JISBOS Buffer - Future Gainstage?
Jul 30, 2009 at 11:50 PM Thread Starter Post #1 of 13

Zaubertuba

100+ Head-Fier
Joined
Feb 3, 2009
Posts
449
Likes
23
Following FallenAngel's suggestion, I'm looking at building a balanced headphone buffer using this JISBOS kit for my next "intermediate" build. Should work well with my low impedance, high-sensitivity phones and my "relatively" hot source.

I've noticed the original SOHA is no longer available as a kit, at least from Glass Jar Audio. If, after getting higher impedance headphones down the road (say I went with a pair of HD600's), and decided I needed to add a gain stage, are there other circuit designs that would synergize with the JISBOS well? I'd be interested in either SS or tube designs.
 
Jul 31, 2009 at 4:57 AM Post #2 of 13
You can add "some" gain for the Jisbos by adding feedback resistors.

I doubt you actually need that much gain. R3=1k, R4=1k (cmoy type feedback) would give you a gain of 3, and with them balanced it's going to hit a gain of 6.

Would this not be more elegant and easier to achieve?
 
Jul 31, 2009 at 6:19 AM Post #3 of 13
Quote:

Originally Posted by TzeYang /img/forum/go_quote.gif
You can add "some" gain for the Jisbos by adding feedback resistors....Would this not be more elegant and easier to achieve?


Thank you, TzeYang. I really like simple and elegant.
beerchug.gif


Quote:

I doubt you actually need that much gain. R3=1k, R4=1k (cmoy type feedback) would give you a gain of 3, and with them balanced it's going to hit a gain of 6.


So...R4 (from the Cmoy schematic) would go from the output to...Q2 on the JISBOS? With R3 going to that point from ground? I guess I really need to take a basic electrical engineering course.
rolleyes.gif
 
Jul 31, 2009 at 12:29 PM Post #4 of 13
From the output to the centre point of the DC Trimpot. Observe the feedback path. It's the point of the 33pF to ground. That point is your inverting input.

For current feedback type amplifiers, the output of the symmetrical voltage follower(first stage) is the inverting input.

And I forgot to mention that since the Jisbos is designed as a buffer, it does not have high enough open loop gain. So it's best if you don't exceed a gain of 2.

My calculations were also wrong on the original post :| This is what you get for posting without enough sleep.

1k/1k = gain of 2. And for balanced it's a gain of 4.
 
Jul 31, 2009 at 1:05 PM Post #5 of 13
Interesting. Just to be crystal clear I've marked on where I think you said the inverting input is...

jisbosschematicv101.png


So you just have to put in a 1k/1k potential divider between that point and the offset trimpot and you will get a gain of 2?
 
Jul 31, 2009 at 8:34 PM Post #7 of 13
You would need to cut the PCB trace that runs to the trimpot wiper in order to have the added resistors to have any effect...
 
Aug 1, 2009 at 7:40 AM Post #8 of 13
Quote:

Originally Posted by TzeYang /img/forum/go_quote.gif
From the output to the centre point of the DC Trimpot. Observe the feedback path. It's the point of the 33pF to ground. That point is your inverting input....1k/1k = gain of 2. And for balanced it's a gain of 4.


Thanks for the little primer. I'm getting a lesson in electronics right here on head-fi. How great is that?

Quote:

Originally Posted by Mr.Duck /img/forum/go_quote.gif
Interesting. Just to be crystal clear I've marked on where I think you said the inverting input is...
...So you just have to put in a 1k/1k potential divider between that point and the offset trimpot and you will get a gain of 2?



...and diagrams to boot. Thanks for the clarification!

Quote:

Originally Posted by TzeYang /img/forum/go_quote.gif
Yesh.


You guys read "Mutts" in Malaysia?
wink.gif


Quote:

And you should remove the C3


Actually makes sense...duly noted.

Quote:

Originally Posted by amb /img/forum/go_quote.gif
You would need to cut the PCB trace that runs to the trimpot wiper in order to have the added resistors to have any effect...


And some key info. from the designer himself. Wow!
beerchug.gif


Thanks so much guys. I'm sufficiently stoked to start ordering parts and get going on this project.
 
Aug 2, 2009 at 12:53 AM Post #9 of 13
When you modify the board for gain, it looks startlingly similar to the CKIII circuit. Without any real knowledge of how the circuits work it looks like there are just a couple more transistors in the CKIII to add gain.

For adding the voltage divider in the feedback loop, it seems like the easiest way would be to replace C3 with a 1K resistor then float the middle pin of the offset pot and wire it to the other resistor. You wouldn't have to cut a trace and there'd only be one component not soldered into the board.

I've been thinking about building a similar balanced JISBOS amp/buffer for a while and the only thing preventing me was that I needed just a little gain. This little mod seems perfect.
 
Aug 2, 2009 at 2:32 PM Post #10 of 13
Quote:

Originally Posted by mugdecoffee /img/forum/go_quote.gif
When you modify the board for gain, it looks startlingly similar to the CKIII circuit. Without any real knowledge of how the circuits work it looks like there are just a couple more transistors in the CKIII to add gain.

For adding the voltage divider in the feedback loop, it seems like the easiest way would be to replace C3 with a 1K resistor then float the middle pin of the offset pot and wire it to the other resistor. You wouldn't have to cut a trace and there'd only be one component not soldered into the board.



It is starting to look a lot like a CKKIII. I think the CKKIII goes without any global negative feedback whereas the JISBOS looks like it has feedback from the output back into the trimpot in the input stage.

If I understand you correctly, I don't think that idea would work because the trimpot lead would easily touch the pad on the PCB meaning you would need to insulate it or just cut one of the traces.

BTW why do you need to remove C3? 33pF from output to ground... used for stability? Why not have it?
 
Aug 2, 2009 at 4:05 PM Post #11 of 13
It's more like a Hiraga le Monstre without cascodes on the input stage than a CKIII since the latter doesn't have global feedback and uses a CFP output stage. This is a fine sounding amp with or without gain.
 
Aug 3, 2009 at 1:44 AM Post #12 of 13
Quote:

Originally Posted by Mr.Duck /img/forum/go_quote.gif
It is starting to look a lot like a CKKIII. I think the CKKIII goes without any global negative feedback whereas the JISBOS looks like it has feedback from the output back into the trimpot in the input stage.

If I understand you correctly, I don't think that idea would work because the trimpot lead would easily touch the pad on the PCB meaning you would need to insulate it or just cut one of the traces.

BTW why do you need to remove C3? 33pF from output to ground... used for stability? Why not have it?




because if you don't remove C3 the gain bandwidth becomes entirely dependent on the capacitor. With a capacitance this small, you get a very low high cut off frequency. The more common way of compensating the circuit is to drop a capacitor in the VAS stage and slow the bandwidth as required.

But really you don't have to do it anyway since the Jisbos is already proven to be stable at gain of 1-2. IIRC bandwidth and gain is not related in a CFB circuit.
 
Dec 23, 2009 at 5:50 AM Post #13 of 13
Sorry to resurrect this, I've been busy this fall and my other efforts have been in blocking out the input/output design for my preamp/headamp.
rolleyes.gif


Quote:

Originally Posted by Mr.Duck /img/forum/go_quote.gif
If I understand you correctly, I don't think that idea would work because the trimpot lead would easily touch the pad on the PCB meaning you would need to insulate it or just cut one of the traces.


Soo...how difficult would be to insulate the the pad, or just float the offset pot enough so it wouldn't be an issue?
wink.gif


Quote:

Originally Posted by TzeYang /img/forum/go_quote.gif
because if you don't remove C3 the gain bandwidth becomes entirely dependent on the capacitor. With a capacitance this small, you get a very low high cut off frequency. The more common way of compensating the circuit is to drop a capacitor in the VAS stage and slow the bandwidth as required.

But really you don't have to do it anyway since the Jisbos is already proven to be stable at gain of 1-2. IIRC bandwidth and gain is not related in a CFB circuit.



So if would it be better in general to just lose C3 or keep it anyway?
 

Users who are viewing this thread

Back
Top