You can calculate the drop across various different paths in the system, across the length of the busboard is unlikely to be the major contributor to noise. Yes that could be a 100x-1000x reduction but all the other paths are not so easy. This is the problem with people making a big deal about bus bars, they are wild overkill in one area while not addressing the other paths. I'd also suggest your calculations are misleading as they do not include contact resistances or the full path between a module and the outside world (if you want to compare interconnection outside the case) or at least between modules. At some point reducing the resistance of one part of the system further (such as the bus bars) doesnt make any measurable difference as other parts dominate the result.What? The exact opposite is the case.That's exactly it, except the magnitude of change can be 100x-1000x reduction in current where as it is impractical to reduce resistances by that same amount.Sure one could reduce the flowing current by using several (smaller) PSUs instead of one (big) to split it so there is smaller voltage drop which of course makes a lot of sense if you have a total current draw of several ampere and that indeed should be done in that case.
For every doubling of psus the voltage drop halves.
Let's assume a single psu, 2 A total power draw and 20 mOhm resistance from psu to the end of a bus board, that is a voltage drop of 40 mV.
1 psu = 2 A * 20 mOhm = 40 mV voltage drop
2 psus = 1 A * 20 mOhm = 20 mV
4 psus = 0,5 A * 20 mOhm = 10 mV
8 psus = 0,25 A * 20 mOhm = 5 mV
16 psus = 2,5 mV
And so on...
You see where that goes?
For reducing resistance instead of amperage lets assume changing the bus board to for example my bus bar solution which is 50cm of 100mm2 (20x5mm) solid copper, as it is connected at the center that means 25 cm till the end.
That is 0,0425 mOhm (42,5 microohms).
It is wired to the psu which is 10 cm away with 12 mm2 copper wire which equals 0,15 mOhm. So from psu to the end of the bus bar that is 0,1925 mOhm.
Mr Hinton provided great detail about how the system 0V had to be controlled inside the case and.. between all the other parts that could interconnected. It's on those external paths where isolated power can reduce the current (and therefore voltage) by the enormous magnitudes. Even inside the case the current along the 0V common between different modules can be massively reduced by use of isolated supplies, dropping the current rather than reducing the resistance. Imagine there being effectively zero power supply current carried along the signal 0V reference, thats where isolated power heads toward.
The isolated example to compare against your busboard would be:
busboard with 1mOhm 0V plane carrying 2A/(n number of supplies) and kept within that path only
leakage current down interconnection paths 10uA.. micro amperes multiplied by n (number of supplies)
Taking the majority of the distribution currents including the hard to control earth loops, and reducing that aggressively. Only possible with isolated supplies. Then the modules and their power cables then contribute the majority of the 0V noise.
Statistics: Posted by Mungo — Fri Oct 25, 2024 2:30 am