DIY amp for AKG K340

well, here's your criteria... 30V output voltage swing, peak to peak. This comes from actual measurements on oscilloscope; music playing loud (too loud for continuous listening but not "I-want-to-remove-this-ASAP" loud either) showed regular clipping on my current amp that can "only" do 25Vp2p. Of course, this is too loud for continuous listening, but it's always nice to have some reserve for transients. Your K340 being more 320ohms than 400 ( http://www.head-fi.org/forum/thread/446294/actual-akg-k340-impedance-measurements ), that means also 15/320= 46mA peak are needed

Not that easy for a simple circuit with tubes. The bijou from Cavalli Audio should bring you quite up there.
Edited by 00940 - 12/17/10 at 8:30am

The Bijou is limited by driving the grid positive WELL before 30v p-p is achieved. Unless Im mistaken (I have never measured one, but cathodynes are not known for driving power tubes into class A2 ;) ) you can only get about 4vp-p out of a bijou before it clips into an infinite impedance load.

 

I'm not sure I agree with 30vp-p to drive the K340. Perhaps if you want consistent 120db peaks, but not everyone does. I would say, build a simple amp like a Cmoy and measure how much voltage swing you actually use. Add a healthy amount of headroom for good practice and go from there. This will also give you a good idea about how much gain you need for cheap. 

 

If you do need a bunch of voltage swing I would look into simple 6AS7 cathode followers: the Bottlehead Crack kit is a nice example.

I stand corrected on the bijou.

On the K340: The service manual of the K340 specs 160mW for 112db @1khz. That's already roughly 20Vp2p and 30mA peak. 115db and 320mW is 29Vp2p and 45mA peak. So 30Vp2p won't hit 120db. 30Vp2p might be not be "necessary", but 20Vp2p is a bare minimum.

Of course, the same manual specifies a maximum handling power of 200mW so you don't want to keep your headphones at those levels too long.

 

Hmm, interesting.  I've not measured my Bijou but I probably should.  I'll try to haul my scope out this weekend and check some voltages into my 600-ohm headphones.  4Vp-p is not much.  It's not even 2Vrms, it seems awfully low to start clipping.  I don't know much about cathodynes and the futterman design, unfortunately.

00940

I wasnt aware that the manual mentioned such high voltage swings. They are pretty crazy headphones.

 

Regarding the Bijou & all that.

I personally wouldn't worry about it. 4vp-p isnt much compared to more, but it will drive most conventional high impedance headphones to formidable levels. I am curious how close my guess was to reality though. It may be as high as 4.4v, and could possibly be slightly less than 4v, but Id bet somewhere between 4 and 4.4vp-p.

 

The cathodyne is the second stage. Its the tube with outputs at the plate and cathode. It splits the signal into 2 out of phase parts (balanced!) to drive the output stage. Something of a problem in designing an amp with a cathodyne is that the balance of the cathodyne is entirely dependent on the load impedance. If you get the resistors wrong things go weird.

 

Now, running class A1, where the grid is never driven positive the input impedance of a triode is basically a lot. Several meg-ohms typically. In the bijou its bypassed by the 470Kohm resistors, but lets just say "a lot".

 

When we drive the grid positive things get weird. The input impedance of the tube DROPS extremely quickly. This is a big problem because while we are driving one tube positive the other one is being driven "more negative" the load impedances as seen by the top and bottom of the cathodyne are no longer the same and the balance goes away.

 

Unfortunately building an amp that will run class A2 is tricky. The crossover from A1 to A2 must be handled seamlessly, which requires a fairly low output impedance driver that does not react poorly to a changing load. Transformer coupling, cathode followers driving power tubes, and pentodes driving power tubes are good bet. Several well respected designers have commented that a lot of "the sound" of tube power amps is actually the driver stage struggling to drive the output stage properly. The other half says "I dont care what you say, I like the tone of my 12ax7 driving a 845" which is Ok, as long as you admit to building an expensive tone control.

 

Now the futterman design is a whole 'nother story.

 

The futterman design is actually a very neat combination of components. When I first saw it I was appalled by the use of a cathodyne (I am normally biased against cathodynes because they suck) but the way futterman integrated the various parts is toooooooo cool.

 

The key to the futterman design is a positive feedback loop nested inside of a negative feedback loop. The positive feedback loop is tricky. Positive feedback buys you silly-low output impedances, but if left uncontrolled WILL cause very serious oscillations. The neat trick is that the negative feedback loop keeps the positive feedback loop in check AND lowers the output impedance even further!

 

I think that there is some untapped coolness in the Biojou circuit, but have never tested. I dont think R5 and R6 are optomized to take advantage of the positive feedback effects of the Futterman. Alex has commented elsewhere (in the bijou thread) that the output impedance before the application of global feedback is (I dont remember what, I think 60ohms) - which is almost exactly the same as a simple white cathode follower without the benefit of positive feedback. Im pretty sure that with some tweaking of the values of these 2 resistors you can get output impedances in the single digit, if not fractions of an ohm range once global feedback is applied. I would say if you want to try this to lock global feedback down too - remove the feedback pot and replace it with a good old fashioned resistor once you settle on values. 

 

I think its better that the stock schematics make no effort to exploit the benefits of the positive feedback loop: the popular mod of removing of the global feedback loop would turn the amp into an oscillator and result in many melted tubes and smoky amps. 

 

Ooh, if anyone out there in TV land is crazy enough to actually try to get the positive feedback loop to work, please dont come crying to me if it dosnt work. I really hope it does, and think it will, but it might go up in smoke.

Edited by nikongod - 12/17/10 at 8:41pm

At reasonable cost and with tubes, the 6AS7 suggested by nikongod are looking fine. Is this the kind of thing you had in mind ?

It's a rough, quick sketch. The CCS gives the amp a quite good power supply rejection, which allows for a very basic power supply (a isolation transformer, a pair of 470uf/200V caps should be enough). Gain is high but it's good for the K340 out of weak sources. The CCS allows also for a lowish B+, making it a bit safer to build and reducing dissipation in the tubes (longer life). 5ma through the ecc88, 50ma through the 6AS7.

I built something very close to that as a tweak of the SP extreme. 

 

6sn7 with single DN2540 plate CCS, and red cathode LED gain stage DC coupled to both halves of a 6as7 in parallel as cathode follower with a DN2540 CCS load. Output idle current was about 70mA. I think BJT CCS's are a better choice for the gain tube, but they would have been harder to work into the extreme.

 

I also added parafeed transformers off of the output MAD ear style for low impedance headphones. It could probably be duplicated for less than $300 in parts if you were really careful with parts selection.

Wow, thanks.  That's alot of info to process.  It sounds like it's a worthwhile tweak to explore.

 

Just thought I would mention that the K340 and Bottlehead Crack was not a very good combination in my opinion. It sounded very weak and flabby.

First project changes everything.

 

A bit more background on your experience level would go a long way in helping isolate a suitable amp project for you to pursue.

 

Lot's of "first build" threads on the forum. Might want to search through a few of those as well.

Are you looking for something portable or is a desktop amp okay?