Meanwhile there were other preamp-projects on the web, with #27's and #56's and direct heated #26's. New to me was the idea to use output transformers, or CCS-load instead of a load resistor. So a redesign of the construction started.

First the power supply design was final: A tube recitifier, the 83 with a swinging PI-style (choke-capacitor-choke-capacitor) regulation section. Chokes are Hammond 193C (20 H 100 mA). The power transfomer is made by Trafo Service Baule in Germany, with 250V at 200 mA on the sec windings. Capacitors for the high voltage section are vintage Dayton paper-oil with 50 µF each, used parallel in pairs. Simulating this power section with Duncanamp's PSU Designer II software gave me a very nice result with a ripple close to zero! B+ voltage is about 245V.

First I've thought about building a massiv filtered DC supply for the heater voltage. Discussing this issue with experienced people on the web I've changed my plans for AC heating. With a 10 gang pot of 100 ohm connnected against ground (a simulated center-tab) it should be possible to have a low hum AC supply. I put my 3,15V heater windings in parallel, each with 4A current, and drop the voltage down to 2,5V by a fat alu housed resistor. I decided now to go for the #27 triode. What I found on the web is saying this has the best 'sonics' of the family.

For the line section I choosed a RC-coupled design again, with some strong influences of George Rigney / Air PhonoGram . George is building microphon preamps with #37 and #76 and did a lot of tests on his proto-board. So I decided to go with his recommendations. Coupling caps are 3 µF paper-oil. Also in the input line is a paper-oil cap of 0,33 µF. It is a very classical design, close to the recommendations in the RCA-10 tube manual (there for #56 triode). Resistors are mostly wirewound, or high quality metal MOX types. 

Volume regulation is be done by a 100k stepped attenuator (based on an ELMA 04A2A00 24t switch), input selection by a Dayton 4p4t rotary switch (which means that also the grounds of the input jacks are switched).

As usual the building of the amp was much more work than designing the circuit ;-) The chassis is made of beech wood with a stainless steel plate on top. Spends a lot of hours to drill about 90 holes in it. The amp made it's first noise on the 15th January 2001. It sounds really well. Amplification was at first a bit low, so I tried cathode bypass caps of 47 µF.  Plate resistor is 40k, cathode resistor was 2k4. Anode voltage was about 110V with 225V above the load resistor (that was using a 5Y3 rectifier tube). It is interesting to see the different descriptions for max. plate voltage of #27 tubes. As mentioned above I'm running 245 Ub+ and the meshy 227 are doing fine since more than 15 months.

Now I've changed the bias of the 27s. I'm using 9V Nicad cells. There are two options for battery biasing: connecting the cathode straight to ground, and connecting - of the battery to the grid resistor, an + to ground. Let's call this way grid biasing. The other option: just connect + of the battery to the cathode, and - to the ground. Let's call it cathode biasing. I've tried both options, and both still work. At the end I'm running battery cathode biasing, with about -9,8V at the cathode and 135V at the anode.
The amp is really quiet, without an input signal I can measure up to 10mV to 20mV AC on the outputs, which should be really good for those old tubes. Interesting to see is the AC is swinging a bit in a slow manner, about all 2 or 3 seconds, looks like the tubes are 'pumping'. I had a similar thing found at the anode voltage with +/- 0,5 DC before changing to the Nicad Bias.

Another option I have already tested: to use a Constant Current Source in place of the load resistor. But this try was not satisfying. From my opinion the resistor loaded version sounds more open, has more dynamic.

The line amp has excellent sonics, things I have never heard  with miniature triodes. It is still different from a more modern design with  miniature triodes, you can not compare it really. The mesh plate Cunningham   327's are glowing fantastique: one can see the red glowing cathode through the mesh plate. But on the other side this tubes are really microphonic, and still nervous without an input signal: one can hear 'plings' throught  the speakers coming from inside the tubes. So I was firing standard ST shaped 27's until I got other very interesting tubes.

My #27 triode test results are:

• Cunningham 327 globe shape/mesh plate: very nervous, extremly microphonic
• Sylvania 27 ST: good sonics, but microphonic
• RCA 27 ST: calmest 27's I have
• National Union 227 globe shape/mesh plate: very good sonics, microphonics are acceptable, the strongest 27's I have
• Arcturus 127 blue and globe shape: pretty babies, tipped (!), but very noisy and very microphonic
• RCA UY-227: globe shape/mesh plate - to be tested
                         

This is the schematic of the finished amp. I've added only an industrial build time delay circuit to give the 83 rectifier tube 30 sec. heating time before the high voltage is switched on. For mercury filled rectifier tubes this is really important.