I think it might be interesting to do a version of this with the OnSemi 2SK715 JFETs, which have recently been discontinued but are still available in quantity from the usual suppliers. They feature around the same transconductance as the BF862 and are through-hole TO-92, which is more user-friendly. Also, at <2mA (as the input stage runs), these parts feature triode-like curves for a SIT-ish characteristic.
I'll give it a try. They are a little more expensive though. A 40 piece roll of 862's cost me $8. These are $10 for 20. But might have an interesting SIT sound.
Thanks for suggestion.
Hi,
using a different JFET isn´t such a bad idea.
The BF862 certainly has a high gm, but that is not required for the input stage SRPP.
To achieve the required gain the source resistors must be chosen large and hence the Id becomes very small.
Fortunately the low Id also corresponds with a comparatively large Vgs figure.
Still though Vgs may be small, at only a couple of dozens or a few hundred mV.
You have to take care that Your input signal peak voltage doesn´t exceed the Vgs as this cuts off the Id completely.
Hence a voltage divider or volume pot is required.
A high Idss/high Vgs JFET on the other hand may allow for several V of input signal before current cuts off.
Where a BF862 allows for example only about 350mVp (@Id~3.5mA) input for undistorted output, a SST4392 allows 2V(@Id~15mA).
For the Buffer-part the Vgs restriction doesn´t apply, but to the (here modulated) current-source it applies again.
Also keep in mind, that a high Vgs part requires accordingly higher rail voltages (Vds), especially for the current source part.
As rule of thumb: the JFETs Vds should be >= 2xVgs (and watch the max. allowed Vdg, with Vdg=Vds-Vgs).
As usual You need to optimize the design to Your needs and choose the devices accordingly.
jauu
Calvin
using a different JFET isn´t such a bad idea.
The BF862 certainly has a high gm, but that is not required for the input stage SRPP.
To achieve the required gain the source resistors must be chosen large and hence the Id becomes very small.
Fortunately the low Id also corresponds with a comparatively large Vgs figure.
Still though Vgs may be small, at only a couple of dozens or a few hundred mV.
You have to take care that Your input signal peak voltage doesn´t exceed the Vgs as this cuts off the Id completely.
Hence a voltage divider or volume pot is required.
A high Idss/high Vgs JFET on the other hand may allow for several V of input signal before current cuts off.
Where a BF862 allows for example only about 350mVp (@Id~3.5mA) input for undistorted output, a SST4392 allows 2V(@Id~15mA).
For the Buffer-part the Vgs restriction doesn´t apply, but to the (here modulated) current-source it applies again.
Also keep in mind, that a high Vgs part requires accordingly higher rail voltages (Vds), especially for the current source part.
As rule of thumb: the JFETs Vds should be >= 2xVgs (and watch the max. allowed Vdg, with Vdg=Vds-Vgs).
As usual You need to optimize the design to Your needs and choose the devices accordingly.
jauu
Calvin
You need more radiating area, not more mass (mass only changes the time until final temp is reached).
Large copper areas at the gate pins, and make sure it is *not* reflective (that is, not smooth and not shiny). Make the copper surface as rough as you can, easily done by etching the whole thing again for a short period after removal of the etching resist. This breaks the reflection and increases surface area.
Thick copper and/or a layer of solder near the pins (but not covering the whole copper pours) of course also helps because of its higher thermal conductivity to spread the heat with mimimal temperature drop along its way into the radiating surface.
Current design is in a small sealed aluminum box and pretty much everything heats up but seems ok as burn in over several days seems ok. I also touch parts when in and they are warm but not burning hot. 12mA current per output device x average 9v drop (18v supply) is 108mW ea - well below max rating. I suppose I could pot it with thermal conducting comping bonded to bottom? So far it's running well heat wise.
In-depth info here:
https://www.fairchildsemi.com/application-notes/AN/AN-1025.pdf
Bottom line is that above 0.3in² (200mm², 14x14mm) pad area the effect of large pads offers little improvement but one should strive to reach more than 0.05in² (32mm²)
https://www.fairchildsemi.com/application-notes/AN/AN-1025.pdf
Bottom line is that above 0.3in² (200mm², 14x14mm) pad area the effect of large pads offers little improvement but one should strive to reach more than 0.05in² (32mm²)
Shrinking it down maybe about 23mm x 16mm now for a mono amp.
Sorry got the size wrong. It's 37mm x 26mm for one channel.
Listening impressions update with balanced armature IEM's. I have very sensitive Aurvana 3 in ear monitor headphones (112dB/mW and 60ohm) which don't need any amplification but are great for checking audible noise/hiss and sound balance.
Well good news: I cannot detect any hiss except a very slight almost inaudible level hiss at the highest (last 90%) volume setting which I would obviously never use as it blow my eardrums and the IEM's given their sensitivity. The sound is like I have never heard from these before - rich, full, great detail and articulation, superb bass. Wow. Class A sound with 2-way balanced armature drivers sounds fantastic. I think one way to describe the sound is superb no distortion and great bass authority with clean and distinct bass lines. You don't know what you are missing until you hear it.
I am also testing out my idea of portable use with a 5v USB battery pack capable of 2.1 amps. Partially good news here: it works fine for the IEM's as they don't take much power but the DT880's lack bass on deep bass notes. It's like the battery pack ran out of juice. Surprising since 2 amps x 5v should be 10w and amp draws about 2.5w assuming 50mA x 22v per channel and 90% efficient on the step up. But it could be that peak currents exceed what can be provided even with 10mF reservoir cap.
But I would never use open-backed DT880's on a commuter bus anyway as they don't block the high levels of ambient noise.
Here it is in use on my bus ride:
Well good news: I cannot detect any hiss except a very slight almost inaudible level hiss at the highest (last 90%) volume setting which I would obviously never use as it blow my eardrums and the IEM's given their sensitivity. The sound is like I have never heard from these before - rich, full, great detail and articulation, superb bass. Wow. Class A sound with 2-way balanced armature drivers sounds fantastic. I think one way to describe the sound is superb no distortion and great bass authority with clean and distinct bass lines. You don't know what you are missing until you hear it.
I am also testing out my idea of portable use with a 5v USB battery pack capable of 2.1 amps. Partially good news here: it works fine for the IEM's as they don't take much power but the DT880's lack bass on deep bass notes. It's like the battery pack ran out of juice. Surprising since 2 amps x 5v should be 10w and amp draws about 2.5w assuming 50mA x 22v per channel and 90% efficient on the step up. But it could be that peak currents exceed what can be provided even with 10mF reservoir cap.
But I would never use open-backed DT880's on a commuter bus anyway as they don't block the high levels of ambient noise.
Here it is in use on my bus ride:
Attachments
Last edited:
LTSpice Model As-Built
I just got a new tablet and installed LTSpice XVII on it and learning LTSpice now. I normally use TINA. Anyhjow, I modified EUVL's LTSpice model provided previously with my as-built values: 18v supply, 33R source resistors, 2200uF coupling cap, 10uF gate cap, 260R load. I am getting about -85dB H2 and decreasing higher orders. The default config had higher H3 than H2 (which I do not like as much as higher H2 than H3).
I just got a new tablet and installed LTSpice XVII on it and learning LTSpice now. I normally use TINA. Anyhjow, I modified EUVL's LTSpice model provided previously with my as-built values: 18v supply, 33R source resistors, 2200uF coupling cap, 10uF gate cap, 260R load. I am getting about -85dB H2 and decreasing higher orders. The default config had higher H3 than H2 (which I do not like as much as higher H2 than H3).
Attachments
What do you guys think of this BF862 and pair of LU1014D for outputs? What cand I do to improve it?
http://www.diyaudio.com/forums/pass-labs/146310-bf862-preamp-49.html
http://www.diyaudio.com/forums/pass-labs/146310-bf862-preamp-49.html