John Curl's Blowtorch preamplifier part II

[wahab]

Originally Posted by Wavebourn

All outputs of all supplies are connected together in one point that is the ground.

This point exact physical position is defined only at iddle
or eventualy in a schematic...
Once the system is in large current mode the said point will be
a moving target.

[jneutron]

Originally Posted by Wavebourn

I have a scope. And when I short tip of the probe with it's ground crocodile near any wire that charges caps through diodes I see spikes.

You are measuring dB/dt, a consequence of amperes law and faradays law. This effect is why star grounds do not work with low impedance circuitry. Bonzai is very knowledgeable on this aspect, IIRC he drew a 4 wire capacitor on a simple supply schematic..

Originally Posted by Chris Hornbeck

If a center tapped transformer is used to establish the center point of a dual polarity power supply, any imbalances, including AC ones, appear as current into the "ground". In a circuit with a single true star ground and no ground loops this is no problemo, but what real circuit has no ground loops?

That is incorrect (in red).

The real problem is this: how does one approach a designated zero potential reference point through a path in 3 dimensional space which has time varying magnetic fields... without inducing differential voltages between two wires ala faradays law of induction?

Simple.

1. via two paths which do not trap time varying magnetic fields.

or

2. via two paths with identical integrated time varying magnetic fields.

For #1, geometry of 3-d space states that there can only be 3 paths which are magnetically orthogonal, so practical circuit design cannot use this. There are some simple ways around this.

For #2, it is almost impossible to do this, and the only practical methods to do this are extremely counterintuitive.

The most important thing here is: star grounds are high impedance constructs. NOT for low impedance with high current slew rates.

ps...btw, this is standard stuff....and works for uV level instrumentation wires buried in 10 tesla magnets running 13 kiloamps, up to 45 kiloamp primary/15 million amp secondary structures over ten meters tall.


j

[Bonsai]

. . . winding temperature measurement with an RTD perhaps?

[jneutron]

Originally Posted by Bonsai

. . . winding temperature measurement with an RTD perhaps?

Milli-kelvin accuracy temperature measurement at 1.8 Kelvin using ceramic-oxy-nitride.

1 part in 10e6 magnetic field measurements in 5 to 10 tesla superconducting magnets.

Quench detection of the 30 gigajoule ITER solenoid during plasma initiation.

As well as some difficult tasks...

j

[Wavebourn]

Originally Posted by wahab

This point exact physical position is defined only at iddle
or eventualy in a schematic...
Once the system is in large current mode the said point will be
a moving target.

Moving in respect to what?

[jneutron]

Originally Posted by Wavebourn

Moving in respect to what?

Detroit.

j

[Wavebourn]

Originally Posted by jneutron

You are measuring dB/dt, a consequence of amperes law and faradays law. This effect is why star grounds do not work with low impedance circuitry. Bonzai is very knowledgeable on this aspect, IIRC he drew a 4 wire capacitor on a simple supply schematic..

Absolutely right. I always solder 4-wire capacitors on breadboards and in prototypes.

[jneutron]

Originally Posted by Wavebourn

Absolutely right. I always solder 4-wire capacitors on breadboards and in prototypes.

When using a 4 wire capacitor, use a braid to carry the current from one end of the capacitor over it's body. This negates the external magnetic field caused by the physical length of the capacitor. No, it cannot negate the solenoidal field of the internal foil wrap if the manu kept them apart. But the selection of a low ESL will take care of that.

Using the braid option eliminates the mutual inductance.

This technique works on diodes and resistors as well.

j

[Wavebourn]

I once even wanted to buy a loom machine to manufacture high-end resistors, including non-inductive multitap carpets for volume controls.

[vacuphile]

Originally Posted by jneutron

When using a 4 wire capacitor, use a braid to carry the current from one end of the capacitor over it's body. This negates the external magnetic field caused by the physical length of the capacitor. No, it cannot negate the solenoidal field of the internal foil wrap if the manu kept them apart. But the selection of a low ESL will take care of that.

Using the braid option eliminates the mutual inductance.

This technique works on diodes and resistors as well.

j

Braid sounds like great idea. However, wouldn't it go some way already to mount axial stuff vertically with the return lead close to the body? Or, what I have been doing in the PS I am working on, take two resistors of 1/2 the desired value, connect them at the top and mount the two which are now in series vertically on the PCB. This will also keep the junctions balanced and as closely as possible at the same temperature. I just surmised this would be better, but the differences are below what I can measure, so who knows?

vac