What was the first Complementary AB bipolar amplifiers?

in my post #19 I mean in my first sentence
"first true complementary medium power transistor pairs in Europe" and not
first true complementary silicon transistor pairs in Europe
- sorry

In Germany while the sixties silicon transistors has been rare and germanium transistors generally in use. One the first true complementary silicon medium power Transistors here in Germany (ITT, Siemens, TFK) are the BC141/161 and 2N2219/2N2905 in the TO-18 outline
for the 2N3055 (TO-3 all brands), BD106/107 (SOT-9, similar TO-66, ITT) and BD130 (TO-3, Siemens) or BDY aquivalent (Mullard) no complement was offered - the offered MJ2955 for the 2N3055 I find not before the seventies.
Actually until this day really true complementary types still not available, although most semiconductor brands claim the opposite.

BTW, in my old databook from"Transistoren 1973/74" from ITT really very explicite datas and a wide range of diagrams are to find for the currently types in this aera, much more than on the scanned datasheets, that I find online. Therefore to search buy/rent such historical databooks isn't nonsense, even with the exist www.
 
Last edited:
Thanks everyone for the AD161/162 output pair. I did not know about it. My portable radio of the time, and even today was the Telefunken Bajazzo with both driver and output transformers. Still sounds great, even today.
I agree - some devices from the Telefunken Bajazzo series I get from time to time to perform maintenance.
The USA equivalent of AD161/162 are the follow:
2N4077/2N2835 (AMP = Amperex Electronics Corporation, Slatersville Devision, George Washington Highway, Slatersville, RI 02876, Rhode Island)

Unfortunately on the web I don't find any informations about original datasheets and amplifier/radio models in English and German, where this types inside.
Perhaps in USA this types was the first true complementary transistor pair and perhaps you have in your own archive informations about this USA equivalents - thank you for looking.

AndrewT: First we must talk about the devices and especially about the first date of release. Only then one could have success to find the first amp models in old magazines and broshures from the right time aera, where such amps mentioned.
 
That is great info... but not fully complementary from input to output. Which JBL was that anyway and what year if you happen to know?

I was thinking it would probably have differential pairs in both NPN and PNP on the input driving either more diff pairs or a gain stage and then on to whatever fully complementary output scheme, either common collector or common emitter. Fully complementary from input to output. Which amp?
 
I made my first complementary through-out power amp in 1967. I amended it in the summer of 1968 with the complementary differential input stage.
In parallel with me, but separately, Jon Iverson of Electro-Research independently developed the complementary symmetry input stage as well.
These were not published designs, but trade-secrets for a few years. Later, Southwest Technical made the first public advertisement of the complementary symmetry input stage in 1972, I'm pretty sure. And then Jim B. worked with SAE to build complementary symmetry power amps around 1973. Bell labs experimented with fully complementary designs in the late 60's, but they appeared to be optimized for some other quality than audio linearity.
 
latala your subsequent question about true complermentarity.
I notice you mentioned JLH's idea in adding the cap in parrellel with the diode on a quasi-configuration. I think JLH's work was in 1973 and the HFN-RR 75 watt amp was the first. In the text of that article he addresses your question and makes the following points. Complementary power transitors were good at lower frequencies but non-complements could deliver better results at higher frequencies. I haven't got the text in front of me but the limiting factor may have been hole storage. He was working with MJ481/491
2N3055/2955 in the lower powered versions and 2N3771(2-3?) or some number near that in the full blown 75 watt edition. Later BDY56's I think.
 
when you say "complementary AB bipolar amplifiers", are you talking about complementary output devices for power (germanium or silicium) or are you talking about "symmetric complementary" structures like having two amps in one, like having one NPN differential input stage, plus one PNP differential input stage, in a mirror-symmetry ?
 
For documentary purposes, the attached pictures indicate what was the status in October 1960, from a AES paper by Jones from GEC.

It is about the 2N656A NPN 60 Volt silicon power transistor, good for several amperes. As you can see, the application is the RCA H.C. Lin inverting audio power amplifier arrangement from 1956, with a bootstrap in the VAS.

Actually, the title about "silicon transistors" is somewhat misleading because TR3 (PNP driver) is a germanium transistor 2N1057. Now you know why Jones prohibed the word "complementary" in his presentation ! The driver stage is completely turning back from complementarity, with TR2 being a silicon NPN power transistor (2N656A selected for minimum beta equal to 20 at 1mA collector current) and TR3 a germanium PNP medium power transistor (2N1057). Amazing !

What was the status regarding germanium transistors, at the same time ? Were there germanium complementary power transistors available yet ? Which ones ?

If somebody could post or point the original paper from RCA written by H.C. Lin in 1956, that will be quite interesting. I guess that the 1956 version is using germanium transistors. Possibly germanium complementary devices ?

Regards,
Steph
 

Attachments

  • AES 1960 - GEC front cover.jpg
    AES 1960 - GEC front cover.jpg
    95.7 KB · Views: 1,233
  • AES 1960 - GEC page 1.jpg
    AES 1960 - GEC page 1.jpg
    260.4 KB · Views: 1,237
  • AES 1960 - GEC figure 4.jpg
    AES 1960 - GEC figure 4.jpg
    76.3 KB · Views: 1,151
Last edited:
For documentary purposes, the attached pictures indicate what was the status in October 1967, from a AES paper by Haas from Motorola.

In his text, Haas is referring to the 1956 Lin arrangement, calling it a "basic quasi-complementary symmetry power amplifier".

The new circuit Haas is presenting is the ancestor of all modern "blameless" amplifiers, still in use today. Haas describes his amplifier in 1968 as an "experimental 60 watt 8 ohm audio power amplifier utilizing full conjugate complementary symmetry output circuit".

Please note it is using a symmetric power supply with a +36V rail and a -36V rail.

Please note the atypic terminology Haas is using, calling the VAS stage the "driver circuit".

The power section is not a Darlington, but a CFP arrangement. The power output devices are MJ2841 (NPN) and MJ2941 (PNP).
They are fed by MJ3006 (NPN) and MJ4006 (PNP).
The VAS is a MM4009 (PNP).
The long-tailed pair at the input is MD8002 (NPN pair).

One may ask why Haas did not used a constant current sink in the long tailed pair, and a constant current sink as VAS collector load. Conceptually, this is a cleaner approach than the simple resistor in the long tailed pair, and the bootstrap as load in the VAS. Maybe all this was available and done at Motorola, but kept secret for keeping an advantage over the competition.

As soon as 1969, Clive Sinclair, with the Project Sixty Z30 and Z50 power amplifiers, was introducing the constant current sink in the VAS, as collector load. But the output devices of the Sinclair Z30 and Z50 were quasi-complementary (ME6101 + BD155 for positive drive, and ME0402 + BD155 for negative drive). And there was still a simple resistor, in the long tail pair emitters. And the power supply was asymetric, using a +35V stabilized power supply (the PZ5). The ballast transistor of the PZ5 stabilized power supply was a 2N3055.

Regards,
Steph
 

Attachments

  • AES 1967 - Motorola front cover.jpg
    AES 1967 - Motorola front cover.jpg
    107 KB · Views: 1,099
  • AES 1967 - Motorola page 2.jpg
    AES 1967 - Motorola page 2.jpg
    287.3 KB · Views: 999
  • AES 1967 - Motorola figure 6 and 7.jpg
    AES 1967 - Motorola figure 6 and 7.jpg
    115.8 KB · Views: 669
Last edited:
Well researched, Steph. The bootstrap was used because it allowed voltage drive ABOVE (in this case, below) the power supply voltage. It was then considered better than just a constant current load, and cheaper too. I thought differently in 1968, but that was the debate between my colleagues and me. Now, only on this website, do you see common examples of bootstrapping
 
Last edited:
I used this output stage configuration in the '67 paper, at Ampex Research in 1969 for a very large motor drive amplifier. I found it difficult to use, due to a tendency to instability. In future designs, especially with Richard Heyser's insistence that the complementary Darlington configuration was actually better, I switched to the Darlington in future designs. Never looked back.
 
Hello,

I'm now realizing that the medium power complementary AD161/AD161 germanium pair was there a long time before the advent of so-called silicium power complementary pairs. And I'm now realizing that using AC127/AC128 pairs and AC187/AC188 pairs, there was the possibility to build mirror-symmetry full complementary amplifiers, with one NPN long-tailed pair at the input, plus one PNP long-tailed pair at the input, one NPN VAS and one PNP VAS, and of course, a fully symmetric complementary arrangement for the output stage, being Darlington or CFP.

Who did that, actually ? Some high-end german equipment manufacturers, maybe ? Was there a market for this, maybe between 1964 and 1970 ?

It took quite a long time to get decent silicium power complementary pairs. I'm now realizing that before the advent of high-power silicium complementary pairs, the whole audio industry needed to fiddle with quasi-symmetry silicon arrangements, while in the past they had the possibility to use a neat complementary germanium AD161/AD162 arrangement, however medium-power instead of high-power.

The 2N3055 was available since long (RCA 1962), but there was no PNP equivalent. When came the 2N2955 ? Was it a decent complementary ?

In my documentation, I can see the BDX18 used as PNP complementary for the 2N3055 (RIM NFK 60 Watt power amp, 1972), with a BD137/BD138 complementary pair as drivers, in a +29V -29V symmetric supply arrangement (about 60V in total).

From what I can see in my documentation, untill further info, the real practical silicon solution came around 1971, not with the expensive Motorola complementary transitors like 2N3716/2N3792 or MJ481/MJ491 or MJ2841/MJ2941, but with the availability of the more affordable TIP31A and TIP32A used in many 15 Watt audio amplifiers powered by a 50 Volt supply.

In 1972, the Luxman SR700X power amplifier used a neat 2SD155/2SA616 complementary pair arrangement with a 2SC495/2SA505 complementary pair as drivers, in a +28V -28V symmetric power supply arrangement (about 60V in total).
There were also the Sony TA-1055 and TA-1066 amplifiers, around 1973, using a neat 2SC1061/2SA671 complementary pair arrangement with a 2SC1124/2SA706 complementary pair as drivers, in a +25V -25V symmetric power supply arrangement (about 50V in total). Those were the big player choices and solutions, in the first half of the seventies.

I have very few audio power amps, in my documentation, fitted with a 2N3055 /2N2955 complementary pair. Is this representative of the reality ?

Around 1971, there were also the complementary Darlingtons, in the form of MJE1100/MJE1090 good for 25 watt, and MJ3001/MJ2501 good for 60 watt.

I have a table here, listing the most common silicon Darlington complementary pairs used between 1975 and 1985, able to cope with a +42V -42V symmetric power supply :

NPN PNP
BDV65B BDV64B
BDX65B BDX64B
BDX87C BDX88C
BD649 BD650
TIP142 TIP147

In 1982, Elektor (a diy electronic magazine) was still describing a simple audio power amplifier using BDX67B BDX66B Darlingtons, using +42V -42V symmetric power rails, very similar to the Haas Motorola amp presented at the AES in 1968, now with Darlingtons instead of the Haas CFP arrangement.

I think that in the second half of the seventies, European audio power amplifiers makers like Philips, Grundig, Telefunken, Braun, Bang & Olufsen had a distinct preference for complementary Darlingtons, is that right ? Mechanical simplicity, thus.

While Japanese, at the same time, had a preference for complementary non-Darlingtons. Speed and precision, thus.

While at the same time, the American had a preference with 2N3055 in a quasi-complementary arrangement ? Very rugged and everybody the same.

And the Ruissians ? Very rugged and everybody the same like in USA ? Haha !

Which may explain why there are so few commercial amplifiers using a complementary 2N3055/2N2955 pair ?

Need to investigate. Interesting theme. Do you agree with me ?

In the late seveties came a double audio amplifier revolution. We got the STK hybrid modules. At the beginning, the STK modules had quasi-complementary devices inside, kind of cheap 2N3055 second-source dies, maybe ? And, in more sophisticated amplifiers, we got the MOSFETs. In reaction to this, the most conservative (and reputable) amplifier makers started a battle, about THD and TIM, with new and sophisticated concepts applied to conventional BJT output stages in amplifiers like the Technics New Class A, JVC Super-A, Pioneer Non-Switching, Sansui Non-Switching, Hitachi Super-Linear. What a change ! Kind of golden age, between 1978 and 1982. Reinventation and sophistication of the full complementary output stage.

There were other silicon advances like the multi-emitter technology (also called perforated emitter technology) for inceasing the safe operating area (less second breakdown) like from OnSemi the MJL21193/MJL21194 complementary pair, quite slow by modern standards with a 4 MHz transition frequency.

OnSemi is now displaying no less than 71 different High Power Audio Transistors.
Audio Transistors

Fast power BJTs with a 30 MHz transistion frequency came with the MJL1302/MJL3281 complementary pair from OnSemi. Beta is nominal 100, matched to 10% from 50mA to 5A.

Some refinements in the geometry enabled the current gain not to drop at high current, those are the "sustained beta" variants from OnSemi like the NJW1302/NJW3281 complementary pairs, and still high speed with a 30 MHz transistion frequency. Their beta is nominal 100, matched to 10% from 50mA to 5A, and only starts to drop at a 5 amp current (Vce as low as 5V), instead of 3 amp for the equivalent non-sustained beta variant.

Then, most valued BJTs, you have the ThermalTrack variants, housed in a 5 pin package, incuding two terminals for the built-in tracking bias diode. Those are the OnSemi NJL1302/NJL3281 complementary pair, featuring the 30 MHz transition frequency, but not the "sustained beta" feature yet.

Of course, you'll find other silicon advances through the Hitachi and Magnatec-Semelab lateral MOSFETs designed for audio, and International Rectifier vertical MOSFETs (not designed for audio).

Magnatec-Semelab is about to launch complementary lateral MOSFET pairs for audio in a 5-pin housing, with the NMOS and PMOS inside. Those are the ALFET 16V5 for 160V applications and ALFET 20V5 for 200V applications. It means there are no source resistors, of if there are source resistors inside, you can't access them for measuring the voltage drop they generate.
Magnatec. ALFET Lateral MOSFETs

Obviously, the MOSFETs killed the Darlingtons.
BJTs are still favoured by many, but for how long ?

Making a power amplifier using MOSFETs is a piece of cake, now that the Cgs and Cds legends have vanished. You DON'T need 50mA in the VAS stage for coping with the more-than-one nanofarad gate capacitance, in order to get a decent slew-rate. Because its effective value gets divided by 40, because of the bootstrapping effect of the source follower (common drain) arrangement. Transconductance of 5 Siemens multiplied by a 8 ohm load resistance equals 40. That's a scientific fact. So you end-up wit an effective 100pF to 200pF capacitive load, plus another 100pF maybe from the drain capacitance. And there are modern topologies, enabling a VAS to operate at a modest 5 mA bias current (like in a BJT amp), and still being able to drive 50mA in the gate MOSFETs, if needed. You thus get slew-rates in excess of 100V/µs, without pain, without high VAS bias currents. See the Kuroda 1982 derived amplifiers. That's an excellent school, if you want to consider audio power amplifiers in a modern, new approach.

Some are now reverting back to vintage concepts, buying 2N3055 transistors again, and assessing the subjective audio qualities of various combinations embedding the 2N3055 magic. Do they know that at the origin, the 2N3055 got designed, certainly not for audio, but as active ballast for DC regulated power supplies ? Which doesn't mean that those amplifiers lack qualities.

Actually, the whole complementary concept is maybe a 20th century technological hoax, as the ear feels uncomfortable when you have some 3rd and 5th harmonics as THD, but no 2th nor 4th harmonics. Unfortunately, that's exactly what you are going to get using full complementary designs. The ear wants to get some 2th and 4th harmonics inbetween. The only way to get this, is NOT to use a full complementary approach. That's what they did with the NAD3020. A semi-complementary design, using 2N3055 devices. And I can tell you, the NAD3020 amplifier sounds great, it gives a sensation of warmness and power, without being overprocessed.

Shall we now hunt and second-hand buy the early STK-equipped audio amplifiers, with 2N3055 dies inside, in a semi-complementary arrangement, and discover that they can sound like the beloved NAD3020, after some tuning ? Life is a joke !

In parallel, there are two other leagues one cannot ignore anymore. On one hand, we have the QSC high-power floating supply audio power amplifier bridge. On the other hand we have the switchmode class-D. Both are non-complementary concepts. They all base on NMOS power devices. Do they represent the future ?

And, at this stage, we still don't know who :

a) first proposed complementary bipolar AB (or class A)
b) first successfully implemented it as an audio power amp
c) first used the venerable 2N3055/2N2955 pair in a product

That was the question, indeed !

Cheers,
Steph
 
Last edited:
To refresh the memory a look at a circa 1968 to 1971 Lafayette or Allied Radio catalog might be a good thing...

On this side of the pond myself, and likely the rest of the USA remained completely unaware that Clive Sinclair ever made an audio amp. The first time I heard of him was with that little (and very good) computer. Can't recall what it was called right now...

_-_-bear
 
To refresh the memory a look at a circa 1968 to 1971 Lafayette or Allied Radio catalog might be a good thing
Good suggestion ! And the radio-shack catalogs Radio Shack Catalogs to see when some transistors arrived on the diy market.

On this side of the pond myself, and likely the rest of the USA remained completely unaware that Clive Sinclair ever made an audio amp. The first time I heard of him was with that little (and very good) computer. Can't recall what it was called right now...
Sinclair ZX-81, Timex 1000 ? If you want to read about Sinclair, that's here : Welcome to Planet Sinclair
If you want to read about the worst preamp of the world (Sinclair Project 60), that's in the post #4 here http://www.diyaudio.com/forums/soli...-complementary-design-sounded-great-time.html

Cheers