X2_ugs3,4

[acaudio]

Outstanding work gentlemen. Thank you for posting. This looks like it could turn into a great thread.

ACAUDIO: It looks like your schematic has twice as many MOSFETS as the PCB. Is this what you meant by "Passlabs use only a single set of secondary mosfets"?

GL

I use at present none compensations caps. I think a cascode how in Mr. Borbely designs can be useful for reduction of input capacitance. Maybe we should try this.

In Europe it´s middle of the night...

 

[acaudio]

Square wave 1kHz...




good night, we see us tomorrow...
sweet "X" sleep

 

[metalman]

Based on a hint Nelson gave me while encouraging my development of my BZLS variation, I have a strong belief that the Jfets are indeed cascoded. That would definitely give the extended bandwidth needed to square up theat square wave. The question then becomes: Cascoded with what? Another Jfet seems unlikely, which leaves either mosfet or bipolar?

Hmmm....

 

[acaudio]

I forgot to say: The circuit above is the aktually test setup.

Adam

 

[acaudio]

... That would definitely give the extended bandwidth needed to square up theat square wave. ...

Now has it a little more contour, it was just a wrong (10x) probe from my oscilloscope.
It´s 1 kHz without compensation.

Regards

Adam

 

[vpharris]

I like the self-portrait with scope trace!

Das ist kunst!

Good work, you fellows!

 

[acaudio]

let us go hot

Hi,
I played a little with the values, the frequency response is better (and distortion lower, hopefully) when the mosfets runs hot.

I changed the values from R6-R9 up to 10k, the diffpair bias is ~1,5mA . It gives aprox 15V at the mosfet-gate to supply,
Ug_abs-Ugs=URs -> 15V-3,5V=12,5V (URs) -> 12,5V/470 Ohm= 26mA

(U-URs)*I=P -> (30V-12,5V)*26mA=0,45W -> Ptot=1,3W ->

I tried too to measure CMRR, both input shorted to another and given a signal. At output ghostly still, no signal, it seems to work good. With my oscilloscope nothing to do.

I removed (shorted) the R22, it´s wrong at place. At present I haven´t idea to correct the absolute offset (~500mV).

Here actually schematic:

Regards
Adam

 

[acaudio]

I added a 10k trimmer from R23-S to GND.
The absolute offset is now adjustable to 0mV, but the impulse response is worse than without the adjustment. ( little over swing by a square wave)
With a compensations capacitor (4,7p) no really better.

Mr.Pass, is this the right way for adjustment ? Should I use a higher value for comp. cap?

Regards
Adam

 

[jam]

You might want to try a trimpot between R1 and R2.....then again it might still be tricky to adjust............or replace R6 and R9 with trimpots.

Jam

 

[acaudio]

You might want to try a trimpot between R1 and R2.....then again it might still be tricky to adjust............or replace R6 and R9 with trimpots.

Jam

I think this will not help. It would be next differential offset adjustment.
Adam

 

[EUVL]

I thought about a silimar circuit a while ago, and these would have been my solutions :

1) Trim R8 / R9, which unfortunately affects the gain of the lower half.

2) Use separate current sources for the P and N diff pairs and trim one of them (for absolute offset).

3) Use two separate servos to feed current to R8 and R9.

Am not far enough yet to report actual experimental results.


Patrick

 

[acaudio]

now, after warm up, it seems better than before. hmm....

 

[zinsula]

Cascoded with...?

Based on a hint Nelson gave me while encouraging my development of my BZLS variation, I have a strong belief that the Jfets are indeed cascoded. That would definitely give the extended bandwidth needed to square up theat square wave. The question then becomes: Cascoded with what? Another Jfet seems unlikely, which leaves either mosfet or bipolar?Hmmm....

Perhaps a folded cascode??
Like this?
I don't know why, (well, I'm a beginner) but I like these folded cascode designs. And it seems easier to me to control the open loop gain.

Sorry when I messed up the designations of the components. I'm in a hurry....oh, and I forgot the gate stoppers for the MOS-FETS

Later, Tino

 

[acaudio]

absolutely unbelievably! It works better with a single ended input than balanced. The square is now really clean.
Is this the trick for X2 design?

Adam

 

[zinsula]

This would be a BJT cascode design. What is the advantage of using MOSFET's as cascode transistors?
Doesn't BJT's keep the voltage at the emitter more constant than FET's while changing currents flowing through the devices?

Just thinking......

Ciao, Tino

Oh, Adam, sorry when I am always jumping in between discussion about your already working gainstage.....

 

[acaudio]

Oh, Adam, sorry when I am always jumping in between discussion about your already working gainstage..... [/B]

Hi Tino,
maybe we should separate the thread in two?

Guys, what do you think about?

We can ask the Moderators. Only an idea.

Regards
Adam

 

[jam]

Adam,

It will work since you mainly affect the gain open loop and the application of feedback will somewhat correct for this.

Jam

 

[zinsula]

Hi Tino,
maybe we should separate the thread in two?

Guys, what do you think about?

We can ask the Moderators. Only an idea.

Regards
Adam

Hi Adam, for me it's OK, I have nothing against that

Anyway, Gute Nacht.

 

[jam]

......or you might want to look at this.....on the other site..........for an implemention of a servo........thanks to Uli.

http://www.diyhifi.org/forums/files/pre5v3_648.pdf

Jam

 

[acaudio]

......or you might want to look at this.....on the other site..........for an implemention of a servo........thanks to Uli.

http://www.diyhifi.org/forums/files/pre5v3_648.pdf

Jam

Thanks Jam,

it looks very interesting, but I´m a minimalist. For me too much devices.
I like the simplicity of my circuit.



Tino,

You are right, it´s better one interesting thread as two boringly.

Gute Nacht
Adam