It is good to go back about 2 decades and see what have changed :
www.passdiy.com/pdf/diyopamp.pdf
https://www.diyaudio.com/community/threads/nelson-pass-discrete-op-amp.51836/
https://www.diyaudio.com/community/threads/pass-discrete-opamp-in-dip-8-package.120445/
Actually, not a lot.
Patrick
.
www.passdiy.com/pdf/diyopamp.pdf
https://www.diyaudio.com/community/threads/nelson-pass-discrete-op-amp.51836/
https://www.diyaudio.com/community/threads/pass-discrete-opamp-in-dip-8-package.120445/
Actually, not a lot.
Patrick
.
Thanks for the links. It's been awhile since I read that one. Obviously my brain is broken (no surprise) and I just need to add that 3rd stage (the follower, with CCS) for my application (figure 14b). TO-126 actives on a small veroboard maybe. Still trying to work out how to package that with the FE. Pay no attention to the ring-of-two current source picture I posted earlier. Made dumb mistake there and was unable to edit in time. Drain bamage again. 
If you are dumb like me and you never understood anything NP was ever talked about, then this whole thing is brand new.It is good to go back about 2 decades and see what have changed :
www.passdiy.com/pdf/diyopamp.pdf
https://www.diyaudio.com/community/threads/nelson-pass-discrete-op-amp.51836/
https://www.diyaudio.com/community/threads/pass-discrete-opamp-in-dip-8-package.120445/
Actually, not a lot.
Patrick
.
Maybe at that time people were being bombarded by ads about 85dB speakers, didn't know horn existed, who knows? Sometimes it's all about timing.
All we needed to do was to wait for all semis in the circuit to go extinct for things to get interesting.Not sure much needed changing
Believe it or not, it wouldn’t be difficult to obtain many of the devices in NP’s article. ZVN/ZVP3310 are easily obtained on eBay and have been used in at least one recent project. MPSA18 are in stock at Digikey. Substitute LSK/2SK170 for the single 2SK389 LTP and we can easily try many of those other NP circuits.
All that said, I’m happy to start with NP’s 2023 iteration
All that said, I’m happy to start with NP’s 2023 iteration
Without small heatsinks, the TO-92 devices (Q4, Q5) cannot handle much more heat dissipation.Thoughts on running higher bias and drive a 300ohm headphone?
Aavid makes a heatsink for the TO-92, and DigiKey + Mouser sell it. Check the parts list for the Noir Headphone Amplifier in the diyAudio Store -- it calls for one of those heatsinks per channel.
Some people report good results using a plastic zip-tie, to attach a TO-92 transistor to a US $0.25 coin ("quarter"), by drilling two holes for the zip-tie and using a bunch of Thermalkote heatsink grease. This merde-rig has the advantage of low cost and zero shipping fees, but the disadvantage of not having a measured, tested, guaranteed thermal resistance Theta-JA to aid design and reliability calculations. There's quite a bit of copper inside a quarter (although the exterior surface is zinc), and copper is a great heat conductor. Certainly I'm not going to publicly recommend sanding/filing/diamond honing down the surface roughness of legal tender coins -- but smoother might perhaps be equivalent to better.
Some people report good results using a plastic zip-tie, to attach a TO-92 transistor to a US $0.25 coin ("quarter"), by drilling two holes for the zip-tie and using a bunch of Thermalkote heatsink grease. This merde-rig has the advantage of low cost and zero shipping fees, but the disadvantage of not having a measured, tested, guaranteed thermal resistance Theta-JA to aid design and reliability calculations. There's quite a bit of copper inside a quarter (although the exterior surface is zinc), and copper is a great heat conductor. Certainly I'm not going to publicly recommend sanding/filing/diamond honing down the surface roughness of legal tender coins -- but smoother might perhaps be equivalent to better.
It's a lower distortion part.
Long story made really short... the K170 and J74 are devices that were made specifically for audio. (or at least there were marketed that way...). The J113 is a switch.
Nelson (being our technical, spiritual, and menu advisor) noticed they had a perfectly useable linear band and with a little creativity, good sorting, and proper setting of operating points, they sound pretty darn good.
Long story made really short... the K170 and J74 are devices that were made specifically for audio. (or at least there were marketed that way...). The J113 is a switch.
Nelson (being our technical, spiritual, and menu advisor) noticed they had a perfectly useable linear band and with a little creativity, good sorting, and proper setting of operating points, they sound pretty darn good.
Much higher transconductance than J113 etc,
2x 2SK209GR is as good as a single 2SK170 except for dissipation.
LSK170 is a poor penthode and very sensitive to Vds changes.
J113 is a good current source, but not as good in a diff pair as 2SK209 (SMD version of 2SK117).
https://www.diyaudio.com/community/threads/njfets-for-source-follower-applications.329131/
https://www.diyaudio.com/community/attachments/udness-revisions-191219-pdf.803355/
https://www.diyaudio.com/community/threads/udness-or-you-dont-need-semisouths.343164/post-5954945
https://www.diyaudio.com/community/...lsk170-jfet-pairs-grade-b.321649/post-7035859
https://www.diyaudio.com/community/...lsk170-jfet-pairs-grade-b.321649/post-5706027
Patrick
All we needed to do was to wait for all semis in the circuit to go extinct for things to get interesting.
Still valid today in present tense.
Even MOSFETs are disappearing everyday.
https://www.diyaudio.com/community/threads/f5-headamp.271926/post-7189183
Patrick
BTW I do not know what Nelson is recommending as maximum rail.
The J113 from Onsemi is rated only at 35V.
For the input stage, it is no issue unless you go beyond +/-35V rails.
Not so for the CCS of the output stage, as it sees the full swing.
So +/-15V is within rating, +/-18V marginally over. +/-25V definitely.
If you want this to power a F4 or the like and use 25V rails, I would certainly install speaker protections.
The KSA992 is rated at 120V, so it will not die of over-voltage.
But maybe long-term reliability when dissipating >200mW continuously.
Junction temperature will be 50°C above ambient.
Same for J113 actually.
Patrick
The J113 from Onsemi is rated only at 35V.
For the input stage, it is no issue unless you go beyond +/-35V rails.
Not so for the CCS of the output stage, as it sees the full swing.
So +/-15V is within rating, +/-18V marginally over. +/-25V definitely.
If you want this to power a F4 or the like and use 25V rails, I would certainly install speaker protections.
The KSA992 is rated at 120V, so it will not die of over-voltage.
But maybe long-term reliability when dissipating >200mW continuously.
Junction temperature will be 50°C above ambient.
Same for J113 actually.
Patrick
Thanks Patrick, and thanks for the references. Lots of discussions in the past, that was very helpful.
I noticed the follower stage was also deleted from the newer design. I wonder why removing the flexibility of that lower output impedance.
Maybe it's only intended to drive high input impedance SIT followers or similar.
I noticed the follower stage was also deleted from the newer design. I wonder why removing the flexibility of that lower output impedance.
Maybe it's only intended to drive high input impedance SIT followers or similar.
Yeah, I read the article a couple of times to find the PS voltage and 50V seemed like alot, as a gut feeling given the parts.
So that Nelson does not have to supply 2 extra pairs of matched J113s per kit ?
We had an extensive listening test by different people on various opamp topologies, based on same parts and bias current. etc.
https://www.diyaudio.com/community/...e-opamp-in-dip8-footprint.337484/post-5780747
The double-differential was voted unanimously as the winner, if it means anything.
https://www.diyaudio.com/community/...e-opamp-in-dip8-footprint.337484/post-5780754
Patrick