the best overall pair is Toshiba 2SK170BL / 2SJ74BL
.. they are reasonably in price, can be found and are good for audio = lownoise
... probably the most used .. ever
they are rated SK170 =40 Volt and 2SJ74 =25 Volt
so you will better have to cascode with some higher voltage bipolar transistor
Another pair is:
2SK246 =40V and 2SJ103 =50V
The problem is often to find the P-JFET transistor.
See for example this good supplier:
http://www.ampslab.com/components_fets.htm
.. they are reasonably in price, can be found and are good for audio = lownoise
... probably the most used .. ever
they are rated SK170 =40 Volt and 2SJ74 =25 Volt
so you will better have to cascode with some higher voltage bipolar transistor
Another pair is:
2SK246 =40V and 2SJ103 =50V
The problem is often to find the P-JFET transistor.
See for example this good supplier:
http://www.ampslab.com/components_fets.htm
http://diyaudioprojects.com/Solid/Jean-Hiraga-Le-Monstre/
In the above, you see 2k resistors divide 12 Volt into two 6 Volt parts.
At the 2SC1775 Emitter is like 5.3 Volt
and so the voltage across the 2SK170 is only like ~5 Volt.
The current from 2SK170 just passes trough the 2SC1775 and gets to the 1k resistor.
In this circuit 2SC1775 and 2SA872 cascodes the complementary JFET pair.
If you use 2 equal resistors, for example 5k, to divide 30 Volt
than each JFET will have like 14-15 Volt drain-source.
Cascoding with bipoar transistors is very often used for to help JFETs.
Not only because they can not take the high voltage
but also because they work best at a certain lower voltage.
In the Hiraga above, the reason is not high voltage (12Volt)
for cascoding .. it is because the JFETs perform better = at 5-6 Volt.
Lineup
Thanks for pointing out this schematic, I have though a question. Why in the schematic we take the output out of the BJT collector and not from the jfet drain as we would do if we didn't cascade them and use the BJT just for consuming power? If we take the utput out of the BJT collector, then we are not take advantage of the quality of the jfets (my opinion).
It is true that some minimal quality will be lost
as the current will have to pass the BJT, as any component will degrade signal a bit.
But in this case, the designer has figured out, that the gain in quality of operating the JFET at the lower voltage chosen
gives a better total result,
than if operating the JFET at higher voltage would have done.
Even counted for the current now has to pass another BJT.
2SK170 works not at best, when the drain-source voltage get above 10 Volt something
It can work, for sure, upto 40 Volt .. but there are some parameters that will be less good.
I do not know exactly WHAT, but some other may tell us the correct reason.
as the current will have to pass the BJT, as any component will degrade signal a bit.
But in this case, the designer has figured out, that the gain in quality of operating the JFET at the lower voltage chosen
gives a better total result,
than if operating the JFET at higher voltage would have done.
Even counted for the current now has to pass another BJT.
2SK170 works not at best, when the drain-source voltage get above 10 Volt something
It can work, for sure, upto 40 Volt .. but there are some parameters that will be less good.
I do not know exactly WHAT, but some other may tell us the correct reason.
The cascode does a double function.
It limits the voltage on the input transistor.
It also operates as a common base amplifier.
We take the output from the collector of the common base amplifier.
You can also use a Jfet as the common gate amplifier with similar results. See Borbely for examples.
There is a third advantage to using the cascode loading of the input transistor; namely, it keeps the input transistor at near constant voltage and this results in almost no change in capacitance of the input device resulting in elimination of much of the non-linearity that device capacitance causes.
It limits the voltage on the input transistor.
It also operates as a common base amplifier.
We take the output from the collector of the common base amplifier.
You can also use a Jfet as the common gate amplifier with similar results. See Borbely for examples.
There is a third advantage to using the cascode loading of the input transistor; namely, it keeps the input transistor at near constant voltage and this results in almost no change in capacitance of the input device resulting in elimination of much of the non-linearity that device capacitance causes.
It can be noted that Nelson Pass drives his 2SK170
at 9 Volt across each JFET.
In this Post:
http://www.diyaudio.com/forums/showthread.php?postid=1657272#post1657272
... 14 Volt means 7 Volt across JFET etc. etc.
From 11-16 Volt Drain-source THD is very low.
Now just THD vs. Freq. without telling at what load does not tell the whole story.
How about 2nd vs. 3rd and higher harmonics.
I am currently working on one amplifier of mine.
At lowest THD 0.008%, the 3rd order dist is pretty high. As high as 2nd.
When I search for the trimming of amplifier bias for lowest 3rd dist
I found out that this occurs when THD is 0.055%.
At this level of THD, the 3rd is :
.....55 dB lower than 2nd order distortion
at 9 Volt across each JFET.
In this Post:
http://www.diyaudio.com/forums/showthread.php?postid=1657272#post1657272
... 14 Volt means 7 Volt across JFET etc. etc.
From 11-16 Volt Drain-source THD is very low.
Now just THD vs. Freq. without telling at what load does not tell the whole story.
How about 2nd vs. 3rd and higher harmonics.
I am currently working on one amplifier of mine.
At lowest THD 0.008%, the 3rd order dist is pretty high. As high as 2nd.
When I search for the trimming of amplifier bias for lowest 3rd dist
I found out that this occurs when THD is 0.055%.
At this level of THD, the 3rd is :
.....55 dB lower than 2nd order distortion
Guys, you are off track here...
Increasing the Vds in a JFET stage always leads to better distortion performance. If you are looking at the nonlinear Cgd, it's dependency on Vds is inversely proportional to the square of Vgd. Which makes the Cgd variation with Vds negative, that is, for the same input signal amplitude and gain, the Cgd variation will be lower. As a result you get lower distortions.
Noise is theoretically slowly increasing with Vds (a ionization impact mechanism) but, usually, that's negligible even in very low noise applications. The ionization impact mechanism becomes dominant only around the breakdown voltage.
Regarding the cascode, yes, you could extract the signal from the drain-emitter junction of the cascode, but only if you can read the AC current. The voltage gain at that point is very small, but the transconductance gain is significant. OTOH, you can read the AC voltage or current in the BJT collector. Reason why we are cascoding is to avoid the Early effect (in BJTs) or the Cgd modulation in JFETs (much better to cascode than increasing the bias voltage as above). No, a BJT used as a cascode has almost no impact on the gain stage performance; the input impedance seen by the gain device in the cascoding BJT emitter is very small. To further improve the performance, the so-called Hawksford cascode was invented http://www.essex.ac.uk/dces/research/audio_lab/malcolmspubdocs/J10 Enhanced cascode.pdf
Increasing the Vds in a JFET stage always leads to better distortion performance. If you are looking at the nonlinear Cgd, it's dependency on Vds is inversely proportional to the square of Vgd. Which makes the Cgd variation with Vds negative, that is, for the same input signal amplitude and gain, the Cgd variation will be lower. As a result you get lower distortions.
Noise is theoretically slowly increasing with Vds (a ionization impact mechanism) but, usually, that's negligible even in very low noise applications. The ionization impact mechanism becomes dominant only around the breakdown voltage.
Regarding the cascode, yes, you could extract the signal from the drain-emitter junction of the cascode, but only if you can read the AC current. The voltage gain at that point is very small, but the transconductance gain is significant. OTOH, you can read the AC voltage or current in the BJT collector. Reason why we are cascoding is to avoid the Early effect (in BJTs) or the Cgd modulation in JFETs (much better to cascode than increasing the bias voltage as above). No, a BJT used as a cascode has almost no impact on the gain stage performance; the input impedance seen by the gain device in the cascoding BJT emitter is very small. To further improve the performance, the so-called Hawksford cascode was invented http://www.essex.ac.uk/dces/research/audio_lab/malcolmspubdocs/J10 Enhanced cascode.pdf
Hello,
Personnaly I found that the best compromise between low consumption and low distortion for a cascoded 2SK170 is to cascode it around 15V. That's true that for lower voltage you get more distortion. According to Hiraga in Le Monstre, the distortion pattern of a bjt cascoded fet is also better than a fet alone (more gradually decreasing with raising harmonics and more even order). I never checked if this is real.
Regards
Personnaly I found that the best compromise between low consumption and low distortion for a cascoded 2SK170 is to cascode it around 15V. That's true that for lower voltage you get more distortion. According to Hiraga in Le Monstre, the distortion pattern of a bjt cascoded fet is also better than a fet alone (more gradually decreasing with raising harmonics and more even order). I never checked if this is real.
Regards
If I recall, Nelson used something like 15-16 Volt Drain -Source in JFET BOZ.
Which is a single jfet 2SK170.
Post
http://www.diyaudio.com/forums/showthread.php?postid=1226532#post1226532
Schematic
http://www.diyaudio.com/forums/attachment.php?s=&postid=1226532&stamp=1181083159
Which is a single jfet 2SK170.
Post
http://www.diyaudio.com/forums/showthread.php?postid=1226532#post1226532
Schematic
http://www.diyaudio.com/forums/attachment.php?s=&postid=1226532&stamp=1181083159
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.