Quad ESL AC Voltage Regenerator

this project / page is under construction

Table of Contents



There is a phenomenon in conjunction with my Quads that I have observed for many years but have only understood in the last few weeks. The loudspeakers have a “day form”, they are sometimes quieter than usual and I have to drive more level to hear the same perceived volume. For a long time I considered it to be my personal daily condition, but the investigations of the last weeks showed that it is the loudspeakers themselves. The reason is the varying mains voltage (allowed: 230V/AC ±10%) which has a direct effect on the bias voltages of the foils. Different bias voltages cause changed deflections – and thus volume – of the Quads. To eliminate this effect I developed the RStAudio ESL AC Voltage Regenerator.

Investigations with Spice


To confirm my suspicion I have once again Spice attempted. With this software it is very easy to determine how the actual conditions on the foils should be.

My 4 Quads are designed for 220V/AC and bring this voltage by transformer to 610V/AC. Then the voltage is multiplied by Villard cascades to a DC voltage of about 1500V/DC for the high frequency panel and to about 6000V/DC for the two bass panels. A first simulation with these parameters calculates the voltages at the loudspeaker modules as provided by the manufacturer.

Direct Voltages on the Foils at 220V/AC

Direct Voltages on the Foils at 220V/AC

This results in foil voltages of 1467V/DC in the high frequencies and 5835V/DC in the bass. These are the ideal values specified by the manufacturer. Of course there was also a tolerance band of the mains voltage at the time of the development of the Quads 57 and I think at that time it was ±10%. But I claim now that the mains fluctuations at that time – at least in country areas – were not as big as today.

Next I took a look at how it looks like with an ideal mains voltage of 230V/AC on the foils.

Direct Voltages on the Foils at 230V/AC

Direct Voltages on the Foils at 230V/AC

The simulation gives 1534V/DC and 6109V/DC. An increase of 67V/DC in the treble and 274V/DC in the bass. This is about 4.6% more voltage than intended and is still within the tolerance band of 220V/AC ±10%. The operation of my Quads at ideal 230V/AC will be quite permissible.

If we look at the voltages in the tolerance band of ±10% around 230V/AC, we get the following picture:

Direct Voltages on the Foils at 230V/AC ±10%

Direct Voltages on the Foils at 230V/AC ±10%

The two middle curves (blue & green) are the results at 230V/AC mains voltage and the two outer curves are the voltages at 207V/AC (-10%) and 253V/AC (+10%), i.e. at maximum utilization of the permitted tolerance band of the mains voltage. The following table shows a summary of the resulting voltages:

Mains Voltage
220V/AC230V/AC - 10%230V/AC230V/AC + 10%
High Frequency
Low Frequency

As can be seen from the simulations, the actual film voltages depend on the voltage currently applied from the grid side. In my case there are voltage fluctuations of more than 10V/AC in the last hour before I wrote these lines. Depending on whether the sun is shining or not, completely different basic values occur – around our house there are a lot of photovoltaic systems installed.

Now, however, the applied foil voltage has a direct effect on the resulting deflection at a defined input signal. The result is a basic volume of the Quads 57 that changes with the mains voltage. This is not surprising if you consider that the foil voltages are generated directly from the mains voltage without any regulation.

Since I don’t have models of bass and treble panels, the loads on the cascade are realized by resistors. Thus the current flow in my simulation is directly linearly dependent on the voltage – Ohm’s law applies. But I think you can live with this limitation and it does not reduce the results.

The EHT units of my revised quads are no longer connected to the secondary 610V/AC winding, but to 590V/AC. At least my 4 Quads provide this voltage additionally. So you get a much better adaptation to the intended operating points at 230V/AC than with the original wiring.

Voltage Regenerator


So what can you do to achieve stable conditions on the Quads? The first thing you think of is to regulate the DC voltage – but with the high voltages this is not very easy and would be a direct modification to the Quads themselves. Another way is to develop electronics that provide the Quads with stable conditions independent of the actual mains voltage at the power supply. This way I have chosen. On the one hand the electronics is much less demanding and on the other hand you can connect an external device in front of the Quads.

The circuit is a 50/60Hz sine wave oscillator with an amplifier stage including output transformer.

The frequency signal is generated by a DDS chip (AD9833). A 1MHz quartz oscillator is used as reference clock. The DDS chip has to be initialized after power on and so an additional microcontroller is needed. I decided to use an Atmel AT89C4051. This controller also initializes the digital potentiometer which is used to set the output voltage.

After the DDS chip, the sine signal is amplified, filtered and freed of its DC component by servo controller. This processed signal is fed to the power amplifier via the digitally adjustable potentiometer. I decided to use two integrated audio amplifiers LM1875 which are connected to a bridge power amplifier. The transformer winding is therefore connected between the outputs of the two amplifiers.

The output frequency can be switched between 50Hz and 60Hz via a DIP switch. In addition, 4 output voltages can be selected via DIP switch.

The circuit naturally also requires an operating voltage. I have designed this in my normal complex circuit technique. After the line filter there is an additional filter with MP2 X2 and Y2 capacitors. A DC filter for the 230V/AC voltage is connected to it. Only after these 3 filters the primary winding of the 25VA toroidal transformer is connected. On the secondary side there are first of all snubber networks before the AC voltage is processed by a discrete rectifier with ultra-fast soft recovery diodes. This is followed by a symmetrical C-L-C filter. The unregulated operating voltage is then generated by capacity multipliers. These voltages supply the two audio amplifiers. Additionally there are ±15V and 5V voltage regulators for the supply of the operational amplifiers and the digital electronics.