.
On the whole yes. Self did write to me that any bad trait should show up somewhere in simple THD measurements. I completely swallowed that. If I draw an analogy with radioactivity it also will show up somewhere somehow. It killed the lady who had the hunch it existed.
I do not have the confidence to say how a JLH dominant pole might be added to the Naim. JLH although not saying exacly as me gives me to think we should not assume the input transistor collector would happilly pump current into the VAS base and the 39pF. Note that the VAS Z in could be below 500 ohms without any additional mechanisms, the 39pf is not as a boostraped so is a nasty problem. The VAS as I said before when a single type is like a one legged cyclist and the VAS is the worst gear he could choose. Self completely shuns the double VAS when it is common to see it in op amps. This wouldn't change the Naim sound if we could improve it, it would help make mid-fi recordings sound nicer.
Very surprisingly the Phono side of Naim needs it also. If you own a Naim pre-amp the cards can be replaced with wire links to make the input line level. Designing a phono stage is good fun, it can be passive 75uS and active 3180/318uS ( no 7950uS please, speakers to balme if you need that ). You could use 2uS passive after the phono stage. D Self says this, I agree ( Quad used it ).
Good to hear from you Hugh!
I'm fairly intermittent on this forum, but have read some scary stuff about your health "challenges". Hope all is well now.
BTW, I still have a very early AKSA55 (actually the Rod Elliott version 😱) running my kitchen speakers. It's never missed a beat, even with my daughters' modern 😱 music at full blast. Still sounds pretty good in the main system too!
Self has made a great contirbution to our hobby , but I think it would be even greater if he opened up his mind a bit. His original series of articles had some errors (e.g. he completely ignored VAS base current when he was balancing his LTP) but he writes it with such conviction that it is very convincing.
I always took his stuff with a pinch of salt... many years ago a boutique hifi manufacturer told me he had built a "blameless" amp and compared it to his own (rather good!) amps. He said it was one of the blandest, most boring amps he had ever heard. Of course the interesting question is; why?
Going back to simple distortion measurements. Even the worst capacitors measure better than any supposed THD we need. Yet their distortion trends directly relate to what we hear. Most tend to think this is true now.
Here is how THD went from science to goodness knows what it is now. In the 1950's the Quad ESL 57 speakers as we call them gave people the chance to see if they could determine THD as they had distortion similar to a 1947 Williamson amplifier, some of the best recordings ever made come from this era so don't fool yourselves the source material was low fi ( Nor a Thorens TD124 ). What they found was 1% THD is hi fi as long as the harmonics fell on a musical curve assumed to be exponential. It was found that 5th harmonic should not be above 0.1% as a piece of advice, especialy so if a feedback amplifier where 2nd 3rd and 4th were suppressed. Here is where a bit of wrong thinking came in. If 0.1% THD any variation would be perfect ( yikes ). The Quad 303 will sound much like an amplifer 100 times worse that has the same near ideal harmonic structure . And that's exactly right as I have built that amplifer. The Quad is <0.03% at 1 kHz and that is nearly all 2nd. The one I built 1%.
Now to pour doubt on any of this. No one every tries to factor in microphone distortion. I doubt it is less than 1%. That is before even thinking of polar and time distortions of how we pretend to have a hi fi recording ( e.g. Beethoven at a concert verses most I have at home ). What bonkers nonsense this is. It's a bit like a 1960's beauty contest.
Here is how THD went from science to goodness knows what it is now. In the 1950's the Quad ESL 57 speakers as we call them gave people the chance to see if they could determine THD as they had distortion similar to a 1947 Williamson amplifier, some of the best recordings ever made come from this era so don't fool yourselves the source material was low fi ( Nor a Thorens TD124 ). What they found was 1% THD is hi fi as long as the harmonics fell on a musical curve assumed to be exponential. It was found that 5th harmonic should not be above 0.1% as a piece of advice, especialy so if a feedback amplifier where 2nd 3rd and 4th were suppressed. Here is where a bit of wrong thinking came in. If 0.1% THD any variation would be perfect ( yikes ). The Quad 303 will sound much like an amplifer 100 times worse that has the same near ideal harmonic structure . And that's exactly right as I have built that amplifer. The Quad is <0.03% at 1 kHz and that is nearly all 2nd. The one I built 1%.
Now to pour doubt on any of this. No one every tries to factor in microphone distortion. I doubt it is less than 1%. That is before even thinking of polar and time distortions of how we pretend to have a hi fi recording ( e.g. Beethoven at a concert verses most I have at home ). What bonkers nonsense this is. It's a bit like a 1960's beauty contest.
JV had the same experience in the '70s. Lots of people publishing "authoritative" conjectures but no musically convincing products.
When I built an amplifer to drive a big motor I went directly to Self and it worked first time. I have to say that's no small thing. Motor drive catches most amplifiers out. I used his basic concept generic amplifer using the type of T03 Darlingtons as JLH. SOA being prime.
One way to test a bland amplifier is using a casette deck with metal tape, Nakamichi would be ideal. Record as high as your ears say is OK. If the sound is the same more or less as a reference amp ( Quad 303 perhaps ) the blameless amp is blameless. If it becomes bland one must ask how come. Maybe the blameless concept is a photograph and music is a movie? I have heard the effect I describe on Sony amps that sometimes were bland. Often current limiting was also to blame. One 170 watt Sony gave 2 watts into 2 ohms whilst a NAD3020 gave 92! And guess what.
I suspect the blandness you describe is in JLH terms an added sound by a mechanism easy to understand. Slewing is the usual name. JLH opened my eyes to the concept of driving the compensation mechanism from a lower impedance and that it should matter. It must if you think about it.
Be very suspicious of slew rates. The Hitachi MOS FET design from memory ranges from about 20 to 45 V/uS ( twin VAS so two legged sourcing and sinking of current ), the latter 600 watts RMS 4 ohms I think. None have a bad sound and exceptionally low distortion even at 50 kHz. What people ignore is the MOS FET greatly helps the rest of the amp and the forbidden 6V/uS 100 watts will actually work if a tiny bit of bandwidth limiting is used.
The cap can be a burden (load) isn't it (just like a zobel is):
The double-VAS Hitachi mosfet amp looks so good. Why? Look at the VAS transistor!! Where did those transistors go?! Modern clones are worse.
JLH 1969 also uses great transistors (from the era) except the VAS transistor (2N1711) that should be used to match the quality of the MJ480 which is only 4 MHz. He did mention that the design requires fast output stage, which of course allows the use of better VAS transistor, which is important for the final quality. Rod's Death-Of-Zen (BD139 for VAS and 35 Mhz MJL4281 for output) is claimed by Andrew and others as the best implementation of JLH amp. Haha! HD, with the right transistors, JLH is not an effect box.
2SD756, there was even a 180 V version. A friend and I designed a BC337-40 + BF720 cascode to be it's equal with a small resistor between TR1 collector and TR2 emitter to offset Early effect. Cob I think was 1.2pF. > 60 MHz is enough and I suspect would be 150 MHz in truth.
The very cunning HH-1200 uses MPSA42 with a 180R VAS 2 tail resistor ( Hitachi usually 100R ). However there is a 100 nF in paralell that takes over at 10 KHz ( and phase shift so careful ). I suspect they used the 180R to lift the input impedance untill loop gain became critical. Some 2SD756 I have read 450 gain on my meter and MPSA 100. HH seems to be a designer who knew how good the Hitachi was and used his brain to make an industrial version. Rumour says it is one of the better clones.
I was shoot down in flames by Mr S once when I said we might use a VAS emitter resistor to mimic the reuslts of a higher gain. He was livid " how many times have I said one can't add local feedback here ". I didn't say local feedback. I said make it easier to drive. It is a type of feedback regardless of it being a current to voltage converter. The idea of adding gain at high frequencies he covers by saying why throw it away at lower frequecies, yes and no.
The very cunning HH-1200 uses MPSA42 with a 180R VAS 2 tail resistor ( Hitachi usually 100R ). However there is a 100 nF in paralell that takes over at 10 KHz ( and phase shift so careful ). I suspect they used the 180R to lift the input impedance untill loop gain became critical. Some 2SD756 I have read 450 gain on my meter and MPSA 100. HH seems to be a designer who knew how good the Hitachi was and used his brain to make an industrial version. Rumour says it is one of the better clones.
I was shoot down in flames by Mr S once when I said we might use a VAS emitter resistor to mimic the reuslts of a higher gain. He was livid " how many times have I said one can't add local feedback here ". I didn't say local feedback. I said make it easier to drive. It is a type of feedback regardless of it being a current to voltage converter. The idea of adding gain at high frequencies he covers by saying why throw it away at lower frequecies, yes and no.
Last edited:
Much as I rate JLH, I was very disappointed with the sound of my JLH1969 amp.
Mind you, I had the same disappointment with an Aleph30 - all hifi and no music**, I wonder what causes this?
The JLH 80W mosfet amp was an excellent amp, although very sensitive to the design and implementation of the front end regs. I'd take one of these over the Aleph and 1969, any day of the week!
(** slight exaggeration, but you know what I mean)
Mind you, I had the same disappointment with an Aleph30 - all hifi and no music**, I wonder what causes this?
The JLH 80W mosfet amp was an excellent amp, although very sensitive to the design and implementation of the front end regs. I'd take one of these over the Aleph and 1969, any day of the week!
(** slight exaggeration, but you know what I mean)
Whilst on this very valuable theme, a VAS for the Naim? If genuine 2SB716 coud be had it should be a winner ( 2N5401 inside some I guess ). It's the gain rather than Cob or Ft.
Only once? You did very well 😀
A friend of mine sat next to him at an AES dinner and DS had all guns blazing but seemed to be missing the point (in this case about common impedance coupling).
Just out of interest Nigel - which of JLH's amps do you think is the best sounding?
The Naim is the design you want to study to answer your question. I wouldn't get too excited about the other designs and sources being discussed here as they won't help much.
Member
Joined 2009
Paid Member
When the phase lead is driven from the VAS collector (as in the AKSA amp) it will represent a load to the VAS and if placed across the feedback resistor from the output stage it will instead be a load on the output (which I hope would easily drive it). In both cases it stops being a load on the LTP. But more importantly, it helps stabilize the amplifier thus permitting a smaller Cdom at the VAS.
My point is that one leg has to drive Cdom together with the base impedance of the VAS device, which in the Naim design uses a relatively high Cob device by modern standards. The other leg of the LTP does not drive anything. So in a.c. terms the LTP current becomes unbalanced even if balanced at d.c.
Last edited:
- Status
- Not open for further replies.