"Odilon"
a transformer-coupled 6DJ8 preamp
This line-level preamp has a moderate amount of gain, individual
volume controls for each channel, balanced inputs and outputs.  The
sound quality is excellent, and offers a performance that is quite
different to my ears than the typical RC-coupled, unbalanced unit.
schematic

The input stage uses a JT10KBD 4:1 step-down transformer from
Jensen because it presents a high primary impedance to the poor
output stage of your CD player, and has a wide bandwidth.  It is
loaded with a 5k audio-tapered
volume control.  This requires a 4.7k
resistor across it in order to lower the total resistance across the
secondary to the 2.43k that Jensen recommends.
the prototype Odilon preamp
The output stage uses the Lundahl LL1660S/10mA unit, wired as a 4.5:1 step-down.  This gives the preamp a low output
impedance without using feedback, and unlike cathode follower outputs the load presented to the tube remains fairly constant -
regardless of the load being driven.  A 1k loading resistor smoothed the high frequency response by reducing peaking and
damping ringing.  Other values can be experimented with, but 1k sounded good to me.  With it in, the LL1660S presents a 22k
load to the 6DJ8 which is over seven times rp'.

Both the input and output jacks are isolated from the chassis.  The circuit ground and ground wire of the power cord are
connected together to the chassis at a single point.  I used one of the power transformer bolts as the tie point.  This unit is
extremely quiet, and I haven't had any issues hooking it up to standard tube amps with direct, unbalanced inputs.  I have
however experienced some noise and low-level hum when using this preamp with a Soundcraftsmen MOSFET power amp.  So,
be aware that although the preamp itself is not generating the hum/noise, there can still be unfortunate interactions - most
likely due to the input stage in the solid state unit.  The transformer isolation does offer an improved listening experience due
to RF rejection, and the breaking of ground current loops.

Either a 6DJ8 or 6922 can be used, as the tube is run at a moderate 90V/12mA.

The unit is housed in a 6"x10"x2" Hammond aluminum chassis.
Front view.  No venting is necessary, as there are no heat
dissipating elements
inside.

Q: "I built it just like your schematic, but my voltages are off.  What did I do wrong?"
A: Voltage differences of up to 15% in most tube audio equipment are perfectly acceptable, and depend on your local line
voltage combined with the tolerances of the circuit components.  My line voltage here in Brooklyn is fairly consistently 124V.  If
yours is this high, or higher, you will need to add series resistance to the heater lines to drop the excess voltage.  I use Dale
wirewound 3W 1% resistors, 0.33 ohm in each leg of the 6.3V heater line, right before the 6X4.

Q: "Can I remove the cathode bypass capacitor from the tube?"
A: Yes, but at the expense of high frequency response, since the output impedance of the tube will increase dramatically.

Q: "What brand of tube do you recommend for this circuit?"
A: I'm using the Sovtek 6922, and I find it to be quiet, completely free of microphonics, and overall a good performer.  You
should avoid the Philips ECG JAN 6922 - all my examples were extremely microphonic.  I also rejected the JJ E88CC - the one I
tried was making odd popping noises.  Perhaps it was atypical.  I haven't tried any NOS european brands yet.
The 6X4 is an 80's JAN Philips.  It's dead quiet, unlike the early 60's Sylvanias I have that buzz and rattle.
Specifications
Input Impedance: 42k ohms
Output Impedance: 270 ohms
Gain: 4.7dB
Frequency Response: 0dB@10Hz, -3dB@36kHz
UPDATE: 11/09/05
I've been listening to the unit with a 2.2k load resistor across the output for a few days now, and I believe it sounds slightly
better than the 1k load.  It's a subtle difference though, and the frequency response is now bumpier, and square waves don't
look as nice.  It's definitely worth trying both, and is a cheap modification.
joeltunnah.com
Joel Tunnah, 2005