Complex Bus Bar Design

You're going to have to do some analysis of the mechanical loads on your bus bars due to the switching of electrical loads, especially large changes in current in a short time. You will also need to consider thermal effects (how much the bus bars will heat, how the heat will be removed and how quickly, etc.). We can't do that for you.

I can suggest some good insulating material. Check out Glastic brand insulating materials. You can get it as large sheets (as in 4' x 8', in various thicknesses from 1/4" up) or in formed shapes, as well as pre-formed stand-off insulators. Here's the web site:

http://www.glastic.com/en/products/glastic-electrical-products.html

Glastic sheeting and forms are commonly used in electrical apparatus for insulated barriers and bracing. The stand-offs are rated from 600V up, depending on model and size.

But I would hope that you have someone do a detailed analysis of the stresses that need to be handled. You don't want something to fail in operation and go up in smoke, and then your insurance company say it won't be covered because you don't have proof it was adequate for the application. I also don't know how something like UL listing comes into play on this. Maybe one of our other gurus can comment on that.

I have software to calculate the heat load. As most of the heat will be between the POS / NEG bars, I plan to add some forced cooling.

Thanks for the information on insulating material.

If memory serves, Brown Bowery has a very nice pdf on it. What concerns me is the quite large dI/dt = 100A/msec can cause large voltage spikes over the inductance of the assembly. You need calculate it properly.

Your drawing is not clear:

I can see the red bars having a gap, each! How does the current flow thru the gaps?

Some contimuity problem in your drawing.

In any case, it is better to have the Bars vertically positioned so that the narrow edges see the other narrow edges: The (+) Bar on top of the (-) Bar (or reverse). They can also be placed flat horizontally if you have enough depth in the cabinet. The outgoing bars or connections should be properly placed with enough distance to minimise the same attraction effect due to the high current flow expected.

This way reduces the inductance/capacitance and mostly the Attractive Forces effect on the bars. At high currents, if the 6" faces are parallel, the attractive forces will try to bring the 2 bars together, and the bars will have the least rigidity (mechanically) to resist the forces. The insulation fixtures will also be happier.

The drawing shows the main power coming in on the left (Red + , Blue -). The red bars +POS must go to the load in series then to the part being tested. That is why the red bars are broken to take the power to the load. The blue bars are the -NEG or return. My main goal is to lower the inductance and per the information I have, placing the bars 6" faces as close as possible gives me the lowest inductance. Is that information correct? I also need to space the other load bars (bars located above or below) far enough away to avoid proximity effect. What is that distance with 833 amps max per bar ¼" x 6"?

Placing the 6" faces as close as possible will increase the Capacity!

As far as Inductance (coil effect) is concerned, Also that will increase. BUT the inductance effect only appears during the surges or brief peaks, in a DC circuit.

In a bus bar, these effect are minimal. The worst case is in the cable runs and the way they are layed, outside the bus bar cabinet. (DC not AC here). To minimise capacitive effects between the + & - bars, You should put them stacked vertically to use, effectively, the rigidity against flesing that the 6" bar side will give you.

Your major concern is the Structure and design of the bars so that they will resist, ptoperly, the attraction forces that will be created between the (+) and (-) bars. At 5000A, these are going to be very significant!

The Heat dissipation you seem to be able to deal with. Forced ventilation or natural convection (if the cabinet is big enough with openings...).

I'm not sure I understand. Let me explain what I'm looking for. Testing requires very high / very quick inrush currents. The product under test must close and open the circuit. Our old system 12vdc @ 7000 amp inrush (long cable runs, bus bars 6ft apart, etc…) , I had to add 10 farads of capacitance with resistive load near the product under test to meet the inrush speed requirements. As capacitors are added, cost and safety issues pop up. I'm hoping to add all the capacitance I can and remove all the inductance I can with the bus bars. The system will be at full power (any capacitance will be at full charge) when the product under test closes the circuit. The product cycles ON/OFF thousands of cycles during a test. When testing near the limit of the power supply there will be full inrush current, then a small dip as it charges the caps while at the same time powering the test. This is OK as long as the product saw the full inrush current.

As for structural issues I can support the bars as often as needed plus with an isolation barrier between the POS & NEG compression should not be a problem. I know the barrier will affect the heat load but with the added cooling it should not be a problem.

That is fine. Now I understand your objective.

In my opinion, if the outgoing Red(+) busbars(loads 1,2 &3) are longer than the remaining bus bars, by far, then these are the main contributors to the inductance. They introduce a series inductance in the circuit, which magnitude is proportional to the apparent frequency calculated from the inrush peak. The remaining busbars, will be contributing much less, and the way you are fixing them, (narrow spacing), will be fine to try and counter-balance the inductance.

To counter the Inductance created by these branche-outgoing, the capacity must be in parallel, which is fullfilled in your design since the (-) circuit is not branching.

To neutralize the inductance, you need to Inpedance match the circuits for the apparent frequency introduced by the inrush speed. In your case, you will have to rely on previous experience and the geometry of the circuit run etc.

In my own experience with high frequencu currents (MHz), which is a continuous load, you can calculate the length of the transmission cables and look at the design geometry to reach a near enough solution. Then it needs fine tuning by trimming the transmission lines...

You might need a Power transmission engineer with experience in DC high current transmission.

An emphatic NO to the less than halfbaked ideas on transmission lines.

In order a transmission line arrangement to work without crazy ringing, it REQUIRES:

1,. A well defined, constant impedance Z thruout, no deviation,no branches, no multiple loads. Those produce well defined reflexions, exactly calculable. Nonetheless, they mess up the waveform big time.

2,. The generator (source of the energy at the input to the cabinet) need to be well defined, and exactly equal to Z. Neither condition is met. More calculable reflexions. Worse, the resistive component in it inherently eats up exactly 50% of the available energy. I suggest, do not attempt to go there.

3,. The termination (your test loads) need to be exactly equal to Z. Am I kidding? Those are TEST loads, hence by definition undefined. More reflexions. Those can be negative or positive voltage excursions.

Radio Frequency engineers obey them regularly. You simply cannot. But, at least you can strive to avoid what you can, and be aware of what you have to live with. You need to build the whole branching construct with the same construction thruout, same spacing, etc. to its long end. Keep the stubs short. Flexible cable connections need to be at Z too. Disconnect them from the stubs when not needed. A large capacitor at the test obiect limits the frequency of the reflexion from it.

Example: load 10 milliOhms, resistive, switched on.

filter capacitor 0,1 Farad, maintaining voltage for a short transient.

Time constant: tau= 0.01 x 0.1=0.001 sec = 1 millisec

That is the time the voltage dropped 30% on the cap. It also the bandwith of this lowpass arrangement, 1kiloHertz, roughly. If you want to know the drop in higher frequency components, it is 6dB/octave or 20dB/decade. That means, a 10kHz component drops to 3% of what it would be without a filter.

As you can see, the very high frequency components of the reflexions are quite low on the busbar, IF.....That is a big IF: it demands a very high quality cap. Its internal impedance needs to be MUCH LOWER, than the load, if any filtering is to take place.

The Internet has decent material on the matter mentioned, while no substitute to full learning.

--------------------------------------------------------------------

There is no free lunch, anywhere.

You should read the OP 1st and then judge my response.

I might seem halfbaked to you, but You seem more cooked than necessary for this topic. This is a DC supply with a transient symptom at the switching. Very short time.

My mentioning anything to do with transmission line theories is only to draw attention that the issue of matching the line to the load at the transient part of switching such a big load is more complicated than just constructing the bus-bar system. His attempt to use the capacity value between the bus-bar flats to improve the circuit impedance will not make the required impact...

Just cool it. If you have any contribution to help the OP in a constructive way, then fine. Otherwise, just be polite when drawing attention to some more sophisticated KNOWLEDGE that only over cooks your own ideas.

Your attempted explanation is out of context: there is no source to match the load to, here. The attempt is to only reduce the impedence that will appear at the transient phase, putting the brakes to the dc current flow. The OP already mentioned that he has very high Capacitors installed at the end of the circuit. He is loking for a way to reduce them or eliminating them.

ANY Ideas on this Subject? stick to the question if you can.

You are entitled to your own opinion, but NOT your own facts.

---------------------------------------------------------------------------

Sonny. It never has been about you, as assumed, but OP's problem. You, as anybody else, can voice whatever, baked, halfbaked or unbaked, as it may be. As long as the intent is help, all contributions are taken as intended.

Hence, it is not about you or me or anybody else on the sideline.

Other than that, your note flaunt a series of arrant nonsenses. I only pick out the most glittering diamonds.

"This is a DC supply with a transient symptoms at the switching. Very short time". What an arrant nonsense! There is no such thing. A step function transient has ALL frequency components active from DC to daylight. NO EXCEPTIONS!! And everything, ,that is larger than a fraction of a wavelengths, IS A TL.

"My mentioning anything to do with transmission lines....." is that convoluted, I cannot make sense of it. BUT, the rest of the paragraph makes no sense, no matter what.

The use of a capacitor IS explained in my note.

The last paragraphs are flat incoherent.

MY summary is, that transmission line theory rules, whenever the size is a large fraction of the wavelength considered.

For those, short on theory, see:

Wikipedia:

Impedance matching

Transmission lines

Reflections of signals on conducting lines

Absorb and internalize the matter , and see me in due time. Take your time.

See you.

I do not wish to argue with you on a topic that is purely argumentative.

Your comment and further direction is out of topic here.

Have a look through these...

http://cr4.globalspec.com/search/sitesearch?do=show&sort=textmatchrank&srch=busbar&order=asc

Thanks for the information. More fun stuff to look through.

As you have dificult to bend bar, sould you use only the main bars and connct the loads with cables, as many as needed. Problably this simplify the construction. Anyway do the proper dimensioning.