Nessun Dorma! (G. Puccini)

Nessun dorma!... Nessun dorma!...
Tu pure, o Principessa,
nella tua fredda stanza
guardi le stelle che tremano...
d'amore e di speranza!
Ma il mio mistero è chiuso in me,
il nome mio nessun saprà!
No, no, sulla tua bocca lo dirò,
quando la luce splenderà!
Ed il mio bacio scioglierà
il silenzio che ti fa mia!

(Il nome suo nessun saprà...)
(E noi dovrem ahimè, morir, morir...)

Dilegua, o notte!  tramontate, stelle!
Tramontate, stelle!  All'alba vincerò!
Vincerò!  Vincerò!

News about circuit theory at the end of this page !!


Puccini integrated amplifier

The intent was to create a "hi-end entry level" object;
there was not much time (it was the end of 1995, the WCES starting at the beginning of January), so we decided to have a look around us, and we found some interesting diagrams.
The first was published on Electronics World (formerly Wireless World) in July 1993 by Marco Corsi, application engineer working with Texas Instruments: it was a simple but sound diagram with one operational amplifier and a pair of complementary darlingtons; obviously the output power is limited by the operational amplifier's power supply (+/- 22V in this case, with a TLE2141 as operational amplifier), which is about 25Wrms /8Ohm.
Diagram of Marco's amplifier
Part list
To overcome the limitation of the low power output, we analysed the second diagram that was in the same Electronics World issue, under the circuit ideas section: Nick Sukhov explains how to increase op-amp output swing by dynamically driving the power supply with the signal.
Diagram of Nick's op-amp
We then thought of "putting together" these two diagrams, in order to obtain more power.
A more theoretical approach was found in the June 1994 issue of "Linear Technology Magazine", in which Dale Eagar wrote an article with the title "Extending Op Amp supplies to get more output voltage". In this circuit it's possible to swing in DC, not only in AC, as in Nick's diagram.
First page of Dale's circuit
So we had everything in our hands to begin the first prototype of the "cheap integrated amplifier" (the name "Puccini" was chosen just a few days before the Winter Consumer Electronic Show).
But while working at this project, we had time to give a look at some similar products; one of these seemed to us to be using the same concept of the diagrams we had studied: here is the diagram.
Well, this last one still had some limitations: it was terminally instable, and the output power was just a little bit more than 30W.
Going further, there was no loudspeaker protection! Anyway it was a good starting point.
After one week the "version zero" was on the kitchen table, ready for shipping to Las Vegas.
We had only ONE prototype that after the show returned to Italy and toured all around the country, visiting all Italian hi-fi shops.
In spring 1996 the first batch of 100 Puccini was ready to be sold.
After one year we received the first "happy news" : 
First of all some data about Puccini:
Voltage and Current in a bi-logarithmic plane
The "conventional" output power is 40W over 8Ohm load, two channels driven.
Having no current protections, there is an intrinsic ability to drive whatever type of load you can imagine, down to about 2 Ohm, if no steady state signal is used (i.e.: only music, no laboratory tests).
The drawback is that in case of a short-circuit of the output, the fusion of the power darlington is quite immediate !
We developed a protected version, used in some "Aida" prototypes (Aida is an "Opera loudspeaker" amplifier, nothing but a Puccini S.E. with minor modifications).
In case of output device failure, there is a protection circuit that "cuts-off" the loudspeaker line; the same circuit is used as an anti-bump at the start-up, and as a "stand-by" effect when the volume knob is turned fully counter clockwise (there is an infra-red barrier that a screw of the volume knob cuts-off).
Other relevant data are:
It's now time to download the zipped poscript (603KB) file of the complete Puccini amplifier.
A deep analysis of the diagram can start from the left channel pre-amp: the first stage is near 10x (20dB) at low noise stage, and the 15Ohm/6n8 RC group is for RFI purpose; the load presented to the MC head is 115 Ohm.
If the MM jumpers are selected, the load presented to the cartridge is 47k, in parallel with 220p; the 1k series resistor (R48) is again for RFI use.
The RIAA network has a center frequency (1kHz) gain of 40dB, and a low frequency roll-off determined by the C2 value.
If you want to implement the RIAA/IEC curve (-3dB@20Hz), change C2 to 100uF.
After the selector switch there is a buffer stage, with a decoupling filter (C10/R12) and another low pass filter (R46/C48): the first sets the -3dB point at 1.6Hz, the second at 1.6MHz; the input impedance is 50k. The out of the buffer (half of a dual TLE2072 op-amp) goes to the tape-out (via R42) and to the volume pot.
Here we have a story: at the beginning (first two or three hundred Puccinis) the pot was a Phier type, but some people reported an unsatisfactory tracking of the two channels; we made some tests, even with some Iskra types, and we decided to change to a more expensive, but more reliable ALPS pot.
It could be a good idea to have a look inside your Puccini: is the pot an ALPS ?
After the pot there is the gain stage (about 10dB), implemented with the other half of the TLE2072; this ends the pre-amp stage.
Having a look at the power left channel stage, you notice FOUR output power devices; they are TIP124/TIP147 Darlington, but only a couple of them are present in the standard Puccini; the other two are reported only for layout facilities (the PCB is the same for Puccini and Puccini Special Edition).
This is the FIRST prototype of Puccini, presented to the WCES 1996.
About the diagram, the entire amplification factor is due to the NE5534 op-amp, configured as integrator with a dominant pole.
The whole amplifier is non inverting, like the pre-amp stage; there are another couple of filters at the input (low pass with C11/R16, and high pass with R49/C70), the first of them being the dominant pole of the entire integrated amplifier.
The out of the op-amp directly drives the power transistors; the 0.33 Ohm emitter resistor has been chosen to ensure a good thermal control, together with the little BC517 darlington, directly connected to the heathsink.
The R31 (56 Ohm) is a trickle for having a local feed-back on the thermal behaviour, as Douglas Self explains in his book "Audio power amplifiers".
The output signal goes through the coil L2, used to limit the request of current in case of capacitive loads; the C35/R37 pair ensures a load even at high frequency, giving more stability margin to the amplifier.
Some of the output signal is feed-back to the input of the amplifier, and the global gain is [1+(R19/R18)], that is near 27dB.
R17 ensures the same impedance to be seen from the op-amp to both the inputs (39k), C27 bypasses its effect at the audio frequencies.
Another part of the output signal is used to drive the "floating" power supply: with no signal present, the NE5534 is powered via R24/R25, and stabilized via D6/D9 at a value of 16V; when a signal is present, the common point of D6 and D9 is driven via R21 (220 Ohm), and the power supply "follows" this signal up (or down), until is doubled (+32/-32).
THD vs Frequency param. Load; not too bad at all.
The behaviour of the power amplifier is typical of a project with a "not so moderate" feedback factor (i.e.: Open Loop gain MINUS Closed Loop gain).
This is a PSpice simulation of the Feedback factor, not a measured one !
Another important study has been performed on the optimal idle current; a PSpice simulation shows how the linearity of the output devices increases as the idle current does so, but there is a limit with the heat dissipation and with the behaviour of the thermal run-away.
The 20mA idle current is a good compromise for the Puccini
The thermal feedback has a set-up time of about one minute, as is visible in this test
Another relevant fact is the value of the four DC power supply fuses: at the beginning it was 2.5A-Fast, but after some field tests it has been modified to a value of 3.15A-Fast.
The last part of the diagram concerns the protection circuit (based on a LM393 window comparator), the stand-by circuit (an NE555 delay circuit and a relay that cuts-off the NEGATIVE side of loudspeakers), and the power transformer: this is a toroidal 150VA, with dual primary (115+115Vac) and dual secondary (2x50Vac, each one center tapped).

Update theory:
October 2000: surfing on the net for another kind of circuit, I've found this PDF about bootstrapping of an op-amp.

Written by Grayson King & Tim Watkins (Analog Devices Inc.), this 8 page document explains in a very detailed way the "how" and "why" of this technique.
The article ran on page 117 of the May 13, 1999 issue of EDN, although here I suspect there is an error in figs. 6&8.

Back to the home page