1970s Design Indulgence

I have a 70's Aiwa AX7600 that sounds good (to my ears) but the lack of remote control is a bug-bear. Not mentioning any names I'm in the process of DIYing a power amplifier based on a circuit design done by someone who hosts quite a big website with hi-fi related projects and articles. When I come to do a pre-amp the remote volume control is a definite must. Input selection and 'bypassable' tone controls can remain manual. If it wasn't for my liking of DIY I'd be buying more of your gear instead btw :-)

Thanks for the information Graham. I'll carry on reading this thread with interest. 

At -1dBu power output is just over 32 WPC and THD 1kHz falls to less than 0.1%.

Evident are the current sharing resistors to lower their power dissipation.

The diode bias spreader is a bit better in my opinion as the quiescent current now only varies 6mA (5% at 120mA Iq) between HTs of 65 and 75 volts.

The local NFB resistor R12 was increased to 180k giving around 52dB gain before global NFB. This gives a frequency response up to about 10kHz before global NFB; this supposedly being a plus-point for transient IM.

Although input stage current is lower, slew rate isn't reduced because the compensation capacitor C7 is not between T2 collector and base (according to JLH).

R10 combines with C7 to produce a zero as per the Bailey 30W amplifier.

Back to R12, and its take off point is from the top of the bias spreader such that there is no variance due to the bias spreader trimmer.

There is no reason why C7 cannot be taken from the same point.

There are two DC negative feedback loops which should aid DC stability: the original one due to obtaining bias for T1 base from T2 emitter resistance chain; and from T2 collector (via the bias spreader) to T1 emitter via R12.

In the Bailey amplifier T1 base is biased from a regulated voltage, but its collector-emitter voltage "floats". In this amplifier T1 collector-emitter voltage is mostly regulated by the zenner stabilizer. I say mostly because R12 influences T1 emitter and its current can alter with HT by around 0.01mA. The 2x DC NFB however, should afford it good stability.

With increasing circuit ambient temperature T1 will conduct harder reducing T2 base voltage, reducing T2 emitter voltage and via T1 base reduce its conductance. With decreasing temperature the opposite will be true. Therefore DC NFB 1 is stable.

Likewise T1 conducting harder causes the voltage at the top of R12 to rise, injecting more current across R9-11 such that its voltage rises, reducing T1 conductance, and vice-versa, so DC NFB 2 is stable.

The main difference however, is that T1 can swing a much larger voltage. Its outside-loop gain is 17.5 and its collector load R8 could swing 4.55V rms and so the input could reach about 1/3rd of input sensitivity for full output before distorting; this being limited by the collector resistor rather than the hard characteristic of the transistor. This is assuming the NFB is doing nothing which according to Otala, is possible for a fast leading edge; this being due to sluggish output transistors.

However, R12 applies local NFB of such magnitude that T1 should be able to swing more than the input voltage. Additionally T2 base resistance is part of T1's collector circuit so gain will not be exactly 17.5. It will be more like 10. By adjusting values of R8 and R14 we could obtain perfect symmetry which is an option which may or may not have any effect.

So far I have not detected any switch on "thump" and the only way of telling it's working is to play something. Restarting from a power cut might cause a loud bang in the speakers though, and this is due to source equipment. Should the amplifier meet with my sonic requirements I will consider ways of muting the amplifier at power-up.

What's in a Naim?

Another question could be who thought of it first, JLH or Julian Vereker?

It can be seen that Naim were using a 0.22 ohm output resistor, at least in 1982 (http://www.audiomisc.co.uk/Armstrong/reviews/finale/1982.html - about 2/3rd of the way down the page).

JLH's books (the ones I have) came out in the 90s, so I guess Naim have it.

JLH explains that it helps linearise crossover distortion, especially in quasi output stages, which Naim was using at the time (perhaps still do?).

Jim Lesurf seems to be of the opinion that using an output resistor coloured reviewer’s subjective findings, and he could be right.

Naim were using the resistor in lieu of an output inductor, but it would be insufficient in removing the pole formed at the output due to capacitive loading, and so Naim instructions warn about speaker cable capacitance (I am told).

But let's not get bogged down by review and/or user politics - it is the resistor and how it can be of benefit (or not) I am interested in.

Depending on the designer's idea of output stage current limiting (I am not talking about protection circuitry here, although it is sometimes related), emitter resistors are used and can be in the value range of 0.1 - 0.47 ohms (I am using 0.33 ohms).

Additionally there is intrinsic emitter resistance which varies with output and is Vt/Iout and at 1W is 0.07 ohms, and this value falls with increasing output.

Adding 0.33 to 0.07 we get a total emitter resistance of 0.4 ohms, but this isn't the amplifier's output resistance because it is reduced by global negative feedback.

If we have 20dB NFB it is reduced (by a factor of 10) to 0.04 ohms. And if we have an 8 ohm load the ratio is 200:1 which gives the conceptual damping factor of 200. If we had 40dB NFB it can be seen that a DF of 2000 would result, but we then have wiring resistance coming into play, so DF is diminished and will never reach such heady heights.

It has to be asked what amount of driving impedance control we really need. Valve output transformers tended to provide a DF of around 20. Because a transformer is an AC device, at subsonic frequencies there is much reduced output and therefore output impedance has risen such that at half power it is 14; quarter power 10; and so on. Any amplifier with a low frequency roll-off will not be able to exert control of loudspeaker excursions below some frequency.

DF is not very linear, and so inserting series resistance will at least make it more linear. Some might see it as bringing all pupils in a class down to the level of the least talented.

Remember though, I told you that substituting a 1 ohm resistor for the 100 ohm resistor whilst doing resonance measurements would make it very difficult to read values, and so an output resistance (or impedance seeing we are talking about AC) of little under 1 ohm isn't so bad.

We can estimate the addition of a 0.22 ohm resistor as giving around 0.25 or 0.26 ohms, and so the ratio is around 31:1 or 32:1, and we may as well round it to 30:1, so the Naim damping factor, if it had been published, would have been around 30.

Perhaps an excellent valve output transformer might do 30 at a push, so you can see the point JLH is making.

How we could try it out in my circuit is a bit in-depth and I will tackle it in my next post.