I found that measuring device gave the same readout for dozens of devices.
I reported that a couple of years ago and surmised at the time it was maybe a limited bit range problem.
But, they are great for checking if a component is still working at low test voltages.
Last edited:
anatech posted an interesting circuit that he devised for his own use. This compared two devices and illuminated an LED if there was a match. It could be configured for PNP or NPN types.
It is a while since I saw this but my impression was it would take a hardly significant outlay and little effort to lash one up.
That's very very easy. The important stereo imformation is the out of phase stuff that defines echo's and most of the spooky stuff we like as stereo. This causes vertical modulation of the the cutter head. This is a major problem and is why the Edison cylinder lost out to the Berliner disc we still use today. Vertical movement is very prone to wear. If kept to reasonable limits record ware is minimised and offers a compatible mono stereo use on all record players. The special modulation is called Hill and Dale.
The LP12 which is a derived design of the AR turntable can be tuned to only move in the vertical direction, lets say 95% to give an idea. The layout of the springs copied from the TD150 tends to stop gyration, often said by " experts " to be a bad layout. Linn also use the arm cable to assist, very tricky to set and often wrong. Thus the turntable being very free to move in the vertical direction makes the work of the pickup arm and stylus easier, this can be seen on warps . The arm will move side to side more noticably if badly set up. If the LP12 is set up to be more free to move it tends to gyrate. This seems to give a different sound which is to my ears is less good although not without other qualities. I would say more detailed, but no good idea of the location of that detail. Two well set up LP12's on a seemingly very solid shelf will share energy like a Newton's cradle. This is why wrongly used mass in turntable designs causes the sense of speed and PrAT to be lost. LP12 is not the best on image stability when heard against a Garrard 301. Getting the springs right makes the difference minimal. However to get the worst out of a 301 is easy. Stone plinths being ideal to get the worst.
To tune a TD150 one just rotates cam washers at the spring bottom. Foam inselts assist. Linn made their springs off centre top to bottom so as to do the same thing. No foam so as to get more isolation. The Linn spring is a swine to adjust. First one of many springs have to be tried. Then adjusted one against another. Often just when sucess is close the LP12 locks solid and one starts again. The TD150 can be set up in 2 minutes and LP12 about 3 hours. LP12 springs in a TD150 do sound better. They fit like a glove. Sorry Linn it's hard to say it's not a copy, albeit better in every way. TD160 is not really as good although looks it should be better.
Naim almost insisted their amplifiers were always sold with a LP12. In the early days a LP12 Rega R200 arm ( not the RB300 which isn't a good LP12 partner ) was often sold. These are very good secondhand buys as the arm is rather good. I seem to remember about £400 in the day. Rega R100 or Ortofon VMS20E II were PU of choice. Interestingly the VMS20 on paper won't work with that arm. In real life it did and very nicely. That was based on 12Hz being an ideal PU arm resonance ( I often think it isn't ). 8Hz can work. 6Hz might be too low. fo= 1/ 2Pi(mc)^0.5. That is, m is total mass of arm and cartridge and c is compliance. Pi is 22/7 to a good approximation. Often these values are not correctly stated and an Ortofon test record used to find the truth. 2Hz wrong is not unusual. The arm bobs up and down at resonance. There is both lateral and vertical tests. Ideally these should not be the same frequency and will be a test of the springs also, wrongly set can be seen as strange movement. Like a car that needs suspension work.
I have no idea if how I test transistors is right. Here goes. Even my cheapest tester will give me a relative value that seems quite close to book specs. For example MPSA42/92 about 100 and 2DS756 about 450, 2N5551 about 160 and BC560C about 520. The meter I sometimes use cost £5 from a model railway shop. I like it as the socket is easy to use with shorter transistor legs, I cut away the plastic to make it even better. I let the transistor settle for about 2 minutes and look out for the battery OK symbol, if on the test is invalid. This seems to give a different reading over a snapshot one.
I keep the transistor out of air currents by using a big cardboard box. When I have plenty of matched samples I do a Vbe test in a similar way. For a LTP this would be ideal. I do seem to get very little DC offset if doing this. Joining those transistors helps, if leads wll stretch use epoxy.
To correct an impression. If this thread is a serrious attempt to do a new amp in the style of the NAP 140 the Cascode is only to make it future proof. If anyone noticed I back tracked to make it a " wrong " Cascode. One that has a little second harmonic. As has been rightly said it is the LTP pair driving the VAS we need look at more than how good the Cascode is. Doing all the right things for none of the usual reasons. To be clear. A BC560C and the Cascode in general might reduce the current the LTP needs to give by a significant amount. Remember in a single VAS we can not source and sink current equally. I call this the one legged cyclist. If one didn't like this improvement a bigger VAS cap should pull it back. An analogy I would give is the modern diesel engine. Often fitted with overhead cam as it is the now way of doing things, not to get 7000 rpm. The Cascode is the same. Not really needed if the range of transistors was what it was, alas it isn't. A Cascode can use a 20V transistor with a 300V type and become a 300 V Cascode with very high gain and very good FT. That's a neat trick. The 300V type then can be a TV type, these tend to be low gain and high linearity to suit CRT drive. To be clear non are truely linear, triode valves get close. That was class A SE without feedback. If non linear colour problems result, typically shadow detail is coloured.
I was asked if a triode valve could be used as a VAS. I doubt it would allow enough loop gain. What we lose in linearity we more than make up for in loop gain, the Naim attempts a high loop gain. The loop corrects the overal linearity up to the Nyquist limit. Cdom mostly sets the Nyquist limit and stops dangerous oscillation.
I have used a Darlington VAS and thought it would be ideal. It had less detail. I then tried a Complimentary feedback pair and found it better and not unstable! It has the advantage of minimal complexity. However it will have very little second harmonic and almost certainly go unstable for others. The " wrong " Cascode with no resistor between TR1 TR2 should be the better choice for us.
The very best Cascode explanation I know of is Morgan Jones on Valves ( he ex BBC ). It's about valves but uses transistors also. His decriptions of transistor CCS is interesting. If you only ever buy one book on valves perhaps this one.
I keep the transistor out of air currents by using a big cardboard box. When I have plenty of matched samples I do a Vbe test in a similar way. For a LTP this would be ideal. I do seem to get very little DC offset if doing this. Joining those transistors helps, if leads wll stretch use epoxy.
To correct an impression. If this thread is a serrious attempt to do a new amp in the style of the NAP 140 the Cascode is only to make it future proof. If anyone noticed I back tracked to make it a " wrong " Cascode. One that has a little second harmonic. As has been rightly said it is the LTP pair driving the VAS we need look at more than how good the Cascode is. Doing all the right things for none of the usual reasons. To be clear. A BC560C and the Cascode in general might reduce the current the LTP needs to give by a significant amount. Remember in a single VAS we can not source and sink current equally. I call this the one legged cyclist. If one didn't like this improvement a bigger VAS cap should pull it back. An analogy I would give is the modern diesel engine. Often fitted with overhead cam as it is the now way of doing things, not to get 7000 rpm. The Cascode is the same. Not really needed if the range of transistors was what it was, alas it isn't. A Cascode can use a 20V transistor with a 300V type and become a 300 V Cascode with very high gain and very good FT. That's a neat trick. The 300V type then can be a TV type, these tend to be low gain and high linearity to suit CRT drive. To be clear non are truely linear, triode valves get close. That was class A SE without feedback. If non linear colour problems result, typically shadow detail is coloured.
I was asked if a triode valve could be used as a VAS. I doubt it would allow enough loop gain. What we lose in linearity we more than make up for in loop gain, the Naim attempts a high loop gain. The loop corrects the overal linearity up to the Nyquist limit. Cdom mostly sets the Nyquist limit and stops dangerous oscillation.
I have used a Darlington VAS and thought it would be ideal. It had less detail. I then tried a Complimentary feedback pair and found it better and not unstable! It has the advantage of minimal complexity. However it will have very little second harmonic and almost certainly go unstable for others. The " wrong " Cascode with no resistor between TR1 TR2 should be the better choice for us.
The very best Cascode explanation I know of is Morgan Jones on Valves ( he ex BBC ). It's about valves but uses transistors also. His decriptions of transistor CCS is interesting. If you only ever buy one book on valves perhaps this one.
Last edited:
Member
Joined 2009
Paid Member
Well, the good thing is that I find this hobby can be a good distraction and I'd like to get my soldering iron fired up again sometime !!
I'm not sure that the parts list I posted was very readable. Anyhow, I'll probably add some more parts, some alternative value etc. so there's stuff on hand to muck about. I do find it easiest if I can get everything from Digikey, one-stop-shop so to speak.
Last edited:
Member
Joined 2009
Paid Member
I heard its predecessor at the Heathrow HiFi show many years ago. The room was empty...just the Lamm playing some innocuous music and me. Pretty damned good. I haven't heard it on rock music but if it is just as good then I would rate this as the best commercial amplifier I have heard. But often these low feedback, multi-transistor contraptions don't do the pace/speed thing well. That is, do they get your nervous system all of a flutter and make you wonder how your speakers ever managed to do that?
Last edited:
My problem with a triode VAS ( or UL ) is thinking that the very simple DC servo action of any op amps style transistor amp might be lost when a valve. Vbe of the VAS being very important. The raw transconductance is missing if triode. ECC88 might be a choice. If pentode a BF series transistor is the better pentode.
A problem we have with hi fi is the speakers that fit nicely into our homes don't really make life easy, they demand an amplifier with damping factor > 10 ( Naim is a bit more into 4 ohms ). I made a very simple valve design to test a theory. One pentode and one 82% UL triode or triode. The 82% UL as it gave better gain with no obvious trade off. Distortion 1% in the Naim style ( better ) that dropped to 0.2% at 1 watt. Sensetivity 620mV. It had no loop feedback. Frequency responce was 15Hz to 62kHz, 20Hz to 35kHz - 1 dB at 5 watts. With cheap speakers like Mission ( 760 ? ) the results were exactly as hoped for. No obvious distortion, no worse than a Revox. Very open and very very fast. I took it to my friends house to try against his NAP250 EPOS ES14 that I knew he was finding hard to live with. To be honest the experimental amp blew what he had away. I didn't exspect that. Where it really scored was showing layers in the music. These are often made by accident and a result of how a mix is made. I doubt very much if this could be repeated with more expensive speakers as the damping factor and modulus of impedance might fight. I could tell this was unhelpful to the friend. He bought some Roksan speakers after that which do change it for the better.
That amplifier works very well with the Quad ESL 63. In fact it is hard to fault it when that pairing. 63's usually sound worse than 57's. Not this time. Often they sound so tame which Quad told me was 63's doing it correctly. I don't think 63's sound right with the NAP 250. 57's do.
After these experiments I feel very frustrated when people reject very OK designs on the basis of distortion. The analogy I would make is leaning about art from a book. It's not telling you everything. A Constable painting isn't anything like it is in an art book. I could imagine the memory distortion is more audible than 0.2% with better than Naim harmonics. That amp took 6 months of smoke and fire to get what I wanted. I call it East to West Ultra Linear. The idea came from a graphical description of Ultra Linear circa 1948 ( North South ). It showed how the curves of a pentode and triode are both distorted and how that distortion could be imagined to have an ideal middle curve. This type of maths suits me. It is this. Tell me and I begin to understand, show me and I can do it. This isn't a radical departure from how UL is taught. It made it far easier for me. Then I said a pentode ( or transistor ) followed by a triode ( or 82% triode ) should also have a UL curve. It does and is not very senstive to valve age if designed with care. The inverting stages helps. Maybe why I don't like cathode followers as they don't. Unlike most valve designs it has 15 mA drive current into the EL34. What I learnt from this is the only real way to get more of what I like is a total rethink. There seems to be no other way.
The Hitachi design is the as near as I ever found to the middle way, that is the double VAS with 2SD756. I feel with a better VAS the NAP 140 clone could do a similar job. Naim amps to me sound metalic ( I don't mind that too much ). I think it must be the VAS. I dare not change the correction networks as that is a radically different amp.
A problem we have with hi fi is the speakers that fit nicely into our homes don't really make life easy, they demand an amplifier with damping factor > 10 ( Naim is a bit more into 4 ohms ). I made a very simple valve design to test a theory. One pentode and one 82% UL triode or triode. The 82% UL as it gave better gain with no obvious trade off. Distortion 1% in the Naim style ( better ) that dropped to 0.2% at 1 watt. Sensetivity 620mV. It had no loop feedback. Frequency responce was 15Hz to 62kHz, 20Hz to 35kHz - 1 dB at 5 watts. With cheap speakers like Mission ( 760 ? ) the results were exactly as hoped for. No obvious distortion, no worse than a Revox. Very open and very very fast. I took it to my friends house to try against his NAP250 EPOS ES14 that I knew he was finding hard to live with. To be honest the experimental amp blew what he had away. I didn't exspect that. Where it really scored was showing layers in the music. These are often made by accident and a result of how a mix is made. I doubt very much if this could be repeated with more expensive speakers as the damping factor and modulus of impedance might fight. I could tell this was unhelpful to the friend. He bought some Roksan speakers after that which do change it for the better.
That amplifier works very well with the Quad ESL 63. In fact it is hard to fault it when that pairing. 63's usually sound worse than 57's. Not this time. Often they sound so tame which Quad told me was 63's doing it correctly. I don't think 63's sound right with the NAP 250. 57's do.
After these experiments I feel very frustrated when people reject very OK designs on the basis of distortion. The analogy I would make is leaning about art from a book. It's not telling you everything. A Constable painting isn't anything like it is in an art book. I could imagine the memory distortion is more audible than 0.2% with better than Naim harmonics. That amp took 6 months of smoke and fire to get what I wanted. I call it East to West Ultra Linear. The idea came from a graphical description of Ultra Linear circa 1948 ( North South ). It showed how the curves of a pentode and triode are both distorted and how that distortion could be imagined to have an ideal middle curve. This type of maths suits me. It is this. Tell me and I begin to understand, show me and I can do it. This isn't a radical departure from how UL is taught. It made it far easier for me. Then I said a pentode ( or transistor ) followed by a triode ( or 82% triode ) should also have a UL curve. It does and is not very senstive to valve age if designed with care. The inverting stages helps. Maybe why I don't like cathode followers as they don't. Unlike most valve designs it has 15 mA drive current into the EL34. What I learnt from this is the only real way to get more of what I like is a total rethink. There seems to be no other way.
The Hitachi design is the as near as I ever found to the middle way, that is the double VAS with 2SD756. I feel with a better VAS the NAP 140 clone could do a similar job. Naim amps to me sound metalic ( I don't mind that too much ). I think it must be the VAS. I dare not change the correction networks as that is a radically different amp.
Damping or drive is the most important variable in my book. For maximum drive, the amplifier should be made very strong and the speaker should be made very easy to drive.
Yes it is mostly by accident. But if the correlation between sound and measurement is well understood then no need ghost chasing.
Many expensive speakers are actually made to be easy to drive. I choose drivers that performs well without series components. Only parallel correction networks.
Latfet is perfect except one thing: drive. Its almost the only determinant factor. The double vas is the best topology that i know that can give proper drive for the latfet. Been interested with ian's CSD
But you need to understand that distortion is one of the most important metrics for sound quality. Not the only one but people have different acceptable thresholds. For one, 0.5% might not be an issue but for others it can be serious. And especially when we can afford better or even the best amp, anything with audible distortion is almost unacceptable.
That metallic sound of the Naim IS the distortion. Can you live with that? I doubt it. You have had experience with better sound to know that there is something better.
I read that the best of Naim is around 0.02%. Even that is audible. But i can live with something like that.
While VAS is usually the weakest part of an amp, i think distortion is the problem. The Hitachi and the Naim has perfect VAS due to the selected transistors. As we can see Naim has always used those ZTX. I simulated mpsa42 but i need bigger cdom (47pf as opposed to 39pf).
So, is it hard to achieve Naim with lower distortion? Yes. Now i know why majority amps with this topology doesnt sound good. There is a certain criteria to be fulfilled for an amp to sound good whatever the topology, and i found out that it is harder when this topology is used. Almost there is no chance that it can be achieved by accident.
Member
Joined 2009
Paid Member
Member
Joined 2009
Paid Member
DDG,
Consider this. A SS amp is not acceptable if its THD is higher than 0.05%, a conventional figure. But with tube amps, THD may be up to 5%, and listening levels are typically 1-2%.
Why is this? What is the THD telling us which is definitive?
HD
Not really. 0.03% or 0.02% is a minimum specification which anyone can read in old reviews or advertising. By conventional steady sinewave testing, even an entry level Nait 2 can still do 0.002%. Select "bias-settings in Naim amps" under the "modification" heading here:
Just one thing about music - when it hits you feel no pain
There is no metallic Naim sound other than that likely from the many cheap clone kits that specify or come with many wrong, even fake semis. In your situation of relying on Youtube sourced sound and unknown re-amplification for reference, you are obviously working under the false impression that what you are seeing of Naim Audio amplification (or any brand for that matter) is what you hear, as if there were a 24bit or better quality sound recording, playing directly through the amplifier and full range, high quality speakers in your listening room. Youtube sound may be convincing to you in the absence of the real thing but for others here, such statements just aren't credible.
I'm sure you understand the truth in that but don't seem to recognize its importance for comparing the real products in real, recommended and appropriate audio systems. It's very important, just like a controlled recording, replay and listening environment is necessary for fair comment on any audio system or component.
Last edited:
Because tube thd is a good kind, better sounding than SS thd.
But none of these is good. Like migrain and cholera both is sickness even tho one is better than the other. What im saying is that people should not enforce his own preference on distortion kind and threshold level because ears are different.
Obviously im not saying that amp with 0.05% thd should sound better than amp with 0.1%. But thats because the other quality variables are different too! If the other differences are not critical then yes, from distortion pov the former is better. Thats definitive.
It will be easier to understand this if we can hear sounds as a group of different sounds. We can hear which one is the sound of second harmonics, we can hear the sound of weak damping, we can hear the sound of intermodulation, we can hear white noise, group delay, etc, then we will not be confused when someone say that X% of imd is too much.
- Status
- Not open for further replies.