Interesting Sony preamp schematic (TA-E88B)

[thermionic]

Hi,

I'm renovating one of these at the moment and thought the topology might interest some here. The preamp was released in 1978. It uses a 150-position stepped attenuator that feels like a pot - you can see it here: http://www.thevintageknob.org/SONY/sonyesprit/TAE88/TAE88.html# If you marketed one of these today it'd have a price tag well into five figures I reckon.

I've attached the basic schematic. You can see the full info here: http://download269.mediafire.com/nm1gdixmdxxg/mm2mthzk4zx/TAE88-TAE88B.zip

Note that the manual goes into exhaustive detail about the topology and is about as techy as any manual I've seen.

Anyway, I know this place isn't really about hi-fi, but I thought the topology might be of interest. I notice very little in the way of caps in the path - it's pretty much direct-coupled throughout. The supposed THD figure of 0.002% seems pretty good for 1978.

All the resistors are metal oxide film. They look like multi-resistor packages, but only have 2 legs.

Justin

 

[Samuel Groner]

Thanks!

Samuel

 

[xmvlk]

TAE88-TAE88B.zip

Thanx a much. Every schematic is better than ton of textbooks.
May be head preamp PS stabiliser is not good for my eyes..., and
feedback RIAA in this grade amp ?

 

[thermionic]

Why is it that feedback RIAA is considered crude? If a design which implements it can get low THD, and RIAA conformity within 0.1dB 20-20, what's the problem with it?

Does anyone see anything particularly interesting in the Sony schematic? What do you think they were trying to do in this case? You may know better than me, but to my eyes, I tend to associate these types of designs with that which you might see in an IC, not a discrete design - i.e. low input + output offset with high input impedance and liberal use of current mirrors. Considering plenty of ICs were around by 1978, what do you think made Sony go for a high-component-count discrete design?

When I get a moment I will take an FFT plot of the RIAA (I have a Lipshitz reverse filter which is good to within 0.05dB).

Justin

 

[abbey road d enfer]

Why is it that feedback RIAA is considered crude? If a design which implements it can get low THD, and RIAA conformity within 0.1dB 20-20, what's the problem with it?

Blasphemy!  Unfortunately, the world-old concept that NFB cannot work is still very much alive with audiophools... You know, NFB cannot work because at the time NFB hits the retroactive node, the input stimulus is already gone. This is very similar to Zeno's paradox about Achille's arrow, which cannot reach the turtle http://en.wikipedia.org/wiki/Zeno's_paradoxes . Some old greek philosopher has already demonstrated the flaw in reasoning of those who condemn the use of NFB. Serious problem here: the ancients have already stolen most of our inventions, in addition, they destroyed our beliefs in advance...

Does anyone see anything particularly interesting in the Sony schematic? What do you think they were trying to do in this case? You may know better than me, but to my eyes, I tend to associate these types of designs with that which you might see in an IC, not a discrete design - i.e. low input + output offset with high input impedance and liberal use of current mirrors. Considering plenty of ICs were around by 1978, what do you think made Sony go for a high-component-count IC design?

I find it interesting because these guys went all the way to design what they believed was the utmost in the state of technology at the time, and they ended up making a discrete version of the 5534 and the TL07x . But they could not decently put on the market a high-profile preamp with IC's... The only thing a monolithic can't do is the 16 large-geometry input devices.
Anyway, thanks for posting it.

 

[thermionic]

Oops! I should've said 'high component count discrete design.

Abbey - it's good to know my observations weren't too far off base. The FET LTP reminded me of the TL07-series, and with the complementary topology throughout, it made me think that much of the 07-series' virtues, i.e. minimising DC offset + high input impedance, would be exhibited.

So, in effect, this is 10kilos' of discrete TL074. They could've done it with a handful of 074s and filled it with bricks

Justin

 

[Samuel Groner]

Equating these designs to NE553x and TL07x ICs is about like saying that anything with a differential BJT input is just a LM741. Both NE553x and TL07x have significantly different topologies than what is shown in the manual (e.g. the TL07x does neither have a differential second stage nor input cascode, and the NE553x is a three-stage topology while what we have here are one- and two-stage architectures), and there are many other considerations (e.g. quiescent current or absence of substrate capacitance) which might favour a discrete design. I don't want to defend this design too much--I'd surely do things differently as well nowadays--but would appreciate it if the discussion would be a bit more differentiated.

Samuel

 

[thermionic]

Hi,

See attachment for an RIAA plot taken via a Lipshitz-design reverse RIAA filter (accurate to within 0.05dB).

Less than 0.1dB down @ 50k - how impressive is that for a 1978 design? I doubt they had access to the kind of component tolerances we have today (did they have 1% caps in 1978?).

BTW - it's probably easiest to open the attachment in another window and shrink / enlarge to fit.

Justin

 

[audiox]

did they have 1% caps in 1978?

Yes they/we had.

 

[abbey road d enfer]

Equating these designs to NE553x and TL07x ICs is about like saying that anything with a differential BJT input is just a LM741. Both NE553x and TL07x have significantly different topologies than what is shown in the manual (e.g. the TL07x does neither have a differential second stage nor input cascode, and the NE553x is a three-stage topology while what we have here are one- and two-stage architectures), and there are many other considerations (e.g. quiescent current or absence of substrate capacitance) which might favour a discrete design. I don't want to defend this design too much--I'd surely do things differently as well nowadays--but would appreciate it if the discussion would be a bit more differentiated.

Samuel

I see that the pinch of salt I had put on my post has passed totally unnoticed to you. The main point I was trying to carry is that these designers made a product based on DOA's, with a generous use of NFB, which seems to be close to anathema for a number of audiophools.

 

[PRR]

It's discrete because it runs 60V rails.

And, in 1978, the available ICs were audibly flawed. It was a few years later that mid-class gear (Crown, Dyna, Gately) used ICs more or less inoffensively, but this beast is clearly aiming higher, sooner.

 

[thermionic]

It's discrete because it runs 60V rails.

And, in 1978, the available ICs were audibly flawed. It was a few years later that mid-class gear (Crown, Dyna, Gately) used ICs more or less inoffensively, but this beast is clearly aiming higher, sooner.

When I posed the question earlier, I was thinking (erroneously) that maybe the 072 was available by 1978.  When did it arrive? http://www.analog.com/library/analogDialogue/archives/39-05/Web_ChH_final.pdf

Why on earth would you want 60v of swing on an unbalanced hi-fi preamp? Solely because the discrete devices used are designed for use at higher voltages, or nice ad copy?

Weren't LM317s used in synths in the late ‘70s? (my ’79 Prophet has them). See attached photo of Sony discrete regulator (complete with vintage Panasonic FC cap). I've had a 'scope on the regulators, and to be honest, they don’t appear cleaner in terms of hash or ripple than a modest commercial monolithic Vreg. Definitely no better than a decently implemented LM317T hash / ripple-wise, but better than an Apps Data-copied 78**-series.

Vreg solders into the PCB by a rather unusual method: it has solder pads along its edge (similar to those seen on Amphenol-fitted PCBs). These sit at a right angle to reciprocal solder pads on the main PCB, with solder blobs in the 90-degree gap between each contact. It looks crude, but is effective / sturdy.

See the ‘scope shot for channel matching via line inputs – this was just a random comparison, taken from a 300mV source. The attenuators certainly do a good job.

Bearing in mind this piece is from 1978, determining which features appear due to engineering necessity and which are there for marketing is something I find fascinating.

“High voltage rails! Discrete regulators!”… it all sounds a bit like modern ad copy…

Justin

 

[thermionic]

Channel matching

 

[thermionic]

Level attenuator tracks (balance control is same design):

 

[abbey road d enfer]

And, in 1978, the available ICs were audibly flawed. It was a few years later that mid-class gear (Crown, Dyna, Gately) used ICs more or less inoffensively, but this beast is clearly aiming higher, sooner.

When I posed the question earlier, I was thinking (erroneously) that maybe the 072 was available by 1978.  When did it arrive?

The TL0xx were just barely available in 1978. This Sony preamp must have been designed a couple of years before being actually put on the market, because of the manufacturing cycles of the time. Probably no CAD, all hand-drawn, no sim, all educated guesswork 

Why on earth would you want 60v of swing on an unbalanced hi-fi preamp? Solely because the discrete devices used are designed for use at higher voltages, or nice ad copy?

I believe you've answered your own question...

Weren't LM317s used in synths in the late ‘70s? (my ’79 Prophet has them). See attached photo of Sony discrete regulator (complete with vintage Panasonic FC cap). I've had a 'scope on the regulators, and to be honest, they don’t appear cleaner in terms of hash or ripple than a modest commercial monolithic Vreg. Definitely no better than a decently implemented LM317T hash / ripple-wise, but better than an Apps Data-copied 78**-series.

IIRC, again, LM317/337 were new kids on the block in '78

Vreg solders into the PCB by a rather unusual method: it has solder pads along its edge (similar to those seen on Amphenol-fitted PCBs). These sit at a right angle to reciprocal solder pads on the main PCB, with solder blobs in the 90-degree gap between each contact. It looks crude, but is effective / sturdy.

That was a commonly accepted way of mounting daughter boards, a PITA for maintenance, but no connection problems.

Bearing in mind this piece is from 1978, determining which features appear due to engineering necessity and which are there for marketing is something I find fascinating.

I don't think these high-part count DOA's and regs can be justified by engineering "necessity". At the same time several companies made much simpler devices without lacking in useful performance. In the end, the cost of electronic components is just peanuts compared to metalwork and some parts like a good mains xfmr or a good reliable volume control.

 

[PRR]

> the cost of electronic components is just peanuts compared to metalwork and some parts like a good mains xfmr or a good reliable volume control.

And marketing.

Marx yammered about the means of production. Bah. If you build it, it just sits there. You need high-energy guys riding around the countryside talking people into buying it, and you need a good distribution system, and ultimately you need a good warranty/repair service.

When I worked in the Tandy warehouse (for $2.15/hour), a $99 retail receiver came into the warehouse at $40. 60% of the retail price was marketing and distribution. I don't know what the ratio is today, that's a hotly negotiated topic between producers and marketers, but I think on general electronics under $999 the just-making (including delivery to target country) is much less than half the retail price.

And just as there is a market for a cheap radio, there is a market for a maximum-cost bragging-rights system. A small market, but the percent and per-unit profit is very high. In that zone it is not about minimizing cost. It is about increasing cost as long as it increases ad-blurb BS.

I have a Yamaha amp with all the sheet-metal deeply stamped in little hexagons. Why? Yes, it knocks quieter than a heating-duct or a $69 cassette deck. But a box of damping-panel would do that, without the absurdly costly stamper. But it looks SO COOL in the full page color magazine ad, and even better in the 8-page brochure at Hi-Fi Heaven. Now in my old age I regret owning an amp that weighs more than both my dogs put together, but it still looks SO COOL.

Oh, it also has a switch to turn-off the phono preamp. Not just audio, it actually takes power off the preamp. So that it does not leak hiss into the rest of the chassis. Well, OK, but if you don't know, and hit that switch, the phono is "dead". You re-engage the switch, but nothing happens. There is a 30-second anti-thump delay, much longer than a frustrated user's attention span.

 

[xmvlk]

NFB cannot work is still very much alive with audiophools... You know, NFB cannot work because at the time NFB hits the retroactive node, the input stimulus is already gone. This is very similar to Zeno's paradox about Achille's arrow, which cannot reach the turtle http://en.wikipedia.org/wiki/Zeno's_paradoxes . Some old greek philosopher has already demonstrated the flaw in reasoning of those who condemn the use of NFB. Serious problem here: the ancients have already stolen most of our inventions, in addition, they destroyed our beliefs in advance...

;D ;D ;D
But seriously. The meaning, that NFB do not work for phono amps
is based on serious psychoacoustics tests { Who do not ever hear MP3 codec
-it is based on the some tests}

Possible explain of this phenomenon is due to M. Otala's nonlinear distortion of
impulsive noise of phonograph plates. Of course, for pure signal, everythink is O.K.
{because of linearity}, but impulsive noise of old plate {all plates are old, because
new recordings are on CDs with much better quality} is of greater amplitudes than signal.
Then amp limits and feedback can not discriminate what is real signal and what is limitation
and generates artefact. This artefact is diferent {and percepted differently} than artefact generated by limitation of wideband amplifier which is followed by RIAA corrector.