Measuring 100A DC current

[tooki]

6$ on amazon - 100 1/2 watt resistors.
https://www.amazon.com/EDGELEC-Resistor-Tolerance-Multiple-Resistance/dp/B07XLBY5RF

I truly don’t get it: why would anyone order electronic components on Amazon? Aside from it being a scummy company, it’s not even a good price! For the same price, you can get name-brand resistors from Digikey for example.

[Martin72]

You're not seriously thinking about using umpteen resistors in parallel here.....

[mzzj]

But that is precisely the problem.
Finding a current clamp that is that accurate.
If you're lucky, you'll find one with 1%, which is already very accurate for a current clamp, but costs much more than a current shunt, which is even more accurate.

Finding <1% accurate current clamps is not difficult, it is just that finding accurate AND cheap wont happen. 

Hioki CT6846A 1000A current clamp is 0.2% accurate but about 3000 usd/eur 

[Martin72]

That's clear...
There is one more question, which is:
How exactly does it have to be and why?

[BillyO]

Yeah, we know practically nothing about the use case.

[pdenisowski]

I truly don’t get it: why would anyone order electronic components on Amazon? Aside from it being a scummy company, it’s not even a good price! For the same price, you can get name-brand resistors from Digikey for example.

Because if you live in a medium- to large-sized city in the United States, you can get many "basic" (and some not-so-basic) electronic components the same day or overnight, with free shipping if you're a Prime member.  If you need something quickly with low to zero shipping cost, Amazon is sometimes a very logical choice. 

But yes, I use Digikey, Mouser, etc. for things that I don't need right away or when I do a "bulk" buy of components, cables, etc.

And I'm still mourning the loss of Radio Shack (I worked there for a short time) ... also not the best prices or selection, but you could find one in almost any town of more than 25,000 people back in the day.

[Psi]

It's very rare that I need a part today/tomorrow. I usually know I will need parts in 2 weeks when the PCB turns up.
But then, I do have a good stock of generic parts already. I guess anyone starting out would have that issue.

The first thing anyone starting out should do though is spend $100 on a resistor SMT/TH book, capacitor  SMT/TH book, box of electrolytics and a collection of generic jellybean ICs. Build up some basic parts inventory.

[egonotto]

didn't look at them till now. yeah would not use those.

6$ on amazon - 100 1/2 watt resistors.
https://www.amazon.com/EDGELEC-Resistor-Tolerance-Multiple-Resistance/dp/B07XLBY5RF

Hello,

1A per resistor gives 1 W per resistor, which has 0.5 W. And 1 V voltage drop is perhaps too much.

Best regards
egonotto

[johansen]

didn't look at them till now. yeah would not use those.

6$ on amazon - 100 1/2 watt resistors.
https://www.amazon.com/EDGELEC-Resistor-Tolerance-Multiple-Resistance/dp/B07XLBY5RF

Hello,

1A per resistor gives 1 W per resistor, which has 0.5 W. And 1 V voltage drop is perhaps too much.

Best regards
egonotto

So spend 50 more cents for 0.1ohm resistors

[nctnico]

didn't look at them till now. yeah would not use those.

6$ on amazon - 100 1/2 watt resistors.
https://www.amazon.com/EDGELEC-Resistor-Tolerance-Multiple-Resistance/dp/B07XLBY5RF

Hello,

1A per resistor gives 1 W per resistor, which has 0.5 W. And 1 V voltage drop is perhaps too much.

Best regards
egonotto

So spend 50 more cents for 0.1ohm resistors

Which then increases the influence of the leads + solder joints even more. Bottom line: you can't construct an accurate, high current shunt from regular resistors in parallel because in the end you can't make the Kelvin connection. Getting the Kelvin connection done right is the pinnacle of constructing a current shunt.

[Njk]

And I'm still mourning the loss of Radio Shack (I worked there for a short time) ... also not the best prices or selection, but you could find one in almost any town of more than 25,000 people back in the day.

BTW there were also a bigger stories like Fry's. I recalled they'd offered affordable Windows OEM DVDs (about $40, IIRC). For distribution only with a computer system but actually it was enough to buy any HW item, like a napkin for monitor. Almost all my Windows licenses were bought in that way

[2N3055]

didn't look at them till now. yeah would not use those.

6$ on amazon - 100 1/2 watt resistors.
https://www.amazon.com/EDGELEC-Resistor-Tolerance-Multiple-Resistance/dp/B07XLBY5RF

Hello,

1A per resistor gives 1 W per resistor, which has 0.5 W. And 1 V voltage drop is perhaps too much.

Best regards
egonotto

So spend 50 more cents for 0.1ohm resistors

Which then increases the influence of the leads + solder joints even more. Bottom line: you can't construct an accurate, high current shunt from regular resistors in parallel because in the end you can't make the Kelvin connection. Getting the Kelvin connection done right is the pinnacle of constructing a current shunt.

Of course you can make a decent shunt from parallel resistors. It has to be carefully designed, geometrically symmetric and you can achieve good enough Kelvin connection, but you have to think in 3D. It can even be quite low inductance.

But before designing it, accuracy targets need to be clearly specified.

[Martin72]

In my opinion, it's only worth the effort if you can't find anything suitable on the market.
Or if you need a shunt on a circuit board, for example we use 5 R50/3W SMD shunts in parallel on a power supply unit for current feedback.
Speaking of shunts, I actually still had 2 lying around, I didn't even remember...
A 60A/60mV and a 20A/60mV shunt, class 0.5, rather “robust” manufacture.
You can see clear traces of machining on the 20A shunt to achieve the resistance value.

[cv007]

See for example the N4646 series from Vacuumschmelze sourceable from Mouser (say the T6044-N4646-X412) for less than 17€/pcs.

Anyone have experience with these, something like-
https://www.digikey.com/en/products/detail/vacuumschmelze/T60404N4646X101/12530732

Its not really clear to me what these are doing, and why they are relatively inexpensive (<$20 for the above part). Obviously not the same as a flux gate current sensor of considerably more cost, but they seem to be using the same idea as they describe it as a closed loop with a zero field detector. Are these essentially similar minus the precision? I can see where precision takes a lot more effort/cost, but if that much precision is not required then maybe these are a good option?

They list accuracy as max 0.5% with typical 0.1% for the above part (at max rated current, normal temperature), but not sure where you end up considering the rest of the variables in play. A typical 100A/100mV shunt is rated at 0.25% error, so it seems these are in the same ballpark. It looks like the downside is you have to come up with a +/- 12-15V power source, the upside is you can eliminate dealing with shunt heat and connections.

Maybe I am missing something obvious, but these look like something that I would at least like to check out since the cost is reasonable. Did not know these existed.