Can somebody please explain me why are those transistors so special for audio use? What exact parameters of those fets are so desirable for our circuits?
Is it high Idss? Low input capacitance? What to look for when choosing a fet for an audio (preamp) project?
Is it high Idss? Low input capacitance? What to look for when choosing a fet for an audio (preamp) project?
Hi,
There are some great threads here. There is a search function and there are some great books referred to here as well. I hope this helps you.
There are some great threads here. There is a search function and there are some great books referred to here as well. I hope this helps you.
Article by Erno Borbely one "must read " classic
http://www.audioxpress.com/assets/upload/files/Erno BorbelyJFETs The New Frontier Part 1and 2.pdf
Greetings
http://www.audioxpress.com/assets/upload/files/Erno BorbelyJFETs The New Frontier Part 1and 2.pdf
Greetings
1) They are very quiet. You can use them for the mc input on a phono stage
2) They have high transconductance at normal bias currents (like 10 mA)
3) With an Idss around the 10 mA zone, they conveniently self-bias
4) The N and P types are good complements in symmetric circuits
5) For such a little part, they do a good job into lower impedances with
low distortion - (witness the Beast with a Thousand JFETs)
6) Toshiba's quality consistency is second to none
2) They have high transconductance at normal bias currents (like 10 mA)
3) With an Idss around the 10 mA zone, they conveniently self-bias
4) The N and P types are good complements in symmetric circuits
5) For such a little part, they do a good job into lower impedances with
low distortion - (witness the Beast with a Thousand JFETs)
6) Toshiba's quality consistency is second to none
Thanks for the answer! I suspected low noise couldn't be the full picture as I've seen them being used in buffers like your B1 where noise shouldn't be of a primary concern.
ad 2) high transconductance means we can achieve high gains with smaller collector resistors, right? But let's say I get 2SK2145 that is easily available from mouser at around 0.6eur a pop as compared to 5eur / piece for an lsk389. I can build a stage with an active load that will surpass any gain figures of a sk170 stage and still pay fraction of the price for the parts. There surely must be a tradeoff, but I'm not seeing one at the moment (I am quite a newbie - bare with me)
ad 3) by "conveniently self-bias" you mean we can use a convenient emitter resistor values? And by convenient I assume lower than in case of devices with lower Idss so we get lower noise from the resistor?
ad 4-6) good points, I didn't think about those.
Just to be clear - I'm not trying to question a choice of the jfet here, but merely understand the reasoning behind choosing this device in spite of limited availability. I started digging in this topic because I want to build a discrete opamp based on the opamp article from passLabs as well as Boberly article presenting EB-604 preamp. They both present circuits where mentioned transistors are used without really justifying why so there's where my question is coming from 🙂
ad 2) high transconductance means we can achieve high gains with smaller collector resistors, right? But let's say I get 2SK2145 that is easily available from mouser at around 0.6eur a pop as compared to 5eur / piece for an lsk389. I can build a stage with an active load that will surpass any gain figures of a sk170 stage and still pay fraction of the price for the parts. There surely must be a tradeoff, but I'm not seeing one at the moment (I am quite a newbie - bare with me)
ad 3) by "conveniently self-bias" you mean we can use a convenient emitter resistor values? And by convenient I assume lower than in case of devices with lower Idss so we get lower noise from the resistor?
ad 4-6) good points, I didn't think about those.
Just to be clear - I'm not trying to question a choice of the jfet here, but merely understand the reasoning behind choosing this device in spite of limited availability. I started digging in this topic because I want to build a discrete opamp based on the opamp article from passLabs as well as Boberly article presenting EB-604 preamp. They both present circuits where mentioned transistors are used without really justifying why so there's where my question is coming from 🙂
For posterity: some useful material I've found so far:
Choosing Discrete Transistors [Analog Devices Wiki]
https://www.fairchildsemi.com/application-notes/AN/AN-6609.pdf
Choosing Discrete Transistors [Analog Devices Wiki]
https://www.fairchildsemi.com/application-notes/AN/AN-6609.pdf
whatever you want to do , you can (Moi working on some solutions exactly with 2SK2145BL)
however , when you want complementary JFet buffer , there is no substitution for sk170/SJ74 ....... especially if you already have them
however , when you want complementary JFet buffer , there is no substitution for sk170/SJ74 ....... especially if you already have them
> solutions with 2SK2145BL
2SK2145 BL is essentially 2x 2SK209BL in one package, but with both Sources on a common pin.
Apart from using them in diff pair, you can use the D-S reversibility and wire them as a follower, as in the "B1".
But they are not guaranteed to be matched in any way (Idss or else).
And the heats dissipation is limited.
The 300mW maximum rating from the datasheet assumes you have generous board area, and is absolute maximum.
In practice, for good reliability, you should not really go beyond 100mW per SOT23 package.
Unless of course you glue on additional heatsink directly on the package itself.
Done that all years ago,
Patrick
2SK2145 BL is essentially 2x 2SK209BL in one package, but with both Sources on a common pin.
Apart from using them in diff pair, you can use the D-S reversibility and wire them as a follower, as in the "B1".
But they are not guaranteed to be matched in any way (Idss or else).
And the heats dissipation is limited.
The 300mW maximum rating from the datasheet assumes you have generous board area, and is absolute maximum.
In practice, for good reliability, you should not really go beyond 100mW per SOT23 package.
Unless of course you glue on additional heatsink directly on the package itself.
Done that all years ago,
Patrick
The 2SK2145 is a great part too. We use lots of them when a compliment isn't needed. Get them while you can.
What dissipation do you design for ?
For BL grade at say 8mA Idss and +/-15V, total dissipation is 240mA.
Typical Rthja for SOT23 is ~300K/W, meaning a junctional temperature rise of 72°C above ambient.
The absolute maximum I would dare to use any silicon based active device at.
Patrick
For BL grade at say 8mA Idss and +/-15V, total dissipation is 240mA.
Typical Rthja for SOT23 is ~300K/W, meaning a junctional temperature rise of 72°C above ambient.
The absolute maximum I would dare to use any silicon based active device at.
Patrick
IMO so-so at best. The beta parameters are quite different which is all that matters.
From August 2010 :
http://www.diyaudio.com/forums/atta...48-f5-power-amplifier-input-jfet-variants-pdf
Patrick
http://www.diyaudio.com/forums/atta...48-f5-power-amplifier-input-jfet-variants-pdf
Patrick
That is a great read EUVL, thanks for posting it!
Regarding interesting devices, U430 from InterFet looks mighty good and it's available from mouser!
InterFET Corp. Page Selector - U431 datasheet
Regarding interesting devices, U430 from InterFet looks mighty good and it's available from mouser!
InterFET Corp. Page Selector - U431 datasheet
Last edited:
The first reference I know of in a diy schematic is by Mr. Borbely in the 1/86 Audio Amateur - part 2 of the Borbely Preamp article. And the K170 was still at Mouser late 2011, maybe a bit longer.
25 years.
25 years.
For use in the audio circuits being presented here, what version of these little guys is best, GR, BL, or V.
Or will it matter greatly?
Thanks
Or will it matter greatly?
Thanks
Bob Cordell's upgraded Hafler, which he presented two Burning Amps ago, gets around this by using different amounts of degeneration for N and P. Picture below.
_
