"Old-style" transistor audio amplifiers.

[G6Tanuki]

I've always had a bit of a thing for first-generation transistor audio-amps - the transformer-coupled class-B push-pull types using a pair of OC72 or OC81 to deliver half-a-Watt or so of audio.

Having recently acquired a few 1960s NOS transformers [RS T/T7 etc] I'm planning to build a stereo-pair - but with a modern twist.

"Classic" Germanium output transistors [OC81/AC128/AC153 etc] are scarce - and were always rather prone to thermal runaway. Finding the heatsink-clips to fit would be a pain too. But I've acquired a boxload of late-1960s TO5-cased Silicon transistors - CV7343 [BCY30-equivalent] made at the Texas Instruments plant in Bedford. I plan to use these [fitted with traditional 'star' type push-on heatsinks] as the output devices.
I'm assuming that they're also likely to be rather better in overall consistency-of-performance than first-generation Germanium devices so I can get away without the need to spend too much effort on the gain/leakage-matching that seemed necessary with Ge devices.

Which led me to thinking: the tradidional amps used an OC71 driver and a pair of OC72 output, or OC81D driver and pair-of-OC81 output.

Was an OC72 just a glass OC71 with a heat-dissipating metal sleeve? And what was an OC81D? I remember these often came with a cream-coloured sleeve over the OC71-style glass envelope.

I also remember OC78 and OC78D.

Was there anything special about the "D" transistors, or did the likes of Mullard just turn out a batch then test them and label them according to market-needs?

[MrBungle]

I rather like them too. It's the mysterious association with things like Queen for me. I actually found a box of 50 OC81D's a while ago at a boot sale and some other Mullard bits and only found data suggesting they were for driving output transformers. They had same gain and leakage as standard OC81's as far as I could measure. Not sure what power dissipation was like as I didn't push them. Shifted them as I had no real use - surprisingly valuable. Nearly fell off my chair when that auction finished.

[paulsherwin]

I'm very suspicious of all the different transistor types marketed by Mullard in the late 50s and 60s. I suspect that Mullard had become accustomed to selling standard sets of valves for radio use and wanted to carry on doing the same thing with transistors. The production process was the same for all those glass encapsulated OC types, so they were probably just individually selected for optimised characteristics.

They carried on doing the same thing with the AF11x series, even though hand selection had ended and the AF114-7 were literally the same transistor (just have a look in Towers). It must have still seemed like a good idea to market one type as an FM mixer, another as an AM mixer, another as an IF amp etc.

The BCY30 looks very like a BC557 or 2N3906 in a big metal can. You could make an audio amp with them, but they're a bit weedy for the output stages, even if you only need 500mW of output. The gain is a bit low too.

[G6Tanuki]

Originally Posted by paulsherwin

The BCY30 looks very like a BC557 or 2N3906 in a big metal can. You could make an audio amp with them, but they're a bit weedy for the output stages, even if you only need 500mW of output. The gain is a bit low too.

Noted. Maybe I'll use them as drivers and go through my batch of genuine RCA 2N3053 to select some output-devices?

[I rather like the 2N3053: run off a 15V supply-rail a suitably-heatsinked pair biased in Class-AB2 work nicely as a linear-amp delivering 1.2 Watts of SSB at 2MHz]

I've got a batch of these kind of heatsinks:

http://uk.farnell.com/aavid-thermall...-c-w/dp/170065

so keeping cool won't be an issue.

[Biggles]

I have built a couple of these over the years. I also found a silicon version using BD13 somethings as the output pair which was surprisingly powerful, and came from a PA system/door entry/intercom from an old control room which I helped strip out in a previous job. My understanding of the "D" suffix is that they were to be used to drive the normally class A phase splitter/driver transformer just before the output stage. I have never checked but I believe they could handle a bit more current than the OC81. Whether this was just careful selection after a normal production run of OC81s I'm not sure. I have used the little transformers LT44 and LT700 to give acceptable results in small receivers and found you can get about 50-75mW output on average, when using a pair of OC72s.
Alan.

[paulsherwin]

2N3053s should be fine if you have some to hand, though they're NPN (BCY30s are PNP).

[Ed_Dinning]

Hi Gents, main thing that was special about the BCY30 series was that they were symmetrical (Veb was similar to Vcb) so they could be used with an NPN as an SCR "look alike" in timing circuits with a reasonable supply voltage.

Ed

[paulsherwin]

That explains the low gain.

[G6Tanuki]

Originally Posted by paulsherwin

2N3053s should be fine if you have some to hand, though they're NPN (BCY30s are PNP).

I'm happy to stand on my head... and use a BC109 as the driver!

[julie_m]

It would certainly be interesting to revisit the "classic" circuit and see just what it is really capable of ..... I suppose transformer coupling fell out of favour as soon as it became cheaper to use a transistor as a phase splitter and add the necessary hefty electrolytic capacitor. Though, transformers did seem to linger on into the 1980s in some public address amplifiers.

As this amplifier is going to be transformer-coupled, there is little stopping you from using a PNP driver and NPN output pair, if you wanted to be really perverse!

[Herald1360]

Originally Posted by G6Tanuki

[I rather like the 2N3053: run off a 15V supply-rail a suitably-heatsinked pair biased in Class-AB2 work nicely as a linear-amp delivering 1.2 Watts of SSB at 2MHz]

Does AB2 have any meaning for transistors? They always have "grid" current so there's no AB1 so isn't a simple AB all that's needed?

[mhennessy]

No, AB2 is definitely a valve-only thing

[kalee20]

I made a 15W amplifier some years ago, to operate from 12V, using two silicon Darlington transistors in push-pull (BDV66 sort of thing, 15A rated, gain 1,000). Collectors fed a centre-tapped output transformer primary, secondary fed the loudspeaker. I interleaved the windings. I also added some partial emitter loading, to give a degree of local NFB and also a few milliohms of DC resistance, to add to DC stability. Output transformer was on a 1.75 sq. in. core.

The push-pull drive came from a Class A driver, reasonably traditional, single transistor, drive transformer primary in the collector (so it handled the DC), centre-tapped secondary feeding the output bases. Drive transformer was on a 0.16 sq. in core (approx 1cm²). Because the output transistors were Darlingtons, drive requirements were pretty low.

It worked well, frequency response 50Hz - 15kHz. Low frequency performance was curtailed by the drive transformer just not enough inductance so current became discontinuous and distortion set in. But the basic topology is sound, as long as you don't mind winding transformers!

Incidentally, the OC81D has always puzzled me... why did Mullard find it necessary to introduce it, couldn't you use an OC72?

I think the only real difference twixt 'classic' and 'modern' output stages was the use of a transformer and collector (classic) or emitter (modern) topologies. All down to the maximum current the poor transistor can supply. Of course you could do an emitter follower and transformer circuit then you would have no voltage gain (less than one given the 'classic' transformer). May as well squeeze all you can from a then expensive transistor.

[ionburn]

Probably not too difficult to have an oldish looking transistor amp with OC71's or 82's, driving an output stage consisting of slightly more modern Silicon transistors (these being effectively hidden away to avoid spoiling the look. There would likely be a bit of thinking to do as DC coupling to the Silicon may be a problem due to the leakage of the Germanium devices. Perhaps a bit of juggling of transformers and impedances. That way the main stresses would be with the transistors that could best handle it.

[G6Tanuki]

I'm not really bothered what it looks like [it's going to be built semi-breadboard fashion using a piece of SRBP with holes drilled in it where necessary and point-to-point wiring between components]. That's how I built these things in the 1960s, When it's stuffed inside a diecast box you can't see it anyway...

One thing that I've thought about over the years is whether it would be possible (or sane) to "super-alpha" the output transistors to get extra gain. Sticking an OC71 on the front of each OC81?

And using another OC71 or similar as a 'bias regulator' in the path between the centre-leg of the driver transformer secondary and ground - as a way to better thermally-compensate the output-pair. That way it may be possible to dispense completely with the resistor between the commoned emitters of the output-pair and ground; even considering the very low value of this resistor it is invariably going to dissipate some of the audio power given that it can't be bypassed at AF because of the intrinsic varying-current-draw nature of Class-B push-pull operation.

[Argus25]

What a terrific idea. I'm equally fond of this type of amplifier too.

One good option for the output transistors, if you did want to stay germanium, is the AC188. At one stage I saw many of these for sale, but I can't recall the supplier.

Another excellent transistor, if you could get some, is the OC74. It out does the OC72.

The Original RCA 2N3053 is a lovely silicon transistor, and also would look great with the type of heatsinks you are planning. The 2N4036 is another TO-5 option.

I think this idea you have is like wanting to build something from "favorite things" (like the brown paper packages tied up with strings analogy)..it is great.

I once decided to build an all germanium radio with all my favorite transistors in it, these included the AF178, which is actually the better replacement for AF11x transistors, an audio amp with an AC188 driver, and two OC16's as the output transistors, you can see them on the hand crafted copper heat sink:

http://worldphaco.com/uploads/THE_AF...OC16_RADIO.pdf

[Karen O]

There is something about the old transformer coupled amplifiers, perhaps because of childhood memories of peeking inside my transistor radio.

There was a practical advantage of these designs: every time I used an LM386 in a radio circuit I had to take precautions to prevent RF getting through the amp, radiating from the speaker leads, and causing infuriating instability. I think that such RF wouldn't get through a pair of transformers.

[Argus25]

Originally Posted by Karen O

There was a practical advantage of these designs: every time I used an LM386 in a radio circuit I had to take precautions to prevent RF getting through the amp, radiating from the speaker leads, and causing infuriating instability. I think that such RF wouldn't get through a pair of transformers.

I agree entirely. Not only do the transformers act as very effective band pass filter, but the transistors (germanium medium power and output audio types) have a low transition frequency compared to silicon devices and are intrinsically "RF proof".

I remember seeing the dawn of silicon transistor stereo amplifiers and initially the designs were not careful to roll off at high frequencies, then people wondered why you could sometimes hear radio communications clearly coming out the speakers !

On the topic of the audio band pass response of iron cored transformers, I once investigated how to mathematically predict the band-pass response for an inter-stage audio transformer used in valve work in the unique case that it has in vintage radios where there is insignificant secondary loading and the usual analysis not appropriate. To my surprise the equation was not readily available, but was derivable from some equations in Terman's book, it is moderately complex. The equation is on about page 15 (the discussion starts on pg 11) of this article on a Grebe vintage radio if anyone is interested:

http://www.worldphaco.com/uploads/THE_GREBE_MU-1.pdf

[ParcGwyn]

What a beautifully crafted radio.

Dave

GW7ONS