Super Shunt topology: any hints?

[Onvinyl]

Hi,
this thread emerged from here
I recently tried the jung-didden super regulator with a balanced composite omp amp stage and am very pleased with the results, escpecially in the mid and high regions: infinte resolution, it seems.
The bass is a bit soft though (or not enhanced compared to a simple opamp preamp), so I thought, maybe a super-shunt-regulator is the eat&keep solution for my cake....
Jonathan Carr mentioned super shunt sonics here
In the same thread he provided a link to a super shunt, which sadly is dead now.

Yesterday, I found a good source concerning shunt reg

The attachment shows a conceptual sketch to a super-shunt. Is my guessing correct that we need a 'beta-booster' (darlington or the like) between error amp and shunt transistor in order to have sufficient regulation capabilities?
thanks,
Rüdiger

[Onvinyl]

...
I forgot: the sensing mechanism has to be swapped, the reference device feeds the inverted input.
Rüdiger

[jan.didden]

Originally posted by Onvinyl
[snip]The attachment shows a conceptual sketch to a super-shunt. Is my guessing correct that we need a 'beta-booster' (darlington or the like) between error amp and shunt transistor in order to have sufficient regulation capabilities?
thanks,
Rüdiger

That really depends on the output capability of the error amp. What you normally do is calculate (or precision guess) the variation in shunt current, divide that by the worst-case transistor beta and then compare that to the error amp output current. With a carefull design the shunt only needs to pass the load current VARIATION (plus some constant bias DC current), not, as in a series reg, the total load current, so its requirements are more relaxed than for the series pass transistor.

Jan Didden

[peranders]

Originally posted by janneman
With a carefull design the shunt only needs to pass the load current VARIATION (plus some constant bias DC current), not, as in a series reg, the total load current, so its requirements are more relaxed than for the series pass transistor.

Are you really sure about this? Could you develop this statement a bit?

[Onvinyl]

hi Jan,

Originally posted by janneman

With a carefull design the shunt only needs to pass the load current VARIATION (plus some constant bias DC current), not, as in a series reg, the total load current, so its requirements are more relaxed than for the series pass transistor.

Jan Didden

Does that mean, that the shunt does *not* shunt noise, ripple residuals and small variations of the voltage source (psu)? I was thinking -apparently wrong- that the additional Hfe is needed for that (so a fast&high bandwith error amp would be very beneficial in a shunt)?

Would that job left to the CCS that sets the total amount of current (both shunt transistor and load)?
Rüdiger

[jan.didden]

Originally posted by Onvinyl
hi Jan,



Does that mean, that the shunt does *not* shunt noise, ripple residuals and small variations of the voltage source (psu)? I was thinking -apparently wrong- that the additional Hfe is needed for that (so a fast&high bandwith error amp would be very beneficial in a shunt)?

Would that job left to the CCS that sets the total amount of current (both shunt transistor and load)?
Rüdiger

Well, with the series CCS, any load current variation causes a Vout variation and the shunt then draws extra (or less) current to bring the total back so that Vout remains constant. So the shunt really only draws the load current variation.

Noise etc from the input side works similarly: it comes in as noise CURRENT through the CCS and thus adds or subtracts from the load current, and the correction mechanism is than the same as described above.

The input noise current variation is small to begin with, and the shunt can use the very large CCS impedance to work against so that is easily corrected. I think the shunt noise itself is larger; just a hunch.

Edit: The dynamic impedance for the shunt to work against is the CCS in parallel with the load so in both cases the work the shunt has to do is the same.

Jan Didden

[peranders]

Originally posted by janneman
With a carefull design the shunt only needs to pass the load current VARIATION (plus some constant bias DC current), not, as in a series reg, the total load current, so its requirements are more relaxed than for the series pass transistor.

Compare this:
0 - max current for the shunt regulator
0 - max current for the series regulator

What is the difference?

0 - far from max current for the shunt regulator
0 - far from max current for the series regulator

Relax, in what way? I can't see the difference really.

[peranders]

Rüdiger, if you want to add some "super" to your regulator you must drive the main shunt transistor with a class A stage. Like the series regulator.

[jan.didden]

Originally posted by peranders

Compare this:
0 - max current for the shunt regulator
0 - max current for the series regulator

What is the difference?

0 - far from max current for the shunt regulator
0 - far from max current for the series regulator

Relax, in what way? I can't see the difference really.

Suppose you have a load of 1A, that varies with the signal from 0.8 to 1.2 amp. The series reg has to carry min 0.8, max 1.2 amps.
Same load for shunt, the shunt only has to carry from 0 to 400mA. It is a bit extreme admittedly, but the difference is clear.
Unless you have a fully class B load, then there is no difference, indeed.

Jan Didden

[aparatusonitus]

My turn now, Masao Noro style Super-Shunt Regulator...recommended by JC

Best, a