@Nige
I think I fundamentally agree with you regarding a classic SE input stage, based on a differential pair, with all the bells and whistles (cascodes, FET/BJT combinations, current mirrors) far too often cannot match a properly done fully complementary input stage on sound quality.
Not automatic, of course, but by and large, I seem to prefer fully complementary on sound basis alone, even if it is more difficult to get just right.
Also, I would add that letting in a little more current simply sounds better. By that, I mean current per trannie of say 1.1-1.4 mA, although Dan D'Agostino didn't shy away from 1.6 mA per trannie in his older series of amps, from the mid 80ies. Sure, your nominal S/N ratio tends to be lower upon measurement, even if you use low noise trannies like 2SC2240/2SA970, or such like, but this can be offset somewhat by using fully regulated power supplies with serious filtering.
Generally, why go single ended and then spend time and money making it symmterical just before the current gain stages, when you can go symmetrical straight off the bat?
All very personal views, of course.
I think I fundamentally agree with you regarding a classic SE input stage, based on a differential pair, with all the bells and whistles (cascodes, FET/BJT combinations, current mirrors) far too often cannot match a properly done fully complementary input stage on sound quality.
Not automatic, of course, but by and large, I seem to prefer fully complementary on sound basis alone, even if it is more difficult to get just right.
Also, I would add that letting in a little more current simply sounds better. By that, I mean current per trannie of say 1.1-1.4 mA, although Dan D'Agostino didn't shy away from 1.6 mA per trannie in his older series of amps, from the mid 80ies. Sure, your nominal S/N ratio tends to be lower upon measurement, even if you use low noise trannies like 2SC2240/2SA970, or such like, but this can be offset somewhat by using fully regulated power supplies with serious filtering.
Generally, why go single ended and then spend time and money making it symmterical just before the current gain stages, when you can go symmetrical straight off the bat?
All very personal views, of course.
Ah, now we cycle back to things we are pretty sure we hear, but can't say what objective measure we can use to quantify it.
Now SELF would say the "traditional" (read that his), input stage is so good you need not go further, and a simple SE VAS contributes so little to the overall signature it is good enough. My understanding of his publications of course. Well, my 1200's sound pretty darn nice; better than any of my "traditional" amps. Only four more old amps to sell on e-bay. I am keeping one Rotel 951 on my test bench as the modern protection circuit in it is a good thing on the speaker test bench where you might connect something wrong and the beguine but un-inspiring sound does not matter. I wish the Z3 power supply was not so terrible as being a small footprint amp gave me more bench space. Just not enough room for caps in there I guess. It hums right along with the music so bad you can't do LF measurements.
Now SELF would say the "traditional" (read that his), input stage is so good you need not go further, and a simple SE VAS contributes so little to the overall signature it is good enough. My understanding of his publications of course. Well, my 1200's sound pretty darn nice; better than any of my "traditional" amps. Only four more old amps to sell on e-bay. I am keeping one Rotel 951 on my test bench as the modern protection circuit in it is a good thing on the speaker test bench where you might connect something wrong and the beguine but un-inspiring sound does not matter. I wish the Z3 power supply was not so terrible as being a small footprint amp gave me more bench space. Just not enough room for caps in there I guess. It hums right along with the music so bad you can't do LF measurements.
Nigel
Re the concept of a phase splitter driven NPN output - Godfrey seems to have arrived at the same point as Linsley Hood 40 years ago with his Class A amplifier.
The only real difference I can see in Godfrey's layout is his use of an NPN input transistor where Hood used a PNP one.
If you are looking for a Class AB development of the phase splitter concept to experiment with, I suggest you investigate MC33078 and MC34080 where there is an additional "pull down" transistor driven in parallel with the lower NPN output device.
With regard to MC1530 - PNP transistors were noisy when fabricated into IC's so all NPN is unnecessary in building a discrete component version.
Re the concept of a phase splitter driven NPN output - Godfrey seems to have arrived at the same point as Linsley Hood 40 years ago with his Class A amplifier.
The only real difference I can see in Godfrey's layout is his use of an NPN input transistor where Hood used a PNP one.
If you are looking for a Class AB development of the phase splitter concept to experiment with, I suggest you investigate MC33078 and MC34080 where there is an additional "pull down" transistor driven in parallel with the lower NPN output device.
With regard to MC1530 - PNP transistors were noisy when fabricated into IC's so all NPN is unnecessary in building a discrete component version.
Not very exciting - about 3% THD @ 4W output, IIRC (I didn't save the sim). Output impedance was about 1 Ohm (again IIRC), rising above 200KHz or so. Looked like it might need a Zobel network for stability with purely capacitive loads.
No coincidence. I started with the JLH design and just changed the input stage to make an "all NPN" circuit. No pretense at greatness - it was just a "for fun" attempt at a simplest possible circuit.
Last edited:
Hi Godfrey . My estimate was that ( >1 % ) . My friend John has hidden his bucket of 3055's which he has begged me to use . Otherwise I would have been soldering it up today .
Mjona . You read my mind . MC33079 is my favoutite cheap op amp . Also JLH . My DC offset adjustment was based on the later JLH . The only problem I had was making a DC point that didn't cause hum , my solution is impractical . I feel the Godfrey circuit in H bridge would be interesting .
I came across this design many years after choosing similar ideas . To be frank if chasing the last wisp of distortion it is hard to see how one could do better . Where would extra complexity help it ?
MJR7-Mk5 Mosfet Power Amplifier
Why did I choose Mc33079 ? I listened to it with my eyes , Motorola often give circuit diagrams . The circuit can match NE5532 except on current . It is a very simple design and under the microscope beats 5532 . So much so it is like one of those annoying football teams from a lower league that any would fear to meet .
Someone said those who do not know history make the same mistakes . I would also say history sometimes had better answers . Fashion has a lot to answer for .
Talking motorcycles . Harley Davidson and Triumph were dominant . Both were seriously flawed designs . The flaws were respected by owners as a price worth paying . The Harley twin is too narrow to get best advantage from it . The Triumph hopeless on vibration also . For the life of me I can not see how they succeeded with such awful designs ? . The only thing I suppose is not going away from a reasonable recipe saved them money . It also meant spares were available cheaply with no shortages due to obsolescence . The Triumph Bonneville is the biggest pile of rubbish I know of ( far better a 1960 6T Thunderbird , almost civilized ) . A Bonneville still is the one I most want to own .
Mjona . You read my mind . MC33079 is my favoutite cheap op amp . Also JLH . My DC offset adjustment was based on the later JLH . The only problem I had was making a DC point that didn't cause hum , my solution is impractical . I feel the Godfrey circuit in H bridge would be interesting .
I came across this design many years after choosing similar ideas . To be frank if chasing the last wisp of distortion it is hard to see how one could do better . Where would extra complexity help it ?
MJR7-Mk5 Mosfet Power Amplifier
Why did I choose Mc33079 ? I listened to it with my eyes , Motorola often give circuit diagrams . The circuit can match NE5532 except on current . It is a very simple design and under the microscope beats 5532 . So much so it is like one of those annoying football teams from a lower league that any would fear to meet .
Someone said those who do not know history make the same mistakes . I would also say history sometimes had better answers . Fashion has a lot to answer for .
Talking motorcycles . Harley Davidson and Triumph were dominant . Both were seriously flawed designs . The flaws were respected by owners as a price worth paying . The Harley twin is too narrow to get best advantage from it . The Triumph hopeless on vibration also . For the life of me I can not see how they succeeded with such awful designs ? . The only thing I suppose is not going away from a reasonable recipe saved them money . It also meant spares were available cheaply with no shortages due to obsolescence . The Triumph Bonneville is the biggest pile of rubbish I know of ( far better a 1960 6T Thunderbird , almost civilized ) . A Bonneville still is the one I most want to own .
Nige, for the n-th time, forget 2N3055 and try out some BD 249C (NPN) and BD 250C (PNP)..
They will sound the pants off 2n3055/2N2955, they are not 60 or 80V but 100V, they continuous current is 25A, impulse 40A (50A if by SGS-Thomson) and they have a very reasonable price.
Well neigh indestructible, and easy monting plastic pack, too. Their only weakness is temperature, in line with 125W devices of that time - you have to cool them seriously.
Mark my words, one of the best transistors for audio, a true unsung legend, despite its relatively modest specs but uncommonly short Ton, Tstore and Toff times for their generation and time frame.
They will sound the pants off 2n3055/2N2955, they are not 60 or 80V but 100V, they continuous current is 25A, impulse 40A (50A if by SGS-Thomson) and they have a very reasonable price.
Well neigh indestructible, and easy monting plastic pack, too. Their only weakness is temperature, in line with 125W devices of that time - you have to cool them seriously.
Mark my words, one of the best transistors for audio, a true unsung legend, despite its relatively modest specs but uncommonly short Ton, Tstore and Toff times for their generation and time frame.
No coincidence. I started with the JLH design and just changed the input stage to make an "all NPN" circuit. No pretense at greatness - it was just a "for fun" attempt at a simplest possible circuit.
Another design shown by JLH has promise for an all-NPN configuration, if not necessarily a power amp: the vacuum tube "ring of three", pg 83 of Valve and Transistor Audio Amplifiers, ISBN 0750633565. It's d.c. coupled and is two cascaded common-cathodes and a cathode follower, with series feedback to the input tube.
You should have seen the transistors I turned down yesterday Dvv . The 3055 in class A is very OK . JLH wrote extensively about other deceives . The big surprise was 3055 was least trouble . Some prefer the non H version !!!
Although 3 % distortion is above the bar ( 0.1 % ) it looks easy to graft it onto another stage . The idea of having feedback that early appeals to me . Perhaps a secondary loop to get a tad better distortion .
It makes me rather proud of my 4 transistor design above . JLH was getting 0.05% up to 20 kHz using Darlintons . If you look carefully I was getting 0.05% up to 50 kHz using 4 devices . It is hard to say exactly what I was getting as the oscillator might have had additive or subtractive traits ? I recon I was in the region of what I show . It was a real struggle to get below - 50 db at 50 kHz initially .
You can forgive anyone for believing negative feedback creates high order distortion as in the graphs I give it looks that way ( instability was miles away ) . If you look at my 0R VAS distortion verses 47 R it looks to be true ( see Dvv we are are not mad to do it ) . Reducing gain seems to have effected a cure ( see graphs ) . I still suspect it was loading the previous stage which caused it . If you take gain to be 400 and 8mA than give z = ( 25/8 ) 400 = 1250 R ( BC547C ) . The resistor feeding it 820 R . That is not too bad . Some would even have 1K feeding 300R in . When I add 47 that is then off the scale . 16 R = 6K4 .
DF 96 said if I prefer 16R it is because it weakens the grip of the amp at low frequencies . Doubtless true . See everyone , I thought in my amp to have covered that ( 820 R feeds > 820 R should work ) and would use 0R to the BC547 emitter . No I hadn't , the scope didn't lie . I would imagine 4R7 to already be a great compromise. Doesn't it make you think to see the horrid little spikes in my 0R graph . OK very low level and 50 kHz . Still , it isn't very nice . Surely if we look at - 80db they will still be there ? At - 80 db I can be happier .
I did try a Darlington VAS . It worked great but did not sound as good ( less open ) . I tried a complimentary feedback pair . It was better . The best was just a high gain transistor . I didn't try a cascode . From my experience with them I would assume it to be excellent ( BC547 , MPSA 42 ) . My design can be up scaled and bias reduced ( 100 ma is fine ) . As voltage increases so distortion is reduces to an extent ( stretching the VAS curve ) . I would guess even 100 W to be possible and far less than 1% distortion ( +/- 50 V , 2n 5551 , 5401 , 2SA 1085 2SD756 ) .
Dvv and DF96 said about fiddling . Nearly all amps use a VAS like this . None as far as I know will do anything wild if a VAS emitter resistor is fitted . I suspect 90% would sound better for 4R7 ? For some amps it would be joyfulness discovered . I will suggest if transistor sound ever existed it was this that caused most of it ? Slewing is a byproduct . A bit like using a shotgun when it needed one well placed bullet . JLH has from memory only about 100 uA to drive the VAS , not ideal . That was typical then .
Although 3 % distortion is above the bar ( 0.1 % ) it looks easy to graft it onto another stage . The idea of having feedback that early appeals to me . Perhaps a secondary loop to get a tad better distortion .
It makes me rather proud of my 4 transistor design above . JLH was getting 0.05% up to 20 kHz using Darlintons . If you look carefully I was getting 0.05% up to 50 kHz using 4 devices . It is hard to say exactly what I was getting as the oscillator might have had additive or subtractive traits ? I recon I was in the region of what I show . It was a real struggle to get below - 50 db at 50 kHz initially .
You can forgive anyone for believing negative feedback creates high order distortion as in the graphs I give it looks that way ( instability was miles away ) . If you look at my 0R VAS distortion verses 47 R it looks to be true ( see Dvv we are are not mad to do it ) . Reducing gain seems to have effected a cure ( see graphs ) . I still suspect it was loading the previous stage which caused it . If you take gain to be 400 and 8mA than give z = ( 25/8 ) 400 = 1250 R ( BC547C ) . The resistor feeding it 820 R . That is not too bad . Some would even have 1K feeding 300R in . When I add 47 that is then off the scale . 16 R = 6K4 .
DF 96 said if I prefer 16R it is because it weakens the grip of the amp at low frequencies . Doubtless true . See everyone , I thought in my amp to have covered that ( 820 R feeds > 820 R should work ) and would use 0R to the BC547 emitter . No I hadn't , the scope didn't lie . I would imagine 4R7 to already be a great compromise. Doesn't it make you think to see the horrid little spikes in my 0R graph . OK very low level and 50 kHz . Still , it isn't very nice . Surely if we look at - 80db they will still be there ? At - 80 db I can be happier .
I did try a Darlington VAS . It worked great but did not sound as good ( less open ) . I tried a complimentary feedback pair . It was better . The best was just a high gain transistor . I didn't try a cascode . From my experience with them I would assume it to be excellent ( BC547 , MPSA 42 ) . My design can be up scaled and bias reduced ( 100 ma is fine ) . As voltage increases so distortion is reduces to an extent ( stretching the VAS curve ) . I would guess even 100 W to be possible and far less than 1% distortion ( +/- 50 V , 2n 5551 , 5401 , 2SA 1085 2SD756 ) .
Dvv and DF96 said about fiddling . Nearly all amps use a VAS like this . None as far as I know will do anything wild if a VAS emitter resistor is fitted . I suspect 90% would sound better for 4R7 ? For some amps it would be joyfulness discovered . I will suggest if transistor sound ever existed it was this that caused most of it ? Slewing is a byproduct . A bit like using a shotgun when it needed one well placed bullet . JLH has from memory only about 100 uA to drive the VAS , not ideal . That was typical then .
MOST power transistors do well in pure class A. However, all don't do equally well
I feel that's not easy no matter which design one uses.
Which is why I believe in more local and less global NFB. Distortion, I feel, is best dealt with locally so that it doesn'r happen excessively, rather than let it happen and cure it after the amp is already "sick". Prevent rather than cure.
Darlington transistors by and large tend to sound warmer than usual, but at the price of some detail. I have yet to hear an amp using Darlington trannies in its output stage which will match a classic say dual transistor pair for clarity and detail.
Ah, but for the one bullet solution to work, one needs to be a good shot.
Personally, I just love degeneration resistors in the input stages and never fail to let the VAS work into relatively high value resistors. In my view, they offer advantages only with no setbacks. They make sure the VAS is always working into a stable, ground referenced load, which reduces the influence of the amp load variations. It also lets you program the gain stage to almost whatever, making sure your stage gain is exactly what you want and what it needs to be. Limiting gain to something sane also reduces distortion, which in turn reduces the need for global NFB, which reduces more or less odd behavior upstream (say, from 50 kHz upwards).
Hi Dvv . They say a little knowledge is a dangerous thing . DF 96 raised two very good criticisms of VAS degeneration . Douglas Self another . In a nutshell I am going to claim transistor sound has always been the VAS if talking about competent designs . Leak and Quad in my opinion never had obvious transistor sound ( the list will be large if I add to it , Armstrong ..... ) . What they probably had was some hint of TID on challenging material. Crossover distortion has no excuse , nor aggressive protection circuits . Designers who get that wrong should resign and run old folks homes ( no joke , that seems to happen , yikes , take me in my sleep , can you imagine the designer of the Amstrad IC 2000 as your care master ? Maybe the IC 2000 wasn't designed . Maybe it mutated from a liver fluke ? I tried to buy one on e-bay , no luck . The old Teleton GA 202 came up ) .
Standard thinking is that degenerating the VAS can only do harm . To quote D S it is not feedback in the sense it would be if a valve circuit . DF 96 says it reduces the grip of the amplifier at low frequencies and raises the VAS collector impedance . Thought about that a bit . The impedance of the CCS is preferably infinite . The output stage is a very nasty load . As D S says making it low impedance will help . That is analogous to an amplifier driving a speaker . So that must logically be right .
Here is my conjecture again stimulated by D S . He gave me to understand TID and IM distortion will always be manifested as harmonic distortion . My graphs seem to prove that . Even committing multiple sins I got my harmonic distortion down ( remember this is not instability creeping in ) .
So what is transistor sound ? Perhaps it is unquestionably seeking transconductance between input stage and the VAS base ? That horrible distortion like a bad triangle wave is not irrelevant . The cure ? Make sure the impedance's are vastly different . Whatever you do , do not match them . If you do then bandwidth limit to take the distortion away . The complementary feedback pair VAS seems ideal . It will give you almost a single transistor with no great requirement for current . One warning . It is not uncommon to have one transistor doing all the work in CfbP . I had absolutely no stability problems when I did it . Somehow I didn't quite get an improvement over the VAS resistor . A Russian guy ( ? ) made a special VAS that looks like a CfbP . I wish I had tried that .
D S makes another good point . The capacitance inside a transistor is not the best . Use a low Cob device and add externally . The upshot is a reduction in second harmonic distortion . On the other hand a bit of second is not so horrible .
That is what the debates we get into are about . Making us all good shots .
One thing someone said . If we trade away loop feedback and see no distortion difference then the distortion was covered by noise ( it has to be true ) . Well I hate to state the obvious . And how does something below noise intrude ? We assume a reasonable amp when saying this . That is sort of saying , dam the - 100 dB distortion isn't up to snuff .
Standard thinking is that degenerating the VAS can only do harm . To quote D S it is not feedback in the sense it would be if a valve circuit . DF 96 says it reduces the grip of the amplifier at low frequencies and raises the VAS collector impedance . Thought about that a bit . The impedance of the CCS is preferably infinite . The output stage is a very nasty load . As D S says making it low impedance will help . That is analogous to an amplifier driving a speaker . So that must logically be right .
Here is my conjecture again stimulated by D S . He gave me to understand TID and IM distortion will always be manifested as harmonic distortion . My graphs seem to prove that . Even committing multiple sins I got my harmonic distortion down ( remember this is not instability creeping in ) .
So what is transistor sound ? Perhaps it is unquestionably seeking transconductance between input stage and the VAS base ? That horrible distortion like a bad triangle wave is not irrelevant . The cure ? Make sure the impedance's are vastly different . Whatever you do , do not match them . If you do then bandwidth limit to take the distortion away . The complementary feedback pair VAS seems ideal . It will give you almost a single transistor with no great requirement for current . One warning . It is not uncommon to have one transistor doing all the work in CfbP . I had absolutely no stability problems when I did it . Somehow I didn't quite get an improvement over the VAS resistor . A Russian guy ( ? ) made a special VAS that looks like a CfbP . I wish I had tried that .
D S makes another good point . The capacitance inside a transistor is not the best . Use a low Cob device and add externally . The upshot is a reduction in second harmonic distortion . On the other hand a bit of second is not so horrible .
That is what the debates we get into are about . Making us all good shots .
One thing someone said . If we trade away loop feedback and see no distortion difference then the distortion was covered by noise ( it has to be true ) . Well I hate to state the obvious . And how does something below noise intrude ? We assume a reasonable amp when saying this . That is sort of saying , dam the - 100 dB distortion isn't up to snuff .
No matter who says what, if it sounds good to me, I'll use it. How many rules I break in the process is of no consequence.
Nige, I have heard I can't even remember how many amps using a single transistor for its VAS, with Miller capacitor, of course, which sometimes had rather low values (e.g. Sansui AU-7800, 22 pF) and sometimes unbelievably high values (a German amp from the 70ies, 220 pF).
The really odd thing for me at least is that amps which I would think of as overcompensated, as when you have 0.5 mA charging a 150 pF Miller capacitor, giving you a wild slew rate of 3.3V/uS, always managed to sound sort of warm and cuddly, even if imprecise.
Nige, I have heard I can't even remember how many amps using a single transistor for its VAS, with Miller capacitor, of course, which sometimes had rather low values (e.g. Sansui AU-7800, 22 pF) and sometimes unbelievably high values (a German amp from the 70ies, 220 pF).
The really odd thing for me at least is that amps which I would think of as overcompensated, as when you have 0.5 mA charging a 150 pF Miller capacitor, giving you a wild slew rate of 3.3V/uS, always managed to sound sort of warm and cuddly, even if imprecise.
An open mind and some common sense is what it takes .
I always say for 4 things that prove a rule there is 1 that disproves it . When you repair things for a living you learn that . I did for 23 years and still do for friends now . Often I find people come to the right answers for the wrong reasons . I am certain even the most educated fall into that trap . It is only when repairing supposedly bad designs that work just fine that a pattern emerges . The Triumph twin motorcycle is a truly horrible design with one or two exceptions . The combustion chamber shape is OK and stroke and timings ( valve and sparks ) . The crank flexes badly . The barrels are seldom square to the crank ( 6 x <1/1000 inch within square was set to one side for racing machines ) . The strange thing is curing the design errors brought new problems . Stiffening the crank made already unbearable vibration even worse ( yikes ) . I said one or two exceptions . The exceptions made the Bonneville a piece of pure magic . I love the oil pump , so simple and excellent . Can be repaired at the road side with improvised tools . Never it's fault if it does , fit a filter and probably never will happen . The pump looks as if a child designed it . I works great .
I am so pleased people who design headlight of cars get it right . My old Rover Montego ( 5 litres / 100 kM , 100 % reliable over 300 000 kM , rust bucket ) I think only ever had 3 bulbs ? I installed a proper transformer in my girlfriends kitchen to try to get the 20 W lights reliable ( and did all the other things one should , good flow of air ). this was to replace the standard switch mode devices . Even at 10 V rms they only last a month or so ( 80 VA 230 V to 9 V run at 40 W and 242 V = 10 V rms ) . That's no 2000 hours and that's no 12 V , come on guys stop lying . I think she needs headlights next ? I might take the light switch out as I don't think it ever gets used ? Now who can explain that ? The waveform shape is OK ( 4 % THD ) and there are no bad surges .
I always say for 4 things that prove a rule there is 1 that disproves it . When you repair things for a living you learn that . I did for 23 years and still do for friends now . Often I find people come to the right answers for the wrong reasons . I am certain even the most educated fall into that trap . It is only when repairing supposedly bad designs that work just fine that a pattern emerges . The Triumph twin motorcycle is a truly horrible design with one or two exceptions . The combustion chamber shape is OK and stroke and timings ( valve and sparks ) . The crank flexes badly . The barrels are seldom square to the crank ( 6 x <1/1000 inch within square was set to one side for racing machines ) . The strange thing is curing the design errors brought new problems . Stiffening the crank made already unbearable vibration even worse ( yikes ) . I said one or two exceptions . The exceptions made the Bonneville a piece of pure magic . I love the oil pump , so simple and excellent . Can be repaired at the road side with improvised tools . Never it's fault if it does , fit a filter and probably never will happen . The pump looks as if a child designed it . I works great .
I am so pleased people who design headlight of cars get it right . My old Rover Montego ( 5 litres / 100 kM , 100 % reliable over 300 000 kM , rust bucket ) I think only ever had 3 bulbs ? I installed a proper transformer in my girlfriends kitchen to try to get the 20 W lights reliable ( and did all the other things one should , good flow of air ). this was to replace the standard switch mode devices . Even at 10 V rms they only last a month or so ( 80 VA 230 V to 9 V run at 40 W and 242 V = 10 V rms ) . That's no 2000 hours and that's no 12 V , come on guys stop lying . I think she needs headlights next ? I might take the light switch out as I don't think it ever gets used ? Now who can explain that ? The waveform shape is OK ( 4 % THD ) and there are no bad surges .
I thought the BBC approach to engineering might be enjoyable . I do get the impression BBC were sponsors of the belief in TID ! Thinking about it they would be .
What I love is the unthinkable . Scientists who happily accept listening tests . More on the bases of testing the reality of it I feel ? Make it bad enough to hear , give it some numbers . The rigorous testing of BBC digital transmission links made one forget it was digital . Stone-age BBC digital better than whizkid/fast-buck CD .
http://www.keith-snook.info/Wireles...s-World-1978/A new distortion measurement.pdf
What I love is the unthinkable . Scientists who happily accept listening tests . More on the bases of testing the reality of it I feel ? Make it bad enough to hear , give it some numbers . The rigorous testing of BBC digital transmission links made one forget it was digital . Stone-age BBC digital better than whizkid/fast-buck CD .
http://www.keith-snook.info/Wireles...s-World-1978/A new distortion measurement.pdf
The simple truth is that conventional specs tell you little about how it subjectively sounds, because they're not measuring what the ear is sensitive to, finds objectionable - merely what is easy and obvious to measure. Until someone finally takes this seriously, and nails the distortion factors that irritate, add greyness to the sound, the enthusiasts will keep going round and round and round in circles, discussing such things ad nauseum ..
There's also this attempt at correlating objectionable sound with a number:
Perception
I propose that, rather than replacing THD with a single weighted-harmonic measurement. it should be replaced with TWO numbers - one that's heavily weighted toward the low-order harmonics (second, third), and the other weighted toward the high-order harmonics (with the ear's rising sensitivity with harmonic number, this one might need little weighting). The first would be "amount of tube sound" and the second would be "amount of transistor sound."
Perception
I propose that, rather than replacing THD with a single weighted-harmonic measurement. it should be replaced with TWO numbers - one that's heavily weighted toward the low-order harmonics (second, third), and the other weighted toward the high-order harmonics (with the ear's rising sensitivity with harmonic number, this one might need little weighting). The first would be "amount of tube sound" and the second would be "amount of transistor sound."
I would defend the BBC and say they often got things right . My feeling is transmitting is plagued with problems . Multipath , hiss , clipping , transmitting over distance ( landlines and via the aerial ) . Whilst multipath is mostly a user problem working with FM must make people aware of how to get the perfect window of performance . That thinking will get ground in . How an amplifier works is much the same . The difference is if a broadcast engineer maybe more searching questions get asked ? For example if it was a transmitter would it be ideal ( it's envelope of performance ) . I know the analogy brakes down because the FM transmitter is a very different animal . The thinking may not be so different . As I say it is no surprise to me that the BBC would understand TID , I am sure it's equivalent is easily found in mildly wrong FM , not least the Nicam 13 bit system .
After all terrible things that have to be done to FM to transmit it is remarkable how good it sounds . That's dedication ( and a miracle ) .
After all terrible things that have to be done to FM to transmit it is remarkable how good it sounds . That's dedication ( and a miracle ) .
The late John Linsley-Hood is one who has defined the problems you refer to - most recently in his book "Valve&Transistor Audio Amplifiers". This covers a host of other areas of audio design in addition to that of immediate interest.
It is an interesting read and I would not be without it.
Michael J.
Best thing about the book is that it's a story. Especially for a novice, seeing how a circuit works isn't the same as knowing why it's used, which is generally a matter of history.
It's probably a very English history, and partisan in several respects, but illuminating all the same. Not up to date though, obviously.
If I remember his story, life was a struggle for transistor power output stages until reasonably matched NPN, PNP pairs became available. Can't much remember why...problem of symmetrical drive for PP I guess.
- Status
- Not open for further replies.