Replacement For Toshiba 2SK170/2SJ74

As I point out in another thread, you can use the Gr grades with much the same quality, and with little or no adjustment. Also you can use the Linear systems parts, including their a grade.

Much of the cost of some of these parts is due to audio guys wanting perfect matching, which is usually not necessary. In addition there are some cheaper surface mount parts.

Finally you can use parts like J113 and J176 which do not have as high a transconductance or low distortion or noise, but which will can still sound good and are cheap. Members here will probably mention their own alternatives.

page 7-35 and 7-37 under
shows approaches to increase gm. Also the attached files. Lowering C-in (C g-s) provide better sonic results - so I think.


  • Sziklai-Darlington-II-Theory and applications of field-effect transistors.pdf
    184.3 KB · Views: 139
  • Sziklai-Darlington-II-Theory and applications of field-effect transistors-58.pdf
    295.4 KB · Views: 109
  • sziklai National Semiconductor FET Databook 1977.pdf
    43.9 KB · Views: 116
  • Ultra High Zin Buffer - National Semiconductor FET Databook 1977 (1).pdf
    43.4 KB · Views: 117
Last edited:


diyAudio Moderator Emeritus
2002-09-20 7:20 am


  • IMG_2546.jpg
    952.9 KB · Views: 771


diyAudio Moderator Emeritus
2002-09-20 7:20 am
Hi I had many original 2SK170BL and 2SJ74GR and found a “new” source. I am inclined to call these 2SK170GR either relabeled or fake.

Positive is that when I compared with original 2SK170 from my stock I found a sealed bag with new 2SK30 :) And a bag of TDA1514 (the amplifier). Both keep turning up.
Last edited:


2005-04-30 7:30 pm
Some additional data points: I recently ordered 10 pairs of J74/K170 just to see how fake fakes can be, apart from the lettering.
I received 10 pcs J74, complained to seller, and promptly received the full refund, so nothing to complain about.
Nevertheless, I tested them using EUVL's jig, Vd=9Vdc and Vg=50mVac:



The moral:
  • Idss is within specs
  • Yfs is only half of the nominal 22mS
  • The dispersion of the parameters is minimal, I don't know how they do it; quite remarkable
So, they probably are useful for something, maybe some line-level stuff

Btw, don't worry about the white paint, I tippex them so I can write on them with a sharpie.