March 2011

Aikido Amplifier...
Or how to twist your electron's wrist.
Article By Jeff Poth

Difficulty Level

  Normally one might expect such an article to begin with the definition of Aikido. If you don't know, infer or Google it, slacker! Instead I'm going to tell you how I wound up taking nearly five years to build my best preamp to date.

5 Years!!!

My good buddy gives me the octal board as he decided to go noval (Thanks Mark!)

I stared at it for 6 months and sourced a DIYenclosures chassis (Thanks Mike!)

I stared at it for 6 more months and sourced the power supply iron for a 6SN7 build, 2 Hammond chokes (193C), and a Hammond transformer (273DZ)

I realized that my iron would be a little too much for the top plate and decided to laminate a second piece of aluminum beneath it.

I worked on other projects and let it sit 3 years (!!)

In mid 2010 I began populating the board with resistors, after deciding to go with 12SN7 instead of 6SN7 due to less expensive NOS tube options. Other than the heater, the tube is the same. Snagged a 25.2 VCT 2A trafo from RadioShack for the job (it's still in-store at the time of this writing, and is one of a very few transformers available for the job, and is fully enclosed which is nice for cosmetics)

I brought the board to a friend with a milling machine (Thanks Scott!) and we laminated the aluminum together using contact cement, and milled the tube and transformer holes. 

Once all this was done, I was fully in build-mode, and could roll ahead with active production, Hooray! Shown is the top plate with the iron mounted. You can see the precisely placed tube holes, designed to allow my 'sprung' suspended board to have the tubes just barely peek out the top of the chassis, and have good ventilation. 

Some slop in the wiring and whatnot as it was still in testing phase, but you can see the heater regulator board, the B+ supply (6AX5, LCLC), and the mounting arrangements. A great deal of effort was put into the 'sprung' suspension, as tubes tend to be microphonic. The well nuts isolate the brass threaded rods from the top plate, and there's a grommet under the plywood (0.5") board that the circuit board is screwed to. Rubber grommets and O-rings are used to damp the mounting points. All this damping ensures that minimal chassis vibrations find their way into the tubes.

So, all's well and we're ready to smoke test! The power supply, loaded with a 5k resistor, fires up properly, the heater regulator works correctly, loaded or not, and we're ready to hook up the circuit and give 'er a spin, right?

The Magic Smoke Came Out!
Needless to say, I was displeased. I replaced some scorched resistors (lucky me, I have AB & B electronics on the corner, an old-school TV repair shop with assorted goodies, and a TUBE WALL!!!!!). Another smoke test, another fried pair of resistors. Naturally, in between smoke tests, I tested all circuit values and everything was good. Further inspection lead me to realize that the darned Chinese tube sockets had their pin labels rotated! ARRRRGH!! I bought those years ago and didn't even remember where, to let them know they had a problem. So, desoldering and resoldering of 32 tube pins. FUN!

Replace the resistors a third time..... Triple check all my work... And... We have...

SMOKE!!!

Same resistors, same problem. As any experienced builder knows, this type of thing gets you into focus- there's no letting up when you're so far into a build and something's off. I began tracing the circuit, and came to an unpleasant realization- the online manual was for the current version of the board, which had resistor number changes. At this point the board had been reworked a number of times and I wasn't real confident in the rest of the resistors besides those that smoked. I contacted Mr. Broskie, the gentleman behind the Aikido circuit, and he generously provided me with a replacement board and parts. Very good show mate!

Another Delay?!
So we have a new, improved, octal aikido board (adds a last stage capacitor to the cct). We have parts. We have problem. The socket locations have changed! The older board was very long, and had an uneven gap between the four tubes -- the gap between each pair (channel) was smaller than the gap between channels, and the new board was spaced equidistant! NOOOOOO!!!!! My precisely drilled holes were for naught… it is a jigsaw time. 

So, I rebuild the circuit, adapt the layout and power supply for it, and forge bravely ahead. A successful smoke test and it's time to drill and mount RCAs, wire up the input wiring (a big job in any preamp), and get 'er ready for signal. Further, the wood mounting had to be changed to be ventilated (to take advantage of the new ventilation holes in the circuit board). You can see the rough drill-throughs and sloppy finish, only intended to seal the wood. This board is wholly contained within the chassis and thus is invisible in use. In retrospect I probably should have taken more effort on making the mounting board pretty since you have now seen it.

Some Details
For signal wiring I used twisted pairs of DH Labs "wrapped" silver and Cardas "wrapped" copper Litz (ground). These are fed through tubes of heatshrink (left unshrunk) to give a little spacing between wires. They're routed and bundled into place carefully with zip-ties to ensure that they have minimal crosstalk and capacitance. You can see that this bundle is routed near the PS, but it's got about an inch in any direction between the wiring and any of the power supply components. There are 2 sets of outputs, one with a 15uF 400V Solen capacitor (sub, or high impedance headphones), which is connected via RCA. The second output is with a hard-wired pair of 1-meter cables, and plays through a 2.2uF Cardas cap. That's the main amp hookup. The hardwired cables are a pain, but the best connector is no connector. In that way there's also a 1-meter pair of ICs for one performance-critical input position, and four RCA inputs. You can see the white terminal strips, which make great, more minimalist (and better sounding) alternatives to RCA connectors, while still allowing you to remove them later if needed. 

Source selection is performed with a Russian military surplus silver plated rotary switch. four poles, so source grounds are switched as well as signal inputs. This ceramic beastie takes a lot of torque to rotatate and would likely support 3A to 4A through the very beefy connections. Good thing I have a giant knob to use (again, thanks Mike from DIYenclosures.com). At the time of this writing the potentiometer is an Alps "Blue Velvet" stereo pot in 100k log but it's a likely upgrade soonish, to a stepped attenuator. 
The heater transformer is run through a fullwave rectifier and Broskie's heater regulator board, which supplies nice clean stable DC for the tubes to run with. A voltage divider biases the heater to about 75V. I originally used the old "mount on grommets" trick for the filament (heater) transformer, but the trafo ran hot. By mounting direct to the chassis, it has a large 0.25" thick aluminum heatsink via the top plate to help dissipate the heat. Oh well, it's not noticeably vibrating anyway.

So... What Is An Aikido?
Well, I'll tell you what the circuit is. The aptly named circuit leverages power supply noise against itself, and minimizes any negative power supply related artifacts. The typical Aikido circuit (there are many variants) has a two triode voltage amplifier section, and a two triode cathode follower section, both utilizing John Broskie's optimization tricks. The basic voltage amplifier/cathode follower arrangement has been used in many preamps, but there is more than that here. Amongst Broskies tricks are using power supply noise against itself, and using one triode's nonlinearity (within a single tube) to counter the nonlinearity of its partner triode. It's a very elegant circuit, not really minimalist but pretty low part count, and more importantly, it's well thought out. There is extensive discussion on the circuit on tubecad.com, and in the various online forums. Between the two tubes per channel worth of B+ and heaters, the preamp draws about 60W of power. Doesn't sound like much, and really it isn't, but you will be putting more power supply into the beastie than you might expect. My build wound up being about 80% as much weight and tube as a Dynaco ST70. 

Performance
I don't expect to build any more preamps for a very long time. I can say with complete confidence that this is the finest preamplifier I've owned, and naturally that includes those I've built. This is not a finicky preamp, but a workhorse that will drive whatever amp you wish. I have it driving a little Gainclone (LM3886 based OpAmps amplifier) and my Dayton SA1000 subwoofer amplifier, off the two sets of outputs. I've built many preamps over the years, including the ancient "Love Kitten" circuit I did for "Valve" a decade or so ago. Upon replacing my 6CG7 linestage (my design), I noted a significant increase in bass authority and a much more open top end, as well as better transparency and instrument delineation. The imaging suffered slightly, very likely due to the combination of a lesser attenuator (the blue velvet, vs. the ladder attenuator in my 6CG7) and the shared last-stage power supply (these are separated in my 6CG7). The noise level is extremely low (I'm running 97dB horns, so that's an important piece), there's plenty of gain (this will be determined by the tubes chosen). With the vast array of options available to you from glassware you should have no problems finding a kit that suits your specific needs. I like octal 'cause they've got nice big plate structures, but with a quality circuit like this, over specified for any real-world linestage use, any tube choice should work well, and Mr. Broskie's provided plenty of documentation on circuit values to use. 'Nuff said.

Difficulty
You'll need to be able to follow a schematic, design and build your own power supply (or use the boards he sells in conjunction with your own iron), design and build your own chassis, and figure out input wiring and whatnot. You'll also need to figure out what style of board and which tubes you want and adjust circuit values accordingly (a chart is provided for common types, and the equations are also available in the online documentation). As I built it, this is not a beginner's project. It wasn't as tough as building a scratch, point-to-point, preamp, but not too far off if you want to be very careful and 'do it up right'. If you got his "all in one" board, and only wanted a single input, you could easily knock up a quickie version in half a day. It is as easy or difficult as you want to make it- many have built these without using the circuit board at all, which would have increased build time somewhat further. Whichever version you build, you can feel confident that you're building a robust, thoroughly thought-out circuit. 

My Version Is The Aikido Octal Stereo PCB
http://glass-ware.stores.yahoo.net/paraikocster.html

www.tubecad.com

http://glass-ware.stores.yahoo.net/index.html