Here's my copy from the RCA designs book. Harmon Kardon Citation 12 sems to be one step before it in the evolution. Maybe Harmon came first.
Mjona. That's a very unusual version! Like the Sanyo STK types. I think some called them thick film circuits as they were made in the fashon of output devices with the chip driver close inside. It shows how a bootsrap would not be practical in a small package and uses a CCS instead as VAS load. I have to specualte that my version 680R = 1 mA is not complete balance. Maybe like Naim RCA wanted that, although not as drastic as Naim.
Your version is very interesting as it has the VAS resistor R12 27R ( and no Cdom ) . I have always advocated this as a possible tweak. If wanting to run with the horses and the hounds one could take the 27R in paralell with 470 nF. This would increase the open loop gain just when most amps need it ( >10 kHz ). This would allow the input stage to see a very high input impedance at real music frequencies into the VAS. Lets say 3000 ohms at 1 kHz which sits nicely with 680/820/1000R before it. This will reduce to lets than 10R > 10 kHz. The HH1200 uses this trick. One has to set the Nyquist stability point using Cdom of the VAS to suit this ( >39pF ? ). The theory is real music seldom will give you 1 watt at 10 kHz, we could get more than we lose as the slewing issues shouldn't happen. As other designers completely reject this, one has to be ones own judge. They talk as if some mysterious demon demands this. Perhaps they are right. Some say this is nonesense and cannot have local feedback ot the VAS applied this way. That never was my point. It's just accepting a I to V stage being easier to drive must change something. RCA also. Could be it also got rid of Cdom by reducing open loop gain.
The usual way to massage the slew rates is to increase the input stage current then reset open loop gain by adding emmiter resistors. This restores the original open loop gain and stability and increases linearity. Then replacing Naims 1K 12 K set up with a current mirror to gain the last drop of current into the VAS. Then tweak the VAS collector to base capacitor to retain stability. In doing that you won't have a Naim sound.
NPN or PNP input stages are not really an issue ( one could say my RCA is most certinaly not the Naim if stretching this point and the VAS CCS ). It often is the VAS transistor that dictates what's best. The input transistors can be of lower voltage than the VAS as they sit mostly at half the total rail voltages all the time. A PNP input stage usually is slightly quieter and a NPN VAS might have higher gain ( good ). I dare say a reversed Naim layout could be better, more open sounding perhaps. Still all NPN outputs as my RCA type.
I never found the special RCA version I wanted to show. It had extra feedback routes. Avondale Audio might be the source of the RCA story as Naim were not best pleased he grew his business on their sucess. Maybe Les was trying to tell Naim none of us can claim to have invented this layout ( uA 741 which started life in 1963 I feel was it's origin, sold to the public about 5 years later ). Here is the other very interesting op amp of that era. Very fast even now and looks very different to how we do things today. I often wonder if it would have been the better thing to clone. BTW, Les is a very nice guy.
http://www.analog-innovations.com/SED/MC1530-TeachingExercise.pdf
Many British amps had the collector resistor of the long tail pair input stage higher than the VAS input impedance. I think for THD measurements this worked well. TID might find it's prime suspect here. What people belived is these stages are purely current amplifiers. The VAS being more rightly called a I to V converter. The silly science that grew out of this was slew rates. Like 0 to 60 MPH times for cars these seem to remain what people like to talk about. I showed a Hitachi design with double VAS. This seems to have come directly from op amps circa 1968 to 1972. The advantage of double VAS is the slewing is symetrical in both sourcing and sinking of current. Sometimes clones of RCA need four times the current they should to work if talking higher slew rates ( selling points if using honest figures showing the up stroke ). The reason is single VAS's are like a one legged cyclist who needs to assist the up stroke of the peddle with just one leg. It's not ideal. The battles with stability to get a paper spec is hard work. Sometimes I felt amplifers sounded worse by doing this. The Naim was sensible.
Your version is very interesting as it has the VAS resistor R12 27R ( and no Cdom ) . I have always advocated this as a possible tweak. If wanting to run with the horses and the hounds one could take the 27R in paralell with 470 nF. This would increase the open loop gain just when most amps need it ( >10 kHz ). This would allow the input stage to see a very high input impedance at real music frequencies into the VAS. Lets say 3000 ohms at 1 kHz which sits nicely with 680/820/1000R before it. This will reduce to lets than 10R > 10 kHz. The HH1200 uses this trick. One has to set the Nyquist stability point using Cdom of the VAS to suit this ( >39pF ? ). The theory is real music seldom will give you 1 watt at 10 kHz, we could get more than we lose as the slewing issues shouldn't happen. As other designers completely reject this, one has to be ones own judge. They talk as if some mysterious demon demands this. Perhaps they are right. Some say this is nonesense and cannot have local feedback ot the VAS applied this way. That never was my point. It's just accepting a I to V stage being easier to drive must change something. RCA also. Could be it also got rid of Cdom by reducing open loop gain.
The usual way to massage the slew rates is to increase the input stage current then reset open loop gain by adding emmiter resistors. This restores the original open loop gain and stability and increases linearity. Then replacing Naims 1K 12 K set up with a current mirror to gain the last drop of current into the VAS. Then tweak the VAS collector to base capacitor to retain stability. In doing that you won't have a Naim sound.
NPN or PNP input stages are not really an issue ( one could say my RCA is most certinaly not the Naim if stretching this point and the VAS CCS ). It often is the VAS transistor that dictates what's best. The input transistors can be of lower voltage than the VAS as they sit mostly at half the total rail voltages all the time. A PNP input stage usually is slightly quieter and a NPN VAS might have higher gain ( good ). I dare say a reversed Naim layout could be better, more open sounding perhaps. Still all NPN outputs as my RCA type.
I never found the special RCA version I wanted to show. It had extra feedback routes. Avondale Audio might be the source of the RCA story as Naim were not best pleased he grew his business on their sucess. Maybe Les was trying to tell Naim none of us can claim to have invented this layout ( uA 741 which started life in 1963 I feel was it's origin, sold to the public about 5 years later ). Here is the other very interesting op amp of that era. Very fast even now and looks very different to how we do things today. I often wonder if it would have been the better thing to clone. BTW, Les is a very nice guy.
http://www.analog-innovations.com/SED/MC1530-TeachingExercise.pdf
Many British amps had the collector resistor of the long tail pair input stage higher than the VAS input impedance. I think for THD measurements this worked well. TID might find it's prime suspect here. What people belived is these stages are purely current amplifiers. The VAS being more rightly called a I to V converter. The silly science that grew out of this was slew rates. Like 0 to 60 MPH times for cars these seem to remain what people like to talk about. I showed a Hitachi design with double VAS. This seems to have come directly from op amps circa 1968 to 1972. The advantage of double VAS is the slewing is symetrical in both sourcing and sinking of current. Sometimes clones of RCA need four times the current they should to work if talking higher slew rates ( selling points if using honest figures showing the up stroke ). The reason is single VAS's are like a one legged cyclist who needs to assist the up stroke of the peddle with just one leg. It's not ideal. The battles with stability to get a paper spec is hard work. Sometimes I felt amplifers sounded worse by doing this. The Naim was sensible.
The time line I see is uA741 1968, Sinclair Z30/50 1969, HK Citation 12 1971 ( guess ) Bob Stewart/RCA 1973 and Naim slightly after. The only amp I saw with some similartiy before that was Franch make Gogny of 1964 that has a long tail pair input and a mild resemblance to the RCA chip amp, it drives 1 ohms for a ribbon tweeter. Sinclair's 1968 amp the Neoteric was typical British Leak style of amp. In one year they changed the whole picture, it was a very cheap amplifier. I feel Douglas Self should claim his blameless amp started somewhere with the Z50. It's problem was using very small transistors usually used as drivers as outputs. This was not too bad when 10 watts. It would not give more. Bob's advice to Julian was to find propper outputs to beef up the design and then think some more ( that's what we see ). The A+R A60 was closer to the Z50 it should be said although more complex in the house keeping circuits.
People say the HK C12 is a special amp. It is a more primitive design. The Crimson was the other great UK design. Many thought it better than Naim. A mix of NAP160 and Quad 303. The C12 went on to use MOSFET's. Could it be Julian knew this? I seem to remember the Beatle not known as a drummer has a HK of this era.
Thias should be useful and at least is from the horses mouth. From the RCA designs also.
RCA data sheet file #647
http://www.firstwatt.com/pdf/art_mos_citation.pdf
People say the HK C12 is a special amp. It is a more primitive design. The Crimson was the other great UK design. Many thought it better than Naim. A mix of NAP160 and Quad 303. The C12 went on to use MOSFET's. Could it be Julian knew this? I seem to remember the Beatle not known as a drummer has a HK of this era.
Thias should be useful and at least is from the horses mouth. From the RCA designs also.
RCA data sheet file #647
http://www.firstwatt.com/pdf/art_mos_citation.pdf
Last edited:
According to web sources the components are mounted on a small board under a removable top cover.
Taking this off would reveal the transistor types which could help servicing the board if required.
There is a chart plotting slew rate against compensation capacitor values - a matter of selecting and fitting the chosen value on the board.
I see Modules can still be purchased online.
The Naim amplifiers were basically a copy of the 1972 75W design published by John Linsley-hood in Hi-Fi News, upgraded by the replacement of the boostrap by a current source. The A&R A60 was the same, as were many Japanese products. The pcb was laid out (Bishop Graphics, no CAD in those days) by his then girlfriend (Dianne from memory) who was a lovely lady. Got pregnant later and Julian ditched her ...he was a total prick.
JV forgot to read the last bit of J.L-H's article noting the likelihood of instability when loaded with small C such as you get from the figure-of-eight zip cable commonly used as speaker conns. The inclusion of a small choke cured that but JV omitted it
. After a series of meltdowns he specified that his amps must be partnered with his special (and grossly overpriced) cable. The 2 r/c networks in the driver bases could be tuned to allow for slow drivers. With newer devices these would probably not be necessary. The only thing which made the Naim amps outperform their similar opposition was a hefty power supply, fully regulated in the case of the 70W version.
JV forgot to read the last bit of J.L-H's article noting the likelihood of instability when loaded with small C such as you get from the figure-of-eight zip cable commonly used as speaker conns. The inclusion of a small choke cured that but JV omitted it
. After a series of meltdowns he specified that his amps must be partnered with his special (and grossly overpriced) cable. The 2 r/c networks in the driver bases could be tuned to allow for slow drivers. With newer devices these would probably not be necessary. The only thing which made the Naim amps outperform their similar opposition was a hefty power supply, fully regulated in the case of the 70W version.
I don't think it helps much to repeat the rumour and speculation on the lineage of the design, already seen in the copious off-topic of the Ebay clone thread. As I see it, the design is entirely generic of its time and borrows more from successful IC opamps and hybrids as suggested earlier. It's even the same topology as Douglas Self's generic model amplifier - albeit with quasi-complementary output stage. Being well known topology now, there are analyses of operation and performance in other publications too. Self's Blameless model of course, is all about the refinements applied to that basic design such as a current mirror, beta enhanced VAS, decoupling caps, speed-up capacitance etc. with the sole purpose of reducing its distortion.
Conversely, had JV not tinkered with coaxing nice harmonic distortion from the generic design instead, Naim would never have had a domestic product worth more that any competent 1970s amplifier - probably leaving him with only the pro. sound and lighting business. That much is an amusing dichotomy and an insight to what UK Hifi industry figures of the day, such Ivor Tiefenbrun, looked for in sound quality 😉
Conversely, had JV not tinkered with coaxing nice harmonic distortion from the generic design instead, Naim would never have had a domestic product worth more that any competent 1970s amplifier - probably leaving him with only the pro. sound and lighting business. That much is an amusing dichotomy and an insight to what UK Hifi industry figures of the day, such Ivor Tiefenbrun, looked for in sound quality 😉
Attachments
I think this is very interesting; it is a warts and all history of the development of a marvellous UK product, a messy, half hearted beginning which slowly evolved into something very significant in the audio industry. People who found these business are driven to do all sorts of drastic actions to keep their cash flow moving, and clearly JV was about to achieve impressively. The tragedy is his early death of cancer under 60, rather like Steve Jobs; I have sympathy for any driven individual who dies early. So many over-overachievers behave this way, as they suffer a boding presentiment even from childhood.
Can't help thinking JV made a mistake dropping JLH's bootstrap......
HD
Can't help thinking JV made a mistake dropping JLH's bootstrap......
HD
I think the fact most people reject is all of these amplifiers are versions of op amps. When the LM1875 we have an op amp that can do anything. I only realised recently that these op amps if driving a high resistance load are almost perfect devices as long as not used for RIAA ( with a Shure M44-7 we might ). From this I might build a preamp that can drive speakers if I choose. It will be a Naim like device with a big PSU out of need. The ECC83 or 33 valve was an early op amp or helped when building one. From about 1935 op amps were known. Harold Leak made to my ears the first transistor amplifier that sound just like the valve versions he made before. With modern speakers perhaps the valve ones were better. Leak pointed out that getting rid of the output tranformers and shipping weight made it possible to say better value. Even so only 10% was saved. Leak eventually went to an op amp long tail pair version about the time of the HK Citation 12, alas Mr Leak no longer owned the company then ( testing was reduced to 1 in 10 units , Source Alan Tisdale now dead, He told me how Mr Leak cried when Rank bought him out ). The first thing Julian told me was he had the identical test everything and every finished thing routine as Leak. He jumped into the vacuum Rank made.
If anyone wants to know about bootsrapped CCS just ask. I suspect 70% to 90% is ideal for the positive feedback ( that's all there is to know really, above 92% add a series resistor to avoid nasties ). The capacitor can be smaller than you might think allowing a film type. As transistors became cheaper a transistor CCS became realistic. The op amp input also became cheaper. Some amplifers were more stable when they did not have a long tail pair input. This has been over played by some people and forgets many VAS devices were high capaitance before the modern devices ( About 1967 ). The same ones who say the inverting input sounds best. With my LM1875 pre amp it might.
If anyone wants to know about bootsrapped CCS just ask. I suspect 70% to 90% is ideal for the positive feedback ( that's all there is to know really, above 92% add a series resistor to avoid nasties ). The capacitor can be smaller than you might think allowing a film type. As transistors became cheaper a transistor CCS became realistic. The op amp input also became cheaper. Some amplifers were more stable when they did not have a long tail pair input. This has been over played by some people and forgets many VAS devices were high capaitance before the modern devices ( About 1967 ). The same ones who say the inverting input sounds best. With my LM1875 pre amp it might.
Both Leak and Vereker caused much jealousy. Vereker said to have used an RCA design and Leak the Tobey Dinsdale design. The big fact that most seemed to forget was these were published designs and if not for copying then what was the idea of them. Leak made a very good statement, in the same way he might have forgotten the Tobey-Dinsdale advancement they seemed to have forgotten Mr H C Lin ( as best I know that's true, there might be a reference I don't know of ). That is a very important point as the H C Lin ideas " still " have considerable merit ( ESP El Cheapo is a version I guess ), Mr JLH showed us that in his wonderful designs the H C Lin ideas could be improved. On another thread someone was saying the modern amps can not be after H C Lin due to feedback arrangements and number of stages. One brave soul said maybe Mr Lin had newer designs. You bet your life he did. He was the father of op amps. If you do nothing else read his history and declassification. It's the mirror image story of Area 51, we are so lucky op amps failed in the theatre of war.
Oral History Lin Index RCA Germanium Transistors Audio
What the long tail pair versions of amps allowed was the feedback and input signal to arrive at opposite sides of the pair. As we have an emmiter follower driving a common base stage the disadvantages are small in adding an extra transistor. The harder to drive common base working with the doesn't give a damn follower. They swap functions as one can see in a differential pair, I choose to freeze time in my snap shot.
If we feed the input signal to the inverting side of the long tail pair we minic the older style amps as the feedback signal and input sinal sum at the transistor base. The advantage of this is we retain the 0V DC reference point with virtually no second thought as to how. Critics don't like this idea and will agrue one input transistor is best . JLH even showed that the DC point can be stable if no long tail pair used. A Naim amplifer could be converted to single transistor. I suspect it might be even better if it was as the 12K 1K trick would not be needed. The JLH DC point idea would be useful if so.
If anyone could have taken Naim's crown away the class AB JLH could, they were miles better than Audiolabs etc. The measured well and sounded more than wonderful, like food that makes us eat more. The Audiolab is fussy about speakers to my ears. BC1's and clones work best. Naim is good with BC1 also.
Oral History Lin Index RCA Germanium Transistors Audio
What the long tail pair versions of amps allowed was the feedback and input signal to arrive at opposite sides of the pair. As we have an emmiter follower driving a common base stage the disadvantages are small in adding an extra transistor. The harder to drive common base working with the doesn't give a damn follower. They swap functions as one can see in a differential pair, I choose to freeze time in my snap shot.
If we feed the input signal to the inverting side of the long tail pair we minic the older style amps as the feedback signal and input sinal sum at the transistor base. The advantage of this is we retain the 0V DC reference point with virtually no second thought as to how. Critics don't like this idea and will agrue one input transistor is best . JLH even showed that the DC point can be stable if no long tail pair used. A Naim amplifer could be converted to single transistor. I suspect it might be even better if it was as the 12K 1K trick would not be needed. The JLH DC point idea would be useful if so.
If anyone could have taken Naim's crown away the class AB JLH could, they were miles better than Audiolabs etc. The measured well and sounded more than wonderful, like food that makes us eat more. The Audiolab is fussy about speakers to my ears. BC1's and clones work best. Naim is good with BC1 also.
Member
Joined 2009
Paid Member
Very interesting!
As I said earlier, I have a lot of joy from cloning these amps because of the "industrial archeology" that it affords.
As I said earlier, I have a lot of joy from cloning these amps because of the "industrial archeology" that it affords.
Here is the nearest thing to a modern amplifer from 1969.
http://www.keith-snook.info/wireles... Amplifer - Ian Hardcastle and Basil Lane.pdf
Here is the Sinclair Z30/50 which has a better spec.
http://media.soundonsound.com/images/forum/web_telia_com_~u31641623_Sinclair20Z30,20Z50.jpg.png
And the Gogny from 1967. Note the use of a 2N3055 to drive a 2N3055 ( the compound gain could still be very low at this current, less than 100 ). This amplifier gives 50 watts into 1R ! It has the more old fashioned feedback type of the 1957 H C Lin design ( the best some say ). If wanting to build this use a current source in place of R4. It might be a bit better than you might think. The 4 diode bias is intersting and could be adapted for the RCA op amp.
http://i.imgur.com/GTtvsHY.jpg
http://www.keith-snook.info/wireles... Amplifer - Ian Hardcastle and Basil Lane.pdf
Here is the Sinclair Z30/50 which has a better spec.
http://media.soundonsound.com/images/forum/web_telia_com_~u31641623_Sinclair20Z30,20Z50.jpg.png
And the Gogny from 1967. Note the use of a 2N3055 to drive a 2N3055 ( the compound gain could still be very low at this current, less than 100 ). This amplifier gives 50 watts into 1R ! It has the more old fashioned feedback type of the 1957 H C Lin design ( the best some say ). If wanting to build this use a current source in place of R4. It might be a bit better than you might think. The 4 diode bias is intersting and could be adapted for the RCA op amp.
http://i.imgur.com/GTtvsHY.jpg
Don't forget Tobey and Dinsdale. I built a silicon version (2n1701s or 2N1485s, not sure) with a positive supply when the design came out (1961?). I was impressed, but unable to afford the rest of the system to complement it. I later modified it with free sample Solitron 2N3055Es. And then the 2N2955 came along and quasi bit the dust for me then.
I always assumed the Leak amps were much the same. The Tobey and Dinsdale phono stage inspired me to try something similar. As you say they do belong in the Naim story. I think KJ in Wigmore Street London told Julian the NAP160 ( ? ) was nothing more than a Leak design with a big power supply. He wasn't best pleased. I imagine if the story if right would mean the last Leak amplifiers before the Rotel ( Roland )designs. The Leak 2200 was a real dogs breakfast of a design. It's nothing like anything before or after although sort of singing from the same song sheet circa 1975. I think that completes the story as best I know it. Never loose sight of the Sinclair in the story. More so than most it set new standards which alas included blowing up. Very fast, very weak. When fitted with the TIP3055 it worked well. The TIP was a 3 MHz device which was fast at the time. I owned A+R A60 SNo 301 bought direct from them ( my first ever cheque ). I remember them saying with help from Bob Stewart they got it to be relaible ( very ). It was a Sinclair and they made no bones about it.
https://www.vinylengine.com/turntable_forum/download/file.php?id=458
https://www.vinylengine.com/turntable_forum/download/file.php?id=458
As you can see the Sinclair is just a memory in the A60. The bit they kept which is so unusual is the resistor bias with current source heat sensor. The Baxendale diode as in the Naim is there. The hand of Bob Stewart I feel.
He converted his output stage from quasi complementary to full complementary symmetry. He used a colloquialism apparently exclusively English.
Yes, I supposed that a transition from quasi to true complementary has been meant. But I stumbled on the question if the 2N2955 really was true complementary to the 2N3055? In the 1970ies we were told, especially by ELEKTOR, that it isn't. They strongly recommended to use the Motorola pair 2N3055/MJ2955 in their EQUA amplifier, e.g. Some time later, the Texas Instruments TIP2955/3055 pair was allowed also.
Contemporary tables listed BDX18 as the 2N3055's complementary.
Best regards!
Julian Vereker is quoted as saying " Complementary transistors are as alike as men and women of the same height and weight ". I was told Julian argued for years that the NPN transistor would effectively become a PNP device if 100 % local negative feedback was used via the collector output ( as said H C Lin circa 1957 although it was PNP germanium then ). This is not absoluty true as the voltage gain of this device is about 0.97 and the emitter follower about 0.92. Stablity also could be less good. Looking back Julian came up against something we get at DIY audio. Some will argue a point because they can. More like a court of law and no respect for someone trying to say something important and needing help. I respect Julian for forcing the correct arguement to now be considered as fact. If quasi has a problem it is the need for correct biasing and a way of keeping it at that point. The remarkable Quad 303 is better on this and has a more ideal input distortion curve.
You guys should look at cloning the Crimson Electrik amps. As best I know a Naim Quad 303 hybrid. The Naim input could be added. If they had had the BDY56 transistors of Naim or whatever we like these days they could be best of the Quasi's . Most NPN PNP amps always sounded a bit slower and darker than the Naim to me. The thing I like about the Crimson is the rhythm and zing. Seeing as the output devices are slow the zing is a big surprise. The Crimson is the only amplifer I knew that seem to please all that heard it. The sort of sound the Audio Lab implied, but didn't have. I only say that as their Lentek amp did.
You guys should look at cloning the Crimson Electrik amps. As best I know a Naim Quad 303 hybrid. The Naim input could be added. If they had had the BDY56 transistors of Naim or whatever we like these days they could be best of the Quasi's . Most NPN PNP amps always sounded a bit slower and darker than the Naim to me. The thing I like about the Crimson is the rhythm and zing. Seeing as the output devices are slow the zing is a big surprise. The Crimson is the only amplifer I knew that seem to please all that heard it. The sort of sound the Audio Lab implied, but didn't have. I only say that as their Lentek amp did.
- Status
- Not open for further replies.