This modification of the Vanderveen MCML05 pre amplifier is totally for free. You have to buy the few components yourself, there is no PCB, it is very simple and I totally leave the implementation to you.
My pre-amp has a recorder output for a recorder or a mixing desk. The output is taken from the selector S-3 via a 1kOhm (R41) resistor, see figure 1.
photo-2: the situation before modification
This solution is very simple and is based on the high input impedance (environment of 100 kOhm) of most recorders. This impedance will hardly load the signal sources like a CD player and a tuner that are connected to the MCML05 inputs. However, I noticed a severe loading under special circumstances, largely distorting the sound. The new buffer is designed to prevent this situation.
Modern recording equipment and mixing desks use integrated circuits nowadays. They do not load the recorder output as long as the recording equipment is switched on. However, when the equipment is switched off, a nasty effect occurs. The IC-inputs are protected by means of diodes to the positive (Vcc) and negative (Vss) supply rails. In switched off condition, Vcc and Vss equal 0 Volt, and then the protecting diodes effectively load the signal source. Figures 2 and 3 clearly show what happens; the diodes create severe distortion.
figure 2: protecting diodes limit the input signal
figure 3: measured limiting of the input sinus;
In my studio, I always disconnected the recorder output to my mixing desk when the deck was switched off. But one day I forgot to do that, with the final result that my client did not like the sound of my amps at all! I lost this client and such a nasty situation should never happen again. That's why I decided to design a buffer circuit that totally separates the recorder output from the input selector. In this article I discuss this buffer circuit.
I tested several solutions for this buffer, like a FET source follower, but I found too much distortion. A valve cathode follower could work, but the filament current demand was too large for the power supply, and it sounded too "technical". The follower also asks for coupling capacitors, which I did not wish to use. So, these solutions were no good.
I decided to use an integrated circuit, then I could meet all the conditions. I selected the OPA2134, which is much better than the standard TL072. However, Douglas Self indicated in his fantastic new book "Small Signal Audio Design" that I might have used the NE5534 as the best solution. I totally believe him, but I had OPA in stock and NE out of stock. By the way, this book of Douglas is of high interest and importance. I advice to buy it, very good and reliable and valuable information!
The buffer circuit
photo-3: the modification replaces the two R41 resistors
photo-4: the modification seen from the top
Explaining the new circuit
Yes, more is going on, although it is hidden. Notice the low resistances of R1 and R2 for little resistor noise. R1 and R2 are almost equal, especially if you take into account the low output impedances of modern audio equipment. For lowest distortion the following condition should be met: R1+Zout = R2. In the manual of the OPA2134 and in the book of Douglas Self and in my own measurements it appears to be of the utmost importance to meet this condition. When met, the total harmonic distortion at least stays smaller than 0,005% at 20 KHz with 1 Vrms in a 10 kOhm load. Figure 8 shows the THD distortion of my computer soundcard measurement system, using ARTA and STEPS (www.fesb.hr/~mateljan/arta. The new buffer circuit actually distorts less, but I can not measure how much less. Then I should use a AP which measures down to 0,0005%, but I happen not to own that fantastic measuring beast.
Suppose you select with S-3 the turntable input. Then you have to deal with the output impedance of the valve RIAA pre amplifier. Its value is around 2 kOhm. Then R1+Zout = 3 kOhm, which is twice R2. Is the lowest distortion condition still met? Not totally, but THD stays below 0,005% and that is good enough for me; at this moment I can't do better. With the given resistances of R1 and R2 almost always the optimum is reached or you are close to it.