Electronics > Projects, Designs, and Technical Stuff
Homemade Coaxial Shunt
(1/2) > >>
13:
Hello,

I recently made a coaxial shunt to measure short, high current pulses.  The shunt consists of two concentric tubes of copper attached together on one end via a machined aluminum end plate as shown in the pictures.  A BNC connector is bolted to the aluminum end plate with the center connector of the BNC connected across the inner copper tube.  I am still working on a way to connect the outer tube to my bus bars so if anyone has a good idea let me know.  I know that I should use manganin for the shunt but I cannot find a tube for a decent price.  I guess I'll just deal with the horrible temp coef of copper  :-// .  Also is it worth coating the outside copper cylinder with silver to prevent corrosion (I have alot of Silver Nitrate)?  Overall, I plan on using this shunt to verify my homemade Rogoswki coils as I posted about earlier.  If you see something I should fix/ improve let me know!


13
T3sl4co1l:
Neato, but copper is the worst choice not just for tempco but also because its bulk time constant is very long indeed -- that's an even better reason to use proper materials (if not manganin, then nichrome sheet rolled up?) especially when high speed is demanded.

To put that another way: the frequency where the reactance dominates over resistance is relatively low for copper, and relatively high for resistance metals.

Simply making the shunt smaller, also helps greatly with inductance.  Inductance goes as the physical length.

If you can afford the inductance in your current path, then there isn't much bad about a single-turn strip (many current shunts are sold in this format) -- what's nice is you can loop the sense line through two turns, positioned inside the current loop, to cancel out the EMF.  You still have the added (10nH or whatever) in the current path, but the signal at least comes out square.

Tim
PartialDischarge:
Why dont you cut the middle tube and solder there 20 or 30 of this resistors WSHP2818R0100FEA  in parallel? Otherwise I dont think your going to measure much
doktor pyta:
I would add that direct contact of copper and aluminium is asking for trouble due to galvanic corrosion.

Could You please specify amplitude, rise time and duration of pulses You want to measure?
JohnG:

--- Quote from: T3sl4co1l on December 19, 2018, 07:35:25 pm ---Neato, but copper is the worst choice not just for tempco but also because its bulk time constant is very long indeed -- that's an even better reason to use proper materials (if not manganin, then nichrome sheet rolled up?) especially when high speed is demanded.

To put that another way: the frequency where the reactance dominates over resistance is relatively low for copper, and relatively high for resistance metals.

Simply making the shunt smaller, also helps greatly with inductance.  Inductance goes as the physical length.

If you can afford the inductance in your current path, then there isn't much bad about a single-turn strip (many current shunts are sold in this format) -- what's nice is you can loop the sense line through two turns, positioned inside the current loop, to cancel out the EMF.  You still have the added (10nH or whatever) in the current path, but the signal at least comes out square.

Tim

--- End quote ---

This is true. The shunt element should be much thinner (4-5x) than the skin depth of the maximum frequency of interest, otherwise the high frequencies will be exaggerated. You can use manganin foil and roll it into a tube. The seam will have negligible effect, even if you don't connect it. In fact, you are probably better off not overlapping it.


--- Quote from: doktor pyta on December 19, 2018, 11:41:43 pm ---I would add that direct contact of copper and aluminium is asking for trouble due to galvanic corrosion.

Could You please specify amplitude, rise time and duration of pulses You want to measure?

--- End quote ---

This is so true. The design of these types of shunts is not trivial at the frequencies where these types of shunts are justified.

Also, sometimes the measurements depend a great deal on the rest of the system, and how the shunt connects to it physically, especially at higher frequencies.

John
Navigation
Message Index
Next page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod