Puccini "REMOTE"(up to v4.3)

You know the story: after some time, someone decides to modify the original Puccini design.
The up side of this matter is the "remote" facility. You can now select the input and the volume while sitting in your seat. The down side is that the original circuit has been substantially changed.
This change affects not only the power stage (the NE5534 with the floating power supply) but even the line stage (the TLE2072).

Don't ask me why! This "philosophy" (modify a design, even if it works fine and people review it enthusiasticly) seems to have been recently applied not only to all the Puccini productions, but even to the Paganini and Maestro CD player: they abandoned the Sony chipset/mechanics and  use a "ready to go" Philips system.

Anyway, talking about the Puccini "remote" and the differences with the original, it would be worth while to review some of the circuit diagrams.
The main issue is a "discrete" operational amplifier, with an FET differential front-end and a transistor output stage.
Lets look at the diagram, considering that this "basic" diagram is not always the same (i.e. some component values are different if we are talking about the preamplifier section or the driving section):

Here's the BASIC (not the exact) diagram used in the preamplifier section (GIF 22k)

As you can see it is a "text-book" circuit; I've seen it as a gain block (GIF 16k) on the Italian Hi-Fi review (Fabrizio Montanucci, July/August 1992), as well as in the MONRIO MC202 power amplifier (PS 116k), but in both case the input was a bipolar, not an FET!

This is a NON trivial detail: get some FETs and measure their VGSoff (gate-source cut-off voltages). You can see the values lie everywhere between -0.25 and -8 volt! A similar problem exists for the Idss values (at a given Vgs).
So, if you want to have a differential pair input stage using FETs you MUST preselect them or severe DC voltage offset will be present at the output.
At first look this circuit seems clever: it is very simple, and on a PSPICE simulation it offers an Open-Loop gain of about 70dB, flat from DC up to 20kHz; the external feed-back resistors (R128, R129) reduce the gain at a value of 13dB (4.5 times); capacitor C118 forces unity gain at DC.
However there are some obscur points that need to be explored in details.

Here you see what happens if the load on the output is varied from  near-infinite to a more realistic value:

Look! Even for a light load of 47k the OL gain drop from 5000 to 500; this is due to the "collector output", that does not isolate the load.
This happens in an asymmetrical way: in fact the output transistor Q101 is somewhat isolated with the help of Q103, and does not "load" the differential stage; this is not true for the other output transistor, Q102: its base is directly connected to Q106's drain, presenting the output load to it, and so affecting the Open-Loop gain.

Another important difference is the "gain" of a differential circuit: bipolar gains are higher than FET's. Period!
Here you can see how Bipolar is better than FET, considering the net gain and the Gain-Bandwidth product:

The only advantage using FET is the "near zero" input current, that is priceless for some applications; but not in this generic audio case.

To end the analysis of this circuit, take a look at the omnipresent NE5534: as you can see the OL gain at 100kHz is still very high (something near 60dB), while the gain of the discrete FET amplifier used in the "remote" has dropped to less than 40dB when the load is less than 10k (as is in the case of the power stage, which has an input resistance of 8.3 kW):

Because the same circuit is used for driving the power output stage (you can imagine how low and non-linear the input impedance is of this), problems will be even worse. More about this later.


Up to now it's been theoretical; now let's look at practical matters.

I'll not spend time describing the input selector (made with relays), nor the volume control (ALPS pot, the ones driven by a DC motor).
The core of all these circuits is a PIC that receives the IR signal and drives the appropriate circuit; other inputs are the Up-Down input selector knob and the "warning: too much current" signal from the output protection circuit
(exactly the same that was developed for the Aida integrated amplifier).
In the Puccini "remote" there is still the Phono circuit. It is the same as in the original version: an NE5532 chip is used for the MC step-up and for the MM+RIAA equalization.

After this there are relays for input selection. Then the signal goes to the motorized ALPS, and at least to the preamplifier section.
Here we have the first big difference with the original Puccini: the impedance seen by the inputs is not constant, but vary with the position of the volume knob! This happens because there is no "isolation" stage BEFORE the volume itself, while in the original Puccini there was a follower ahead.

It is now time to download the diagram (as usual is a zipped PS, 28k)

This is not the entire Puccini "remote" (it lacks the logic part), but it is detailed enough to understand how it works.

As you can see the diagrams start to be different than the original Puccini after the volume knob, where the first active stage is the discrete Op-Amp previously discussed; the Closed-Loop gain is set to [1+(11500/3320)] = 4.46 times; but due to the insufficient OL gain the real measured value is near 4 (12dB). DC gain is set to 1, due to the decoupling action of C118.
In this circuit (as well as in other part of the "remote") there is a large use of "audio grade" electrolytic capacitors: the ELNA Starget.

All the resistors are Welwyn RC55 metal film.

(note that diode D? was forgot in the PCB, and was thus added in series to R120 with a "flying weld")

The power supply is regulated at +/- 15Vdc with two ordinary 7815/7915 ICs; input is taken from the raw DC supply (about +/- 40Vdc), with the help of a series 390W 5W resistor at the input; since the current is 40mA, the voltage at the input of the regulators will be about 25Vdc.
Two toroidal transformers are used (200VA each channel), but for the preamplifier section (ChL and ChR, +/- 15Vdc) the supply is taken only from the Left Channel transformer.

There is another AC input (the ones devoted to power the logic part), derived from a supplementary winding of the ChR toroidal transformer.

Well, nothing much to say about this preamplifier stage; it really does its work (apart from the input resistance problem), the bandwidth is more than necessary, the gain is adeguate.

On the contrary the power stage is difficult to understand!

Here we have again the same discrete FET stage, in place of the NE5534 original IC.
All the resistors are changed to higher values in order to be powered by a +/- 32Vdc supply, obtained with the help of a simple transistor stabilizer (made with Q107/Q108 and zener diodes) from DC.

But a detailed look at the currents reveal that the FET input stage is biased with an Id of only 150uA, about five times less than in the preamplifier section.
The following pair of transistors are biased with 3.7mA, and the last two at a higher value of 16mA; I suppose this is due to the very difficult load offered by the output power stage (very low, at least near 1kW, and highly non linear).
BTW the power dissipated by Q113 e Q114 is very high (0.5W), dangerously more than the maximum allowable (0.35W) for these TO92 case transistors.
Many complaints have been reported about this dissipation problem (after some time the amplifier become "dead"), and someone found a "home" solution.
There is a global feed-back network (R144, R140, C151, C152, C153) that sets the Closed Loop gain at 18.72 (DC gain=1) but even here, due to the low OL gain, the real gain is lower (about 16.6).
Note the presence of C155 (1n5) that acts as integrator, shunting a lot (if not all!) of high frequency signals, and overriding the global loop.
That means: at low (audio) frequencies the loop is closed in a standard way; at high (ultrasonic) frequencies the output power transistors are "not controlled", and the system is closed in loop with the aid of C155.
This brings the problem that the global frequency response of all the Puccini "remote" is not flat (like the old one), but has a -3dB point at about 40kHz, so -1dB at 20kHz and -0.3dB at 10kHz. Talking of phase, this is about 27 degrees of lag at 20kHz and 14 degrees at 10kHz.
Not only! All the distortion generated in the high frequencies by the output transistor (e.g. the crossover distortion) will not "killed" by the feed-back.
About C155 I made some tests, and the system seems stable (with whatever kind of load) even with a value TEN times lower (150pF).
I suggest to make some tests with 560pF anyway: the -3dB point will rise to 120kHz and there will be no risk of oscillations.
Do you remember the "Errors revealed" section in the Donizetti web page? Well it seems that someone has not read it!
In the Puccini "remote" still there are two capacitors (C156/C157) with the wrong value!
Take a look at the Puccini "remote" THD plot, compared with the original Puccini: as you can see the THD value is worse, even with the capacitors changed with the new 10uF value.

This is due to the low Open-Loop gain. The high frequency rise is due to the C155 capacitor local feed-back.

Whats left to say?
The declared output power is 60W. It is reached with a 230Vac mains; no more is possible due to the 32Vdc stabilized supply of the driving section. In case of a short circuit there is a current-limiting circuit copied "as is" from the
Aida integrated amplifier, even with two "non sense" resistors (R159/R164); these were used in the original circuit to obtain a divider (with other two series resistors), reducing the sensitivity of the limiter.
In the "remote" circuit the value of R160/R163 has been increased, lowering the maximum frequency the circuit will still work (see it as a low pass filter, made with R163 + C159). So doing only a "steady-state screwdriver short circuit" allows Q115/Q116 to conduct.
At the same time the photocoupler IC103 pulls down a pin of the microcontroller, opening all the relays and stopping the music.
Nothing happens in case of voltage clipping.

I'm not sure, but it seems to me that there is no DC sensing of the output. In case of transistor failure (short circuit), a DC current will flow through the speakers; what if the fuses don't blow?

That's all, folks.

Back to the home page