Ideal shunt wire?

[swbluto]

I've seen tutorials on the internet detailing how to make shunt wire and while that's just oh so useful, copper has a pretty sorry temperature coefficient where a 10 degree change will cause your current reading to be off by 3.34%!(Somewhere around there)

So, the next question then becomes, what are some practically ideal but yet inexpensive shunt wires? It seems nichrome has a temperature coefficient around 20 to 30 times lower than copper, so there'd be much less temperature related error, but I haven't seen wire for it in the <16 awg range(Almost all I've seen are above 20awg). 100 ft. of 21 awg ~.35 ohm/ft. nichrome wire was going for about $14 on eBay. Any ideas on what type of wire is used in controllers and BMS's?

 

[rkosiorek]

many are made from a manganese/nickle/copper alloy called Manganin. mostly because you can solder to it.

roughly the same resistance per foot is a nickle/copper alloy called Monel and it is sometimes used for shunts.

Nichrome comes in several flavors. nichrome-A and -C being the most common. you can sorta solder to it. first it has to be cleaned by sanding to remove an oxide coating that forms on it. and you have to use lots of heat and an aggressive acid based flux. if you don't tin it fast enough with the solder the oxide will form again.

you can buy reistance wire at some industrial supply places like MWS wire and Mcmaster Carr. nichrome will be the cheapest.

rick

 

[wrobinson0413]

.

 

[rkosiorek]

<snip>

We had one big problem with using wire shunts, and that was getting a consistent resistance after soldering them in. A little bit of solder can change you shunt resistance quite a bit as people have found out by tweaking their controllers.

i have found that even changing the position of the probe on the solder blob will whange the resistance reading. that is why when i am setting up a Cycle Analyst that uses the controller shunt i take a voltage reading at the CA connector while passing a known current through the shunt. i ca then calculate the shunt resistance using Ohms law. that is the only way i could get repeatable result.

rick

 

[Toorbough ULL-Zeveigh]

i know this isn't what you're asking for, but when u said 'ideal shunt' it brought to mind this article tripped across recently about using an instrumentation amp instead of a passive shunt & might worth a look.

http://www.edn.com/article/CA6615605.html?spacedesc=designideas&industryid=44217

 

[methods]

i have found that even changing the position of the probe on the solder blob will whange the resistance reading. that is why when i am setting up a Cycle Analyst that uses the controller shunt i take a voltage reading at the CA connector while passing a known current through the shunt. i ca then calculate the shunt resistance using Ohms law. that is the only way i could get repeatable result.

rick

I agree. I take it a step further.
I wire the CA in and power up the controller with 100V
I set the CA shunt value to something close (say 1mOhm) and zero out the current.

I then flip the board over and take an alligator clip and hook into the 100V buss.
I run that through a calibrated ammeter and then into a big oil heater on my floor.
I take the output from the oil heater and "poke" that into the ground circuit upstream of the shunt.

100V / Oil heater = apx 10A

I read the current through the ammeter (say it is 10.62A)
I read the current displayed on the CA (say 8.54A)
I calculate the percent difference and multiply the CA calibration value by that amount.

This brings me to within 1%
I replete and usually get to within 1/10 of an amp.
Sometimes I repeat one more time.

For me, this is the easiest way to absolutely ensure that the CA reads the exact same current as the ammeter given the same stimulus.

-methods

P.S. Dont know how I got off on that Tangent... The real reason I responded to this post is because I too am looking for good thick shunt wire. Unfortunately I realized too late that Nichrome is very difficult to solder to. There are a ton of precision commercial shunts on the market but it is like searching for a needle in a hay stack. In my fantasy world I would find a 250uOhm shunt good for 100A cont. More likely I will have to find 4 1mOhm shunts good for around 25A each and run those in parallel... which would be fine by me.

I have found the best way to locate shunts is to take stuff apart that already has a shunt and read the part number. I have found quite a few good resources this way. Digikey and Mouser are lame in the shunt department... nothing powerful enough or cheap enough for what I am trying to do. I have received some parts from China that had what appeared to be manufactured wire shunts but no luck determining the source.

-methods again

 

[swade]

Would it be possible to parts of an existing shunt from a scrap controller I have laying around? I've got a 30 amp controller that I want to beef up to about 40-50amps and am not quite sure how to be that precise...

 

[dnmun]

if you really wanna use shunt wire, you can buy them from mouser and digikey and newark. but most people just solder up the shunt that is there, not add more shunts. i never even worried about the value of the shunt after soldering it up. but i don't use the direct conect CA.

 

[swade]

if you really wanna use shunt wire, you can buy them from mouser and digikey and newark. but most people just solder up the shunt that is there, not add more shunts. i never even worried about the value of the shunt after soldering it up. but i don't use the direct conect CA.

Yeah I've done that before too. I was hoping to be more precise with using a wire off of an existing controller.

 

[rkosiorek]

do you know what the value of that shunt is to begin with? is it in any way similar to the shunts you are adding it to? is the sensitivity of the error amp in both controllers the same?

sure you can do it. but to expect a more precise setting than just adding some solder is rediculous. just too many variables. even if you take an identical shunt wire from an identical controller and add it to a different controller there is no way that you can mount it the same way as the original one is wired. the resistance of the new connection itself could be higher than the value of the shunt, negating any contribution.

the only predictable way to do it is to calculate the mV drop across the shunt that your controller uses to trigger the over current event and then to modify that level either in software for digital controllers or the reference value in analog systems to match your new conditions.

rick

 

[methods]

Yep.
You have got to calibrate the shunt no matter what.
Its not as bad as it sounds... Especially if the CA is already working.
Think of it as a good excuse to buy a new meter

The only reason to use something other than solder is if you are running huge current and you are worried about temperature drift.
I solder up the shunt on all my controllers and I run 100V and 100A - 150A. It is fine. Not perfect, but fine.

The reason I am searching for off the shelf shunts is to keep the readings stable and to shorten the time it takes me to assemble boards.
Tuning 1 shunt to 250uOhms is one thing... Tuning 30 in a row gets old

-methods

 

[Gregte]

For measuring current, why is the method of measuring voltage drop across a shunt resistor used on ebikes rather than the method of passing the high current wire through the center of a toroidal transformer?

 

[rkosiorek]

For measuring current, why is the method of measuring voltage drop across a shunt resistor used on ebikes rather than the method of passing the high current wire through the center of a toroidal transformer?

the toroidal transformer will only work with AC. very little good hanging one on the DC input to the controller. using a split toroid core with a hall element placed in the gap could be used for DC and is used in some very expensive controllers.

the other reason is simple economics. a couple of chunks of manganin or copel wire is a lot cheaper than a toroid coil and a sensor. especially in quantity. lastly physical board layout is smaller and easier with wire shunts.

rick

 

[methods]

I use hall sensors to measure 0-15A DC current at work.
The biggest problem I have noticed is that they are very sensitive to gravity (though the better ones are not affected) and temperature (again, probably due to the low cost).
I ended up dumping them because they would swing 15% when the fans came on in the chassis

Part# MS15A0112
3 turns of wire around the sensor.

-methods

 

[Gregte]

the toroidal transformer will only work with AC. very little good hanging one on the DC input to the controller.
rick

Now you are making me feel foolish!! I knew that! Color me stupid

 

[rkosiorek]

methods,

i have a couple of A1302 sensors that i use inside a split ferrite core as a current sensor. just for laughs i powered one up today to see if it would be affected by the earths magnetic field. and it was not. 0 was 0 no matter the direction or orientation of the sensor.

i guess those sensors you have are much more sensitive. i tried looking up the part number (MS15A0112) but i could not find it. do you have a datasheet for those? the A1302 looks good as a current sensor, but i may want something more sensitive to use for a different application.

rick

 

[methods]

I tried to find a datasheet and found nothing as well.
If you are really itching I could probably go to my old lab and dig up the datasheet.

I have 16 of them on a board that I designed 7 years ago.
They are nickle size black pucks potted with some sort of dark epoxy.
You can see 3 turns of wire when looking down through the epoxy.
They are rated to either 15V DC or +/-15V DC (cant remember)

The reason we chose them was because (at the time) there were not too many active components available that could read large currents. ( that I could find )

Cant say that I would recommend them though :?

Another great drawback to this sensor is that its output is directly proportional to its +5V input source voltage.
Thinking back - I now remember that it was my 5V linear regulator that was swinging with temperature - which in turn - caused the output of the sensors to swing.
So I guess I cant really fault the sensors on that account.... Just make sure that you have a rock solid source that does not dip under load.

I designed several current sensing boards around that time (all of which were failures for our application). There was one that needed to be calibrated in the vertical position (vs the horizontal). IIRC this particular sensor claimed to be immune to gravitational excitation but I think it was still affected to some degree. If you rotated the board you would see the offset float around.

If you really want to try one out I suppose I could de-solder one and send it to you.
I can map the function of the pins by the board layout.

(I) in
(I) out
5V
Signal
GND

I am sure it could handle more than 15A - the output would just saturate.

-methods

 

[rkosiorek]

thanks, methods.

but i think i will stick to the A1320 i know hat they work well in my applications. 15A does not sound like a very high current though.

i did do one for 400A using a pair of ACS754-200 sensors bolted to a buss bar. these are 200 amp bi-directional sensors from allegro semi. they were located in such a way that the current was shared equally by each. the outputs were summed together to get the 400A range.

for votage stability i used a TL431 as a voltage reference for an LM317. the TL431 in its 0.5% form is quite stable for temperature.

but enough stories.

rick