2SK79 VFET gain stage

Okay, so I’ll start with a brief introduction of my current headphone setup:

Source is a Vref modded dam1021 (rev2) DAC that’s hooked up via and XMOS USB interface to a WIN10 minimac running foobar2k with an ASIO plugin. The dam1021 raw output goes into a dcb1 unity gain jfet buffer, which in turn goes into a 10k Valab stepped attenuator, the output of which goes into another dcb1 buffer. So basically an r2r DAC and a buffered volume control. The dam1021 is specified to put out 1.4Vrms (unbuffered, single ended). The dam1021 dcb1 pre combo is around 0.007% THD at 1Vrms output, mostly 2nd order.



Currently my headphone amp is a scaled down clone version of the well-known FW F4, with a single pair of output transistors. The F4 clone itself has no voltage gain. The F4 clone itself has a THD of 0.0019% or -94.9dB for an output of 1V (0.002% at 3V), with the 2nd harmonic dominating the 3rd (by maybe 10dB).





This setup has enough voltage swing to drive the HD800. No surprise, Volts RMS required to reach 90dB SPL as reported by innerfidelity are around 0.25 Vrms. With the HE6, which according to innerfidelity needs 1.018Vrms to reach 90dB SPL, I found a subjective lack of oomph.


Given my interest in VFETs, I have been eyeing 2SK79 based preamps for quite some time. There’s an SRPP design by Yasui Akira of Wakamatsu-tsūshō, Akihabara that apparently was published in the June 2002 issue of MJ magazine. That simple design has caught my particular interest, and I’ve been looking for a kit for a looong time, couple of years actually. Finally, this past March, I found one for sale on yahoo Japan.

It arrived two weeks ago and looks like this:





It’s a total hype beast, sporting Riken carbon film (?) audio grade resistors (an oxymoron) and Rubycon Black Gate filter caps at the Zener regs’ outputs.



The actual circuit differs slightly from the schematic enclosed with the kit, but it was easy enough to figure out that it runs on 40V, as hinted on the packaging. Since I didn’t have a 35-0-35 VAC transformer on hands as needed for the onboard center tapped full-wave-rectifier PSU (50V unregulated, 2*(35*2^.5)/2 for the center tapped full wave rectification), I grabbed a 250VA 2*18V toroidy audio grade transformer off the shelf and whipped up a supply with parts from previous builds that I had stashed away in various boxes. So now there’s a discrete full wave bridge with two 10000uF 50V caps and an R47 wire-wound in a CRC configuration. The regular PSU wasn’t ridiculous enough for my tastes anyway.





The circuit came alive without a hitch, and I’ve been listening to it for several hours now. Listening to the HE6, before adding the gain stage between the source and amp, the volume pot was a few clicks from max (so maybe 6 to 9 dB of headroom). Now it’s just shy of 12 o’clock. Subjectively, adding the VFET gain stage has greatly improved dynamics, brings out lots of details top to bottom, and provides the desired oomph. This is, however, kind of expected; after all I paid for it, waited in anticipation, built it, and even think it looks pretty.



I usually don’t do this, but this time I listened before I took measurements relating to sound. So after listening for a few days, I pulled out the measuring gear (Boonton 1121 analyzer, THD baseline 0.0008% at 3Vrms 1kHz) and pulled some numbers from the circuit yesterday. Here’s what up:

Vin=1Vrms, Vout= 10,55Vrms, 20.47dB gain (a lot)

Vin=250mVrms, Vout=2.826Vrms (this would do 1W into 8R with the F4 clone amp)

here the THD=0.154% or -56.24dB
and S/N=92.60dB


This is of course far beyond headphone level listening. So I put on the HE6 and turned up the analyzer’s output until the volume was uncomfortably loud. This was at Vin=100mVrms, resulting in Vout=1.145Vrms. The 90dB SPL Vrms=1.018Vrms as per innerfidelity.

So for Vin=100mV, Vout=1.145Vrms we get THD=0.034% or -69.25dB



Looking at the spectrum, we see a mix of 2nd and 3rd harmonics, with the 2nd just a tiny bit above the 3rd (markers at 2kHz and 3kHz).

At higher levels the composition remains pretty much the same, with the 3rd going slightly above the 2nd.


I initially built the F4 clone to get as close to the source as possible. Turns out that’s not really the sound that I prefer. This is good news however, because that gives me a reason to build a whole fleet of preamps now.

 

Hey, a fellow dam user. I am also using similar dam board, only the rev1 with some substantial vref mods and custom filters and separate shunt power for everything.
Are those Würth poly caps on the VREF? If so, I have the same ones, only underside.

Have you a-b'd the dac output with and without the dcb1 buffers?
If so, have you noticed loss in fine details, dynamics etc ?
Now that you have the VFET gain stage this should not be issue to try out. If I didn't make a mistake reading the schematic, the VFET gain stage has input load of about 100k ohms, this will cause no trouble for the dam raw output to drive directly.

 

The caps indeed look like the Würth, yeah, but they’re Nichichon FP Aluminium Organic Polymer Capacitors, RR5 820uF 6.3V.

https://www.mouser.at/ProductDetail/Nichicon/RR50J821MDN1?qs=bW1g5Sys4u0yvYlAG5xXUQ==

The rev2 of the dam1021 had the vias filled with resin, so I had to add the caps directly to the bypass caps.



Most of my opinion on the dam1021 stems from my experience with it in my stereo system, where I had it running with a dcb1 based preamp and a unity gain JFET buffered active line level crossover (the mess to the right):



You get the picture, I’m a fan of JFET buffers. Adding one to the dam’s output in lieu of an opamp seemed like the natural choice. No A-B-ing could convince me otherwise. Being the dense mfker I am, I went as far as removing the opamps from the pcb just so that there’s none in the chain (not even idling) and I can burn more current in the shunt…yeah.




I thought about putting the VFET pre right after the dam1021, having it work into the volume control, and having a buffer at the volume’s output. I think I will try a few different gainstages first though before I move further upstream again.

 

Gorgeous builds!
Have you ever compared the mezmerize to the original B1 jfet buffer with really good caps?

 

Thanks

I used to have a B1 clone built after the original schematics, but this was in a very different system and too long ago to remember anything useful about the sonics.

 

forget the preamps! This opens a whole world of VFET / SIT power amps for you to try

great work on your builds..dB

 

Most VFET power amps have a jfet/mosfet/bjt voltage gain stage though. So far I've built two VFET power amps, the DIY Sony Vfet from diyaudio, and this single ended design:



Going from memory, I'm still undecided wether the VFET pre plus source follower combo beats the other amps I've tried. I'll have to revisit some pre/amp combos I guess.

There are a few preamp designs I'm currently working on (three different Jfet, another with a Russian VFET, and possibly a tube pre), so we'll see how well the 2sk79 srpp holds up to other topologies. So far I'm impressed.

 

Yes, i have not seen a vfet gain stage before ( not before your preamp post). I wonder if it has more to do with the SRPP design as my all time favorite tube preamp is an SRPP based design.

I do like the visual affect of the l'amp. If you already have the Sony vfet power, then you already know the general sound of this genre.

Looking forward to hearing more about your exploits into the preamp world.

.. dB

 

>>Rubycon Black Gate

Are these still lusted after in the DIY world? I thought they went out of production years ago

 

I think that's part of why they're lusted after. If you want a good laugh (or if your head needs scratching), go look at some completed ebay auctions...

I thought about flipping them, but the whole build is so silly I just left them in place. I like to think they're part of the magic