Weird measurements on a shunt resistor

[Eheran]

Oscilloscope: Rigol DS1054z, 4 probes RP2200 all compensated to their channel and a fresh self-calbration after running for ~1h.
Simple powerfull 12V LiFePO4 battery connected to a 12V 55W lamp and i want to measure the current, mainly the "inrush" current, with a 50A/50mV shunt resistor. Now take a look at those values - i dont know what to say. They are all over the place and after about 0,1ms settle to the correct ~4,2A the lamp draws. At 10x Ch. 2 even shoots above 0V for some reason.


Picture 1 to 7 (53-61) are all the same, where in Nr. 4 (57) i swaped the probes from Ch. 1 and 2 and in Nr. 5 (58) those of Ch. 3 and 4 [no new compensation was done], Nr. 6 and 7 (60+61) i changed the physical location of the probes.
Picture 8 to 13 (68-73) are all the same (but different scale and 10x), 8-10 (68-70) are with the Scope on 230V AC, 11-13 (71-73) for shits and giggles on 60V DC. Notice how Ch. 4 bows down on DC but up/straight on AC
Picture 14 shows the setup. Connecting the Lamp was done manually, no switch.

[Performa01]

The probe hooks and ground leads form nice magnetic loop antennas, with different loop area and -orientation per probe as can be seen in the last picture.

When you connect the light bulb, it is cold and its resistance is low initially, hence there is some enormous current pulse, generating a magnetic field in the wiring of the lamp. This is of course picked up by your 'loop antennas' and adds/subtracts to/from the shunt voltage.

Use a twisted pair of wires instead of the probes to connect the shunt to your scope and it will be much better. The switch pulse will not be totally eliminated though, as the two connectors of the shunt are too far apart, so you cannot avoid having some loop area at the last few centimeters of the twisted pair wire.

[Kilrah]

At the very least, connect only one ground clip and remove the others!

[Eheran]

At the very least, connect only one ground clip and remove the others!

That causes a lot of noise on the not-connected channels.
I used Math this time, so no ground at all just 2 Probes at the 2 ends. About -1100mV and thus 1100A inrush-current. Can that be correct? I had about 500-600A from the other measurements.
Channel 4 ist a current clamp, going crazy too... it peaks way too late. However Its correct (4,2A) later on too.

How would you connect twistes pair wire?
Or how else would you measure "huge" inrush currents? After all this is just a 55W lamp, nothing speacial...

[mos6502]

An incandescent lamp's cold resistance will typically be 1/10th of the hot resistance, so a current over 50A is completely implausible.

[TurboTom]

It seems the initial inrush current overdrives the o'scope's channels. I found the DS1000Z series to show some funny behavior in the instant after a heavy overdrive. Moreover, my RP2200 probes are complete crap in x1 mode. Rigol should have omitted the useless switch and just manufacture them as fixed x10 probes. In x1 setting, they can also produce all kinds of funny waveforms and intermittent signals - just typical of a switch that doesn't make proper contact. At least some of your measurements appear to have been done at x1 probe setting.

Cheers,
Thomas

[Eheran]

All the x1 and x10 in the screenshots are the correct values.
Maybe you guys can try to measure such inrush currents too?
Everybody (with a car) got a car battery and 12V headlamps. So we could compare scopes and techniques as well as the results.
Im going to try a few more things and report back.

[mos6502]

Those measurements are definitely bogus. I just measured the cold resistance of a 55W H4 bulb - it's around 0.4 ohms. I=U/R. The inrush current will be around 30 amps. That's a fact. If your scope tells you otherwise, it's lying. That's why it's important to always do a plausibility check on anything you measure.

[Eheran]

I already did that, >100A is impossible, indeed.
It seems like i cant measure the first few µs for some reason.
However, both shunt and clamp do a good job after about 400µs of start.
Check this out:
89: Ch.4 Clamp 10mV/A = 34,4A  -  Ch.1 Shunt 1mV/A = 36,4A (after 400µs)
90: Ch.4 Clamp 10mV/A = 13,6A  -  Ch.1 Shunt 1mV/A = 13,6A (after 15ms)
91: Ch.4 Clamp 10mV/A = 5,1A    -  Ch.1 Shunt 1mV/A = 4,8A   (after 85ms)
92: Ch.4 Clamp 10mV/A = 4,3A    -  Ch.1 Shunt 1mV/A = 4,2A   (after 500ms)

The resolution of 400µs is okay for me, there are going to be few machines that require it to be faster.

PS: How to measure the "avg. voltage"? V_mid, V_avg or V_rms?
They all give different values even tho mid and avg should be the same as the voltage never goes inverse... right?

[mos6502]

Well, something is very wrong then. This isn't a fast event. Also, a ~30mV signal can't be overdriving the ADCs at 10mV/div. Bug? Defective Scope?

Maybe you should do a better setup and switch the lamp with a 555 as a monostable and a MOSFET as the switch to eliminate connection/switching/wiring issues. Like this:

[Eheran]

I just placed a 1mH inductor (Ch. 2) 2,5cm from the shunt and measured 9V over a 100k resistor. The little inductor thus got about 1mW of power.
94: The first 50µs
95: same but 550µs

So the huuuge inductive spike seems to be the problem. The clamp goes into negative for the first 50µs and slowly reaches a real value after about said 400µs.

I dont have a 555 laying around. However, as long as i make good contact right away (50% of the time) it looks the same. So i dont think it has anything to do with the "switching". (aka jaming the gold contacts onto each other)


Edit:
I added two measurements showing the points at which the current seems to be correct. 400µs for the current clamp and ~50µs for the shunt. (yellow=shunt, blue=clamp and turquoise=inductor voltage over 10kOhm @ 10cm)
Now all i need to know is how a faster response time could be achieved and how exactly the inductive effect messes everything up