Friday, November 30, 2012

Project: The Command 1625 tube amplifier

In the world of single-ended amplifiers, the venerable triode still gets preference.  It's easy to see why since the 300B, 2A3, 845, and 211 - among many others - has a certain catchet and a sonic signature that is honestly very hard to beat.  The triode is quite the linear device with the downside of needing a bunch of voltage and current to make relatively low power.

The other option when running single-ended, minus the odd solid-state design, is to use a pentode/beam-tetrode in ultralinear or native operation.  This not only gives plenty more power, but these tubes also have the benefit of being inexpensive to buy.  Based on my experience with various SE amplifiers, most of the magic of single-ended is the circuit simplicity - no phase-splitter - and the Class A output stage.  Now this design I'm about to introduce was not meant to be an assault on the very best amplifiers, but instead offer a budget alternative to the sometimes stale and now very mundane 300B amplifier.  It's also an attempt to replicate 1930/1940s technology' but with modern components.  Much of the 'sonic footprint' of those pre-war years was not from inferior amplification, but the bandwidth limited source material. 

Of course running in pentode mode requires some negative feedback - a new fangled concept back then - in order to reduce the output impedance to something that modern speaker can use.  An alternative would be to experiment with Schade feedback - taking voltage from the plate of the output tube and tying it to the plate of the input tube. 

The first decision was tube choice.  If I wanted to be just a cheapskate, there are plenty of TV tubes that could be used.  Instead I went with a vintage classic that is still widely available - the 1625 output and 12J7 driver tube.  (Note that the 807 and 6J7 can be used - just don't use the voltage doubler for the filament supply.)  Expect to pay $5-$10 per tube, which is a bargain considering the quality of these vintage beam tetrodes.  Also note that the 1625 has a low screen voltage maximum, so running ultralinear isn't an option unless one can live with reduced voltage and the resulting lower power output.

The UX7 base top-plate cap 1625 is part of the famed 807/6L6GC family and was used in WW2 to provide radio communications - the AN/ARC-5 Command transmitter - for the majority of the planes in the United States Army and Navy.  This tube was made to military specifications with wonderful assembly and a high quality control - when it came to radio pilots lives were on the line. Since there was a war on, scads of them were made and every example I've seen is an impressive piece of engineering.  In tribute, I have given this project the name 'The Command 1625'.

The grid-cap 12J7 pentode is the 12V filament version of the 6J7 tube which is the octal version of RCA's earlier 6C6 tube, which in turn was their answer to Western Electric's famed 310A.  These tubes take the signal voltage and bring it through the cap on top of the tube - a nice way to keep the fragile signal away from noisy power supplies.  The 6J7 eventually begat the 6SJ7 which eventually became the smaller EF40/6AU6/EF86 types.  The 12J7 was selected for the fact that NOS versions are still cheap and plentiful in metal or glass versions.

(right click to open and see larger version)

A quick look at the signal schematic shows nothing too exciting, though using a pentode on the front end is considered a heresy these days of current sourced triodes and LED biasing schemes.  My idea was to keep this design as simple as possible - in the spirit of the vintage tubes selected - with minimal parts count and matching required.  Eagle-eyed reader will note that this same type of front end was used on my EL156 amplifier.  It's a good sounding front end so it was easy decision to use it again.

The feedback resistor for a 4-ohm output transformer is 6.8K.  For an 8-ohm secondary, increase the value to 8.2K.

(right click to open and see larger version)

For the power supply I stuck with tube rectification which has less switching noise than solid-state and better visual aesthetics.  The 5Z3 - a precursor to the 5U4G - was selected on looks and the decent prices, plus the ability to easily handle the current requirements for two output tubes.  Plenty of other options are available here.  Two Triad chokes and a smattering of generic capacitors provide the raw filtering for the output tube plate.  Of interest is the regulation for the screens of the 1625 and the input stage.  This stiff front end gives more control and separates it from the wranglings of the output stage.  For easy implementation, shunt regulation is accomplished with simple VR tubes: the OD3 and OC3 are put in series for 255 volts.  The small bypass cap here is just to clean up noise - going for a larger value will cause instability with these types of regulators.

The voltage doubler circuit is used to take the 6.3VAC filament tap and boost it up to ~12VDC required for the 1625 and 12J7 tubes.  Note, do not ground the center-tap.  If one wishes to use 807s, 6L6GCs, 5881s, or KT66s along with a a 6J7/6SJ7/EF86 input tube, the voltage doubler can be skipped.  In this case, do ground or lift the center tap of the 6.3VCT winding.

Some of the passive-parts decisions were based on what I already had in stock.  For example, the PEC volume control, Wima coupling caps, and power supply electrolytics were chosen because I just happened to have them in my stash.  Builders can select their own goodies, provided they meet the electrical requirements of the circuit.  Other than that, I went with Kiwame and carbon-composition resistors, Nichicon cathode bypass capacitors, and a few Solens.

For the the output transformers I went with the Edcor CXSE model.  Alternatives are James, One Electron, or even Hammond.   For those on a tight budget, there are cheaper alternatives, but these will limit bass and treble response, not to mention that the core will be taxed more heavily.  I've found that overkill output transformers lead to a more relaxed, transparent sound.

Top aluminum panel is from the steadfast Front Panel Express.  Size was selected since I already had a leftover wooden chassis from an older project.  Layout was done to minimize power and signal circuitry interaction, while also giving a vintage "juke box" aesthetic.

Potential modifications:
Different Output tubes: With socket changes and 6.3VAC from the transformer instead of the voltage-doubler filament circuit, the 807 or any of the 6L6GC family can be used - 5881, KT66, 350B, etc.

Different input tubes: With wiring changes: the 12SJ7.  With filament and/or socket and wiring changes: 6C6, WE310A, 6J7, EF86, EF80. 6AU6, EF37, etc

Different rectifiers: 5U4/G/GB is the Octal version of the 5Z3.

Listening Results:
Turning on a new, freshly built amp is always an exciting moment.  With the test speakers - Pioneer SB-21s - place on the ground, I was immediately taken with the fast quality of this amplifier.  After checking over the power supply voltage and bias readings, I moved on to my main system consisting of Spica TC-50 speakers, a Quicksilver full-function preamplifier, and a VPI HW19 Mark III turntable with a Denon DL-110 cartridge.  This is a smaller, more budget orientated system than the setup at my old digs, so I'm still feeling my around the various changes.

First impressions - revealed an amplifier with a nice, neutral midrange.  This is not a romantic sounding unit but I would never mistake it for solid-state either.  Bass quality was very good: controlled and with no tubbiness.  This is hardly a surprise due to the heft of the Edcor output transformers.  The already mentioned speed was evident in the treble too - giving a slightly lean presentation which may be more a character of the Spica TC-50s than anything else.  Soundstaging, also strong point of these speakers, was wide and deep.  Texture and inner-detail was also top-notch.  With more dynamic music, the amplifier had no problem delivering the power while keeping everything in line.  This amplifier sounds much more powerful than its expected 10Ws, indicating good clipping characteristics common with Class A output stages.

Overall, I would still give the overall nod to the EICO HF-60s, but those monoblocks usually cost three or four times as much, and also sport much more exotic output iron plus increased power output.  Having said that - and it's always difficult to fairly judge your own creations - the Command 1625 amplifier offers great sonics for such a low price point.  It is also a great reminder that pentode outputs can compete in the world of hi-fi.

Parts list and Front Panel Express file are available upon request.  Shoot me an email.