Supply Topology/Grounding for Pre-amp

[Bonsai]

I am working on a preamp and have been thinking about the supply topology after reading some of the threads on the forum.

Here is what I've basically come across
1. Straight series pass linear voltage regulator - LM317 thru to low noise discrete variants (examples from Borberly, Walt Jung et al)
2. Voltage + Shunt variants (Curl - I think if I read comments correctly
3. Constant Current/Shunt distributed toplogy
4. . . ?


For # 3, I envisage a constant current source (one for + and 1 for -) that feeds the 3 or 4 preamp stages through a resistor of between 150 - 200 Ohms at each stage. At each amplifier stage, a localised shunt regulator maintains a constant voltage.

The advantages as I see them are:-

a. All the supply lines are constant current - there are no radiated magnetic fields
b. The return lines to the power supply are also constant current except for the load current returns, which would be run separately back to the central return ('mecca')
c. Only place where magnetic fields (=cross coupling) are generated is right in the vicinity of the shunt regulator. with careful layout, the loop area can be kept very small.

I've done a bit of simulation work - see attached LT Spice diagram to confirm the topology benefits. for the shunt regulators, I am thinking about straight Zeners in parallell with 100uF and 1uF WIMA. This weekend I am going to do some practical tests using the zener/cap combo vs an LM317 and a NE5534 and some headphones.

Any other schemes/feedback?

[Bonsai]

No takers?

Can I assume everyon'es vote goes for the LM317/337 voltage regulator approach?

[Nordic]

Afraid your picture is to small to see on anything but a 13" monitor.

[AndrewT]

Hi,
the three resistors after the regulators completely ruin the regulation into each stage.

The statement

All the supply lines are constant current - there are no radiated magnetic fields

only applies during quiescent conditions.
As soon as a signal is passed the lines carry variable current. Is the pre-amp single ended or push pull? Is it fully complementary and/or balanced?

[Bonsai]

Andrew,

the low value resistors in the ground lines simulate the track resistances.

The resistors connected to the voltage sources at each shunt regulators simulate the load current drawn by the various op-amp stages.

The resistors in series with the shunt regulators isolate the shunt regulators from the current sources.

Without these, the lowest voltage shunt regulator would carry all the non-load current - so we would end up with the same situation as would be the case with a normal voltage regulator - varying current in the supply lines - i.e. noise.

Since all the stages are fed from a current source, the current in the lines to each the shunt regulator (zeners in this model) is constant - the only place it varies is between the load and the local shunt regulator. Because these are in very close prximity, the loop area can be kept small, so radiotion, and this coupling to other stages minimized.

[Bonsai]

Just to re-iterate, the current in the supply lines is constant - the only place you see change is between the shunt regulator and the load. The current in the ground returns line is also constant.

[AndrewT]

Originally posted by Bonsai
Just to re-iterate, the current in the supply lines is constant - the only place you see change is between the shunt regulator and the load. The current in the ground returns line is also constant.

are you sure?

The current supplying a ClassA stage varies with the output current.

[Bonsai]

Andrew, yes. Because the whole supply is driven by the current sources, the ground return current is also constant.

I'll try to zip up the simulation and attach it later.

[Bonsai]

Here is the model if anyone wants to play with it.

[Piercarlo]

Originally posted by Bonsai
Just to re-iterate, the current in the supply lines is constant - the only place you see change is between the shunt regulator and the load. The current in the ground returns line is also constant.

I don't understand where is the concrete advantage of displacing variable current (and somewhat variable magnetic fields) from a point to another of the same circuit.

Further, if a magnetic field shielding is needed is better worth pairing the going/return path of the currents with bifilar connection as ever used from the early time of electronics... and left the current itself behave freely as required by circuits and loads...

Bho!

Piercarlo