The return of the 01a stage

I remember my first 01a pre-amplifier to be one of the best sounding ones I ever had. The uniqueness of its tone, detail and clarity was astonishing. Perhaps it is due to the warm tone it provides and I guess this is the reason why Thomas Mayer branded his design as the “sound processor”.  I fell in love with the sound of a CX-301a and the joy of listening to this stage was so great that I found a fantastic excuse now to re-build this stage. My Starlight Discrete DAC has a very low output due to the step down transformer it has. I can only get 500mV as maximum output level. Not enough to drive my system to full level.  This was a perfect argument for me to look at building a simple amplifier stage that could add the sonority of the 01a in my system.

 The concept was very simple. I wanted to maximise gain whilst minimising the number of additional components. The transformer coupling was discarded as a proven design sprang to mind to achieve the maximum gain and reduce size of the preamp.

Firstly, I looked at the curves again when filaments are starved. I found the 01a microphonic noise to be reduced significantly in this condition. I looked at my stock of 01a and selected a pristine globe SX-201a made by Sylvania. This measured really well:

26/10/2014 13:26:39 uTracer3, GUI V3.11.6 Triode Quick Test
SX-201a Silvertone Vf=5V, If=250mA
SECTION 1
 
Test conditions:
Va : 136.7 (V) Swing +/- 2.73 V (2%)
Vg : -9 (V) Swing +/- 0.18 V (2%)
 
Test results:
Ia : 3.09 (mA) 103 % of nominal 3 (mA) 
Ra : 10.7 (kohm) Ra = dVa/dIa
Gm : 803 (uA/V) Gm = dIa/dVg
mu : 9 (-) mu = Gm*Ra

Then, I dialled the filament current down to 200mA:

26/10/2014 13:36:57 uTracer3, GUI V3.11.6 Triode Quick Test
SX-201a Sylvania If=200mA Vf=3.4V
SECTION 1
 
Test conditions:
Va : 136.7 (V) Swing +/- 2.73 V (2%)
Vg : -9 (V) Swing +/- 0.18 V (2%)
 
Test results:
Ia : 2.51 (mA) 84 % of nominal 3 (mA) 
Ra : 14.04 (kohm) Ra = dVa/dIa
Gm : 588 (uA/V) Gm = dIa/dVg
mu : 8 (-) mu = Gm*Ra

As expected the emission was reduced and anode current diminished by 16%. The anode resistance increased from nearly 11KΩ to 14KΩ (+26%) and the transconductance reduced from 800 μmhos to 588 μmhos (-27%). The mu is expected to stay the same however the rounding effect of the uTracer shows a difference however is close to 8.7-8.8.

The saturation of the valve given the low emission is clear when tracing the curves:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

The SPICE models can easily be derived from the above set of curves:

 The preamp design

To achieve the additional gain I need on my system, I opted for the classic hybrid mu-follower I used many times before. This topology provides great drive and lower output impedance given its push-pull nature and maximise gain, close to mu. The cascoded FET topology also improves the PSRR. I will tap the HT from the 4P1L line stage I have which gives 235V of raw DC.

The filament resistor was increased to 20Ω from my previous design given the reduced filament current. I aimed to bias the valve at 3mA for best sound as per my notes:

 

The anode voltage should be 100-104V depending on how the cathode point is derived from the filaments. In practice, the anode current will be adjusted by Rpot part of the high-impedance voltage reference made up by M2 and R3.

This stage has very good HF response given the mu-follower topology and also the low capacitance of the 2SK170 chosen in the lower device instead of a classic DN2540. The low frequency response is given by both C2 + output impedance and C1.

Quality of C1 and C2 is paramount. I’m using Russian PIO NOS I have at hand and sound very nice. You could spend whatever money you want on “boutique” capacitors, which is not my cup of tea. Equally, you want to keep the value of C1 and C2 as low as possible to minimise the impact of these capacitors to the sound of this stage.

Next step is to build this in a breadboard and listen to it!