6V6 HiFi project from the archives February 08 2013, 0 Comments

I just had a look at this amazingly awesome project over at Circuit and Signal, and felt guilty that the original 6V6 HiFi project was no longer on my site (insert hard drive failure sob story here).  fortunately the Internet Archive WABAC machine has my back.

This amp was in my main system for a bunch of years, then a second amp, then loaned out, then back in the basement, and finally given to my buddy Kevin.  It even went to VSAC in Seattle.  It has been in steady use for 10 years, with no parts failures.  I have rolled a few different tubes through it, but not because of failure.

As one researched tube amp designs, it is very easy to go down the road of "ultimateness" where you start trying to eliminate capacitors, buy really great transformers, experiment with novel topologies, agonize over subjective parts reviews, and finally end up spending a ton of cash for a really great amp.  This is truly a noble pursuit, but I would like to suggest that there is a particular beauty and enjoyment contained in a very simple, classic, and textbook example of the 1940s state of the art.

So here it is, circa 2002:

Since I started all this audio DIY insanity, I have been trying to learn to design tube amplifiers. After much reading, websurfing, and a few discussions with old-school gurus, I have learned enough to get myself in trouble, as they say.

This amp is my first ground-up design. I had the power transformers from another project that never got built, and wanted to use tubes that are cheap and easy to work with. I settled on the 6v6 because I have a bunch of them, and it should be able to provide a couple watts even in triode mode. I also had a bunch of 5y3s so that will be a fine rectifier. Since this is starting to sound like a Fender guitar amp, the logical preamp tube would be the 12ax7, but I thought it would be cool to have an all-octal amp so I used some 6sl7s. Octals are much easier to work with too, since there is so much more room on the socket.
I found a schematic for a design that used the tubes I wanted, and even in the single ended configuration I wanted, so I felt confident that the design would work and that I had a fall-back in case my design sucked. The schematic is from diyaudio.co.kr, which I believe is a Korean site - how cool is that? They have a single ended 6v6 amp with lots of different driver tubes, which is neat. The design of this amp is very straightforward. I wanted to build the simplest amp possible to minimize the things that could go wrong :) Therefore it had to be single ended, had to be triode mode, and had to have a simple single triode driver stage. I didn't use any negative feedback, and didn't try any tricks like parallel feed output or an SRPP driver stage. Perhaps the sound will suffer as a result, but I want to start at the beginning!

I breadboarded the power supply with clip leads and a dummy load to draw roughly the current I thought the final circuit would draw, and so was able to see approximately what my B+ voltage would be. I got about 300V on my test rig, which would be fine. I was hoping for 250-275V, but that looks impossible from a 275V power transformer without adding a dropping resistor which I felt woud compromise the design. Elegance at all costs! :)

Now that I knew my B+ I was able to plot the operating point and load line for the 6v6. I decided triode mode was the way to go since I really would have preferred a real triode anyway. At 300V and 40mA bias the 6v6 would sit at it's maximum dissipation of 12W. The load line looked pretty good, but a little less voltage and more current would have put it in a slightly more linear region of the curves. A quick search for a power transformer that could provide less voltage and still have 5V and 6.3V windings showed that there is no such thing (that I could find) so 300V will have to do.
For the 6sl7 driver stage, I used the tables from a GE tube manual to find a circuit that would let me use my 100k input potentiometer, and provide enough voltage swing for the 6v6. I'm not sure what is involved in matching a driver stage to an output tube, so I basically just crossed my fingers for this part. The values I came up with are very different from any other designs I have seen, but they seem to work just fine. On the input I used VoltSecond's "shunt mode pot easiest" input attenuation scheme.

So with that I started building it. I had a pair of Hammond 270BX power transformers, so I had to go for a dual-mono design. The rest of the iron is Hammond too. 156L chokes in the power supplies, and 125ESE output transformers. I found a really neat old chassis made by Digital Equipment Corporation! Ironic, don't you think? The chassis was nice and big so I would have some room to work. I used a small hole saw for the socket holes, and a combination of drill bits and step drill for the rest of the holes. Octal sockets are salvage, binding posts are Radio Shack, input jacks are the teflon insulated ones from MCM, resistors are vishay 1/2W and 3W from Mouser, coupling capacitors are Angela brand, input pots are leftovers from my Bottlehead Preamp, and electrolytic capacitors are Sprague Atoms also from Mouser. The parts selection is inexpensive, but the quality seems to be plenty good.

Once built, I powered everything up and discovered that I had overestimated the current draw of the tubes, so my B+ voltage was 335V instead of the 300V I was planning on. Damn, I wanted 275V but had settled for 300V which had turned into 335V! I guess that is how it goes :) So now the 6v6s are idling at 14W, 2W over thier maximum rating, and the operating point is well out of the sweet spot. But what can I do about it? Fender runs the hell out of 6v6s in thier guitar amps - the Blackface Champ runs at 350V on the plate at 40mA which is 14W dissipation, so what the hell! Maybe the lifespan of the 6v6s will be shortened,but I'm using good NOS American 6v6s so they won't implode or anything. Let's have a listen!

Running my Radio Shack 40-1354 fullrangers with Motorola piezos, and the lowly discman as a source, the amp sounds absolutely beautiful! Yes, I may be a little biased, having built it and all, but I never expected it to be this good. It sounds like a real amplifier! I figured there would be *something* wrong at least! The biggest suprize was that the bass is quite strong and it plays suprizingly loud. There is definitely some compression as the volume increases, but given the design, that is to be expected.

Thanks to everyone who helped me with this project, most without even knowing it. Marty, all the folks on the Bottlehead Forum, The fullrange driver folks, and all the crazy bastards with clipleads coming out of everything. That's what it's all about :)

Update [2002-02-06]:
I had accidentally left the non-center-tapped 6.3V filament connections floating instead of grounding one side of the winding. I measured the AC hum into 8 ohms before and after grounding them. Floating was 1.9mV, grounded is now 1.1mV. Both are acceptable into my allegedly 90dB efficient speakers, but grounding the filament connection did decrease the hum significantly.

[2012 postscript]:
I mention Marty in the article above, and this is a reference that needs a bot more backstory.  Marty is my vacuum tube sensei.  He is one of those true geniuses that expresses himself best with a whiteboard and profanity, delivered in a classic Brooklyn accent.  I went to him when I first started looking into building tube gear, and he spent roughly an hour explaining everything I would need to know to build a tube amp.  This was all extemporaneous, off the top of his head, and included a ton of advanced information that I simply was unable to absorb.  This whiteboard session started in red, and once the whiteboard was full, Marty didn't pause to erase, but just drew over everything in black.  My favorite Marty quote is (and you have to say this out loud in your head using your best Brooklyn accent, and a tone of resentful amazement) "gravity is a motherfucker" - meant in a completely mathematical sense.