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.
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..
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
Last edited:
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
Absolutely right. I always solder 4-wire capacitors on breadboards and in prototypes.
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
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
I purchased 1000 feet of mike cable from parts express, and pull the braid off as I need it.
I also have a few kilofeet of flattened braid half inch wide, looks about #12awg equivalent. It opens to about 1 inch diameter when I birdcage it. I used it to make a line cord a while back, and use it to carry ground currents over low signal level cables. (There is no internal magnetic field inside a circularly shaped cylindrical conductor, be it the outer coax of a shielded run, a piece of copper tubing, or a length of conduit. If the cross sectional geometry strays from a perfect circle, magnetic field will occur internally.)
For my resistors, I now use three pieces of single side clad perfboard..
I did not understand "multitap carpet", but from context is sounds like what I call a resistive ladder.
j
Huh, what do you mean by that? Any pictures?
To maximize information interchange polysyllabic utterances denigrate efficiency. Simple words work best.
ES
What you do does indeed lower the inductance. But there will still be some. The minimum inductance of two cylindrical conductors will be about 150 nH per foot, about 10nH per inch (Note that this is external inductance only, there is still a 15 nH per foot internal inductance for a solid cylindrical conductor...braid eliminates that part as well, based on the aspect ratio of the braid to it's thickness.). And that inductance expresses itself as magnetic field external to the device as a dipole field. (inductance is the relationship between current and stored field energy). The problem is, your solution does not eliminate the external field, just lowers it. Using a coaxial return path over the cap body will confine the magnetic field to the volume under the braid, with none outside. With none outside, a star ground has a better chance of working.
Look at my avatar, it is a field intensity map of a braid conductor inside a braid conductor... note there is no internal field to the inner braid, and there is none outside the outer braid. Energy is confined to between the braids.
Use two resistors of value twice that needed. Run them parallel, tied at top and bottom. Run another conductor down in the geometric center of the two. This will be the minimum inductance configuration for two resistors without going to a body braid for the return current.
j
Last edited:
I'm still waiting for yours.
As mentioned I did show this before in this very thread. I am working on an article that will cover all of this. I know you get the publications where it may appear so that shouldn't be a problem for you.
When I do the schematic I will post it here, for those who don't know what a dual winding transformer with each winding driving a diode bridge looks like. Of course it may just be the .001 uF capacitors placed across each of the eight diodes that confuse some folks. Maybe it is that one side has it's - connected to the other sides + and becomes the common that looses folks.
Of course a search of this thread would show they schematic, but then I was unable to find any L.E.D. noise data using the search function.
Since I was paid to measure some of them I can't post my results. That is why I was curious to see if the correct results were in the public domain.
Alternatively, you could learn the polysyllabic words..
j
That could be too much like work. ESL has more than one use.
Work is fun..it has always been fun...
Those embedded in this discussion already understand terminology such as ESR and ESL.
And, it is important to stretch one's understanding. Questions are not wrong.
Think of e/m theory as a crossword puzzle.
j
ps...Is "ESL" considered polysyllabic??
I used a swept filter style spectrum analyzer. It covers DC (Badly) to 1,000,000,000 hertz. It is designed for 75 ohm TV use.
I connected the shield to the common connection and used a .01 uF capacitor from the + rail to the input. A 75 ohm coaxial cable was used for the test lead. The connections were soldered using 63/37 solder.
I used my noise generator previously shown here as the source. It is an E-I core 18 volt transformer feeding a diode bridge that powers a relay through it's own N.C. contacts. It sits in a small metal box and is connected to the same outlet as the power supply under test. The AC power line impedance is 400 mOhms at the point of connection.
- Status
- Not open for further replies.