Replacement For Toshiba 2SK170/2SJ74


  • F89BF2C0-99DF-407B-8A11-08BC31CF02AF.JPG
    262.8 KB · Views: 1,237


2010-06-08 6:15 am
Greetings to everyone who passes this thread by. I wanted to share my jfet experience if I may. I started building my first (class A) jfet buffer 12 years ago as per fellow Japanese engineer's schematics, it was made of 4 para for channel- BL grade Toshiba jfets (8 para total). I bought 16 matched jfets off ebay for $31 in 2009, genuine. I never finished the project due to pot delayed in transit. When the pot arrived I already found other things to do in life so I carefully put all the silver wires, gold plated jacks, pure solder, pots and jfets into a small box and put it aside. For 12 years. Few months ago I wiped the dust off the box and finished the project. I was blown away by the sound I could not believe my ears. I listen to music at home everyday since i hooked this buffer up to my Sony pcm-d50 line out. The sound is so natural and clear I just cant stop enjoying the music, this is a pure joy. So I decided to get another 16 matched jfets and few high grade pots to seal them for later (12 more years who knows). The forecast for music devices industry is not looking promising, it is getting harder everyday to find genuine semiconductors. So I want to preserve the sound to rediscover it later again. I looked through this thread recently and found it quit entertaining and informative so thank you all for input. I also checked for ebay auctions and got frustrated about the number of fake clones on the market and the quantity of sold items as per statistics.
The message from Nelson about new stock arrival gave me hope. Now the question is if it is possible to get 16 matched jfets in case of stock arrival? Will Nelson be able to ship abroad? Can you direct me to a shop web link? Thank you all again, this is my first message on this forum please bear with me))



Paid Member
2004-11-10 8:31 pm
"Some day when we woke they will be all gone."

I worked on really simple active crossover, and build few with real k170/j74 and few identical circuits with j113/j176/j177. I had hard time to measure distortion for real toshibas, had to get better ADC. Distortion with j113/j176 was just a little higher, but nothing I would hear or loose sleep over. Sleep well my friend.
  • Like
Reactions: 1 users

Mark Johnson

Paid Member
2011-05-27 3:27 pm
Silicon Valley
The M2x project's goal was to eliminate the 2SK170 / 2SJ74 while retaining the wonderful sound of the First Watt M2 amplifier. By all accounts it was wildly successful, and thanks to the decision to put the K170+J74 circuitry on an interchangeable daughter board, it's easy to compare the Nelson Pass original (WITH K170+J74) versus the legion of replacements (WITHOUT K170+J74). About three quarters of the diyAudio members who built an M2x, report that they like a non-Toshiba-JFET daughter card more than they like the Nelson Pass original. Getting rid of the JFETs improved the sonics, for these members.

Conclusion: it's possible to eliminate the K170+J74 and get as-good or better sonic results. At least in some cases. Perhaps (?) in most cases or even in all cases.

Mr. Morpheus recommends, "Free your mind, Neo." Good advice.
> Getting rid of the JFETs improved the sonics, for these members.

You have chosen an example where the Toshibas in source follower mode were asked to drive a heavy load.
That is not a typical application for the Toshiba's.
Perhaps you care to show us an alternative in e.g. a F5-like input stage where getting rid of the Toshiba's results in better performance ?

Hello All,

Last night before I unwoke I laid my head on the pillow and read or reread Papa’s papers about the sweet spot, J fringe and a couple of different applications of active crossovers.

The concept of sweet spot was most instructive for me. If you fix as many variables as you can; input impedance, input impedance of the next stage or filter you can adjust the other remaining variables to limit any change in gain. Distortion is change in gain.

With the passing of the complimentary pair, K170/J74, Papa has moved away from the push pull buffer at least for the current crop of crossover circuits.

In the Papa papers I read last night there was considerable discussion of matching of PF5102 style N channel JFET’s and bias resistors. There were AP “distortion” Plots of PF5102 style N channel JFET’s with bias resistors. The plots were scaled with only “distortion”. I assume that is THD+N.

There was qualitative discussion of K170/J74 pairs performing better for noise and distortion than the new version buffers without the P Channel JFET half of the complimentary pair being included. In terms of both distortion and noise the K170 is the better of the J113.. In comparison of the J74 P Channel to the J176 the J176 does even worse. The discussion started out with objective AP distortion plots of the J113 JFETS and ended with a “qualitative” the J113’s are good enough, this is all you are going to get with affordable current production parts.

There were no side by side comparison plots of the K170 with the PF5102 style JFETS or the J74 with the J176.

What I am looking for or will do on my bench is side by side noise and distortion FFT’s of the noise and harmonics for the now RIP K170/J74 and the current production J113/J176 in the real world J Fringe and crossover circuits.

Thanks DT