MPP

As far as i can see you get too little gain. One reason may be that the output impedance of
this alternative input stage is not high enough and the other is that the RIAA is much lower impedance then the one simulated by Sampler.

no matte what RIAA I put on it gives pretty much same results.
I personally think that this circuit doesn not work like this, but I am wondering how come sampler was getting 60dB gain with his simulator.
What simulator does he use?
Which result is correct though?

If the voltage drops and everything looks alright I can try to build this but usually I don't like to build something that doesn't even show the minumum functionalities on the simulator though.
Any thought on this?
 
I will make more experiments today if i can make an input stage based on the proposed topology that has enough gain.
Two years ago i made a parallel symmetric version of it that worked very well.
I designed a very high output impedance, low noise CCS but i forgot exactly how that was designed. I will try to find it in my files but i may need some hours.
There is a ton of circuits i made the last 5 years.
 
I will make more experiments today if i can make an input stage based on the proposed topology that has enough gain.
Two years ago i made a parallel symmetric version of it that worked very well.
I designed a very high output impedance, low noise CCS but i forgot exactly how that was designed. I will try to find it in my files but i may need some hours.
There is a ton of circuits i made the last 5 years.

why is the current source here a concern?
I mean the cascade current source with LEDs is very low noise and high impedance as well.
My main concerns would be whether the topology works properly or not, in other words, if it puts out the gain is supposed to and is not inverting and has decent noise and distortion performance.
For some strange reason the proposed topology doesn't work on the LTSpice simulator.
Is there any volonteer that wants to check that out as well?
I can share my project file.
 
why is the current source here a concern?
I mean the cascade current source with LEDs is very low noise and high impedance as well.
My main concerns would be whether the topology works properly or not, in other words, if it puts out the gain is supposed to and is not inverting and has decent noise and distortion performance.
For some strange reason the proposed topology doesn't work on the LTSpice simulator.
Is there any volonteer that wants to check that out as well?
I can share my project file.

Like this? Does not work (for now).
 

Attachments

  • CroMagnon.PNG
    CroMagnon.PNG
    28 KB · Views: 199
Well, Sampler made it work. How, i do not know. Maybe he comes back on this.
In my book the gain is Gm x RIAA / output impedance of the input stage
I also though to add a folded cascode but then we have the problem that this amount of current ( say we run the folded cascode also on 50mA ) drops too much voltage over the
bass resistor of the RIAA. Say the bass resistor is 100kOhm, then we have 0,05 x 100000 = 5000 V !
 
Last edited:
Like this? Does not work (for now).

yep it is the same model I have and sampler has simulated.
I agree with you that it doesn't work and I would be astounded to see that such a simple circuit boosts 80dB with relatively low distortion, it would be the MOST minimal circuit you can ever think.
I would rave to try such a topology if it was working.

However I still don't understand why Sampler got that result.
 
Frans, your circuit does not work because there is only ca. 25mA idle in the input stage and your CCS delivers 50mA. That makes the voltage over the CCS break down.

Good hint.
I got it to work on the basic idea. It is a matter of having the right amount of current droving on the top current source of the cascade.
Here is a screnshot.

:cool:
 

Attachments

  • Sim-2.JPG
    Sim-2.JPG
    97.5 KB · Views: 193
Frans, your circuit does not work because there is only ca. 25mA idle in the input stage and your CCS delivers 50mA. That makes the voltage over the CCS break down.

Yes, and now it works :) I got the 50mA from Stefano's work (V(D7)+V(D8)+V(D9)-Vbe(Q3))/100 = +/- 50mA :)

With the CCS at 25mA the output voltage is (purple) about 0.002V but at 30mA (and more) it goes to 6V (or there about).

My approach to a problem like this is to first simplify it to the max (and that is what I did :)).
 

Attachments

  • CroMagnon.PNG
    CroMagnon.PNG
    103.6 KB · Views: 186
Last edited:
Yes, and now it works :) I got the 50mA from Stefano's work (V(D7)+V(D8)+V(D9)-Vbe(Q3))/100 = +/- 50mA :)

With the CCS at 25mA the output voltage is (purple) about 0.002V but at 30mA (and more) it goes to 6V (or there about).

My approach to a problem like this is to first simplify it to the max (and that is what I did :)).

And this is the output at 30mA (1mV in)
 

Attachments

  • CroMagnon.PNG
    CroMagnon.PNG
    78.3 KB · Views: 185
Well, what is the optimum idle in the input stage ?
We want high gain and the gain of Fets goes up when the idle goes up.
50mA from 4 x BF862 gives a transconductance ( Gm ) of ca. 160mS.
Gm x R is the gain so we want high Gm and high R. The problem we then run in is DC stability and we need a high impedance node at the output of the input stage or the gain will drop again because the output impedance Rout appears in parallel to R in the formula Gm x R. You can than say G ( gain ) = ( Gm x R ) / Rout.
This high impedance comes with the problem that even a small DC shift in the input stage is transferred into a big DC shift over R, in our case the RIAA transimpedance components shunted to ground. The task is to find a low drift, low noise, high impedance CCS.
 
In fact the whole input stage including the Fet-BJT cascode should not drift much.
The problem is when the voltage over the input cascode goes down the BJT will eventually saturate. Even before this happens, the less DC voltgae there is over the input cascode the less voltage there is to be modulated so we end up with thermally drifting dynamic range.
We want the dynamic range as high as possible though so that the output never saturates even with a loud input signal. With transimpedance RIAAs this is a bit relaxed because the loud treble signal on the vinyl ( ticks, pops, dirt, amplified 20 dB @ 20kHz by the RIAA ) are shunted to ground by the treble cap.
( clips ).
 
In fact the whole input stage including the Fet-BJT cascode should not drift much.
The problem is when the voltage over the input cascode goes down the BJT will eventually saturate. Even before this happens, the less DC voltgae there is over the input cascode the less voltage there is to be modulated so we end up with thermally drifting dynamic range.
We want the dynamic range as high as possible though so that the output never saturates even with a loud input signal. With transimpedance RIAAs this is a bit relaxed because the loud treble signal on the vinyl ( ticks, pops, dirt, amplified 20 dB @ 20kHz by the RIAA ) are shunted to ground by the treble cap.
( clips ).

good thoughts.
Jung paper about CCS might come handy for noise measurements.

Anyhow, I am always concerned about thermal drift I find it to affect sonical performance as it changes bias over time which in terms changes slightly the sound over time.
Isn't there a way to swervo control this? I guess that given the simplicity of the circuit servo into this would be very invasive if ever.

Can we start laying on LTSpice a suitable circuit based on the basic idea?


I will try anyway later to mirror FwD's simulation and try to pass that into a "real" circuit with a basic CCS structure (which brings us back to the basic idea of JG).
 
Sampler has already shown a mosfet gyrator that has low drift. If that also provides low noise is yet to look in. I will search for my design from 2 years ago. That solved the problem i think .
A servo CCS ? Why not, maybe we can get a patent on that, maybe not C:
I think Scott Wurcer came up with something like that. I think it was an opto coupler.

yep that is what I was thinking about..