Mullard LP1173 amplifier module

[WaveyDipole]

I was wondering whether anyone has a circuit diagram for one of these modules? They use a germanium AD161/162 pair for output. I did have a look at the UVR service data but there doesn't seem to be anything present in the library.

There seems to be around 14v on the speaker output, but the output transistors on both modules seem to be fine. Indeed, it would seem that the voltage is arriving via the 220ohm feedback resistor. These seem to be fairly simple circuits and I have attempted to reverse engineer the circuit and model in LTspice, but the model will not work for various reasons. I have retraced and gone over the circuit several times now (see attachment) but am still left scratching my head, so a circuit diagram to confirm would be of great help. According to the simulation, the voltage drops to millivolts once the speaker is connected, but even so, should not the feedback line be connected before, rather than after the output capacitor (yes unfortunately it does have one)? Why is there no decoupling capacitor between speaker out and feedback line? I am uncomfortable with the idea of DC voltage being present on the speaker terminal. This is the same on both modules I have to hand.

Would I be correct in thinking that since these are kit items they would have been engineered to a budget and kept as simple as possible so compromises will have been made to save cost?

[Cobaltblue]

This previous post will shed some light I hope

https://www.vintage-radio.net/forum/...d.php?t=104289

Cheers

Mike T

[WaveyDipole]

I just realised that Q2 has a bit of a strange arrangement in that it has no DC supply until the speaker is connected, which then provides it with a path to GND via R14 and the speaker. That seems to correspond to the observations on the Deccalian A600 where he mentions components at 26v until the speaker is connected.

However, I also realised that I might have been overloading the input. I was simulating injecting 1Khz at 1v, based on the idea that line in can be up to 2Vp2p. The result was that my input looked more like square rather than sine waves after Q1 and Q2. Reducing the input to between 100-200mA seems to have produced much more sensible results in the simulation. However, my question regarding DC on the speaker remains. Wouldn't have been more sensible to provide DC to Q2 via a resistor to GND, and have the bootstrap the other side of C7? That way, DC to the speaker is blocked via C7. No?

[paulsherwin]

It's certainly an odd design, and I can't see any advantages to it. I'm not a design engineer though.

[WaveyDipole]

Just downloaded the Deccalian A600 service data from UKVRSD (thanks Cobaltblue). It confirms that my diagram is correct, except for one minor error in capacitor value. There was a small square ceramic capacitor marked 'n15' which contrasted with a Mullard mustard marked 15n 160v. I assumed this was also a 15nF but at a much lower voltage, but on the A600 diagram it is marked as 1500pF.

Mullard also seems to have made some changes. The transistor for Q2 in my module is a BFX88, whereas in the A600 version its an AC128 and has a resistor and capacitor in parallel between the emitter and the positive supply. There is also a 25uF capacitor on the input instead of a 4.7uF. The 680uF is a 640uF, 47uF is a 50uF and 470uF is a 500uF. Otherwise it seems that the design is identical together with the weird ground via the speaker arrangement for the driver transistor.

I will need to have a think about how I address this issue as I would really prefer not to have DC through the speakers, even if its only milliamps.

[Refugee]

The bias earthing through the speaker might be something to do with what happens if the unit driven hard into distortion.

[mhennessy]

This is actually quite a common arrangement, especially on cheaper designs.

It is using the output capacitor C7 as the bootstrap capacitor, hence saving the cost of a separate bootstrap capacitor.

So the feedback via R14 is actually positive feedback. It raises the effective value of the collector load seen by Q2, hence its gain. As I say, this is normally done by a separate bootstrap cap.

The feedback via R6/R7 is the expected negative feedback.

But as noted, the trouble with this arrangement is that it requires the loudspeaker to be connected (in designs with a separate bootstrap cap, there is another resistor which is obviously always in circuit). So if you have a headphone socket, this can be "interesting".

[Herald1360]

Originally Posted by WaveyDipole

There was a small square ceramic capacitor marked 'n15' which contrasted with a Mullard mustard marked 15n 160v. I assumed this was also a 15nF but at a much lower voltage, but on the A600 diagram it is marked as 1500pF.

n15 is 150pF just to add to the confusion (0.15nF)

[WaveyDipole]

Originally Posted by mhennessy

This is actually quite a common arrangement, especially on cheaper designs.
It is using the output capacitor C7 as the bootstrap capacitor, hence saving the cost of a separate bootstrap capacitor.

So the feedback via R14 is actually positive feedback. It raises the effective value of the collector load seen by Q2, hence its gain. As I say, this is normally done by a separate bootstrap cap.

Cost cutting as suspected then. As you say, capacitors were much more expensive back then.

Originally Posted by Herald1360

n15 is 150pF just to add to the confusion (0.15nF)

Well the Deccalian module used a germanium AC126 transistor in that position with a 1500pF between be, so maybe when it was later changed to a silicon BFX88 then the cap was reduced to 150pF?

Thanks for you input on the the meaning of n15 though. That does make sense as you put it in brackets, i.e. .15nF

[mhennessy]

Originally Posted by WaveyDipole

Cost cutting as suspected then. As you say, capacitors were much more expensive back then.

On the other hand, there is a certain elegance about it. The signal on the end of a separate bootstrap capacitor should be identical to that on the LS side of the output cap, so if you can be sure the load is connected at all times, then fitting the separate cap is arguably over-engineering. One less part to fail...

Incidentally, C3 is also boot-strapping, raising the input impedance. It's an engineering approach that isn't seen so much these days in audio designs thanks to (often misplaced) concerns about electrolytic capacitors in the signal path. Taken to extremes in specialist gear, some bootstrapping schemes float the entire power supply of a stage in an effort to maximise impedances.

The only downside here is - as you mentioned earlier - the 50mA current going into the voice coil. It's not a big problem in practice, but as a rule of thumb I don't like to see more than about 50mV (~6mA) in my designs. Increasing the value of R14 to 1k8 isn't feasible given the Hfe of the output transistors, and making them Darlington pairs would have added too much to the cost at the time (no-doubt more than the cost of a bootstrap cap and extra resistor). Everything is a compromise