Simple method to measure inrush current of power supplies of computers etc.?

[nightfire]

Situation: One or more computers are connected via a power strip and then the plug is put into the outlet. BANG! and the circuit breaker switches everything off.
Reason: Most probably a case of capacitor inrush, where the Y-capacitors of a PSU will go for maximum draw for about 20ms.
(Location: Europe/Germany, 230V single-phase 50Hz)

Where it comes from, I know- just had a discussion with some people where this happens more frequently at work and thy would like to measure the inrush currents etc.

I did this at my last employer (datacenter) with a Fluke 435 network analyzer, where I had a prepared cable to wrap around my clamps so that the analyzer could trigger on inrush flanks - so I am familiar with the steps to do so. But as this equipment is expensive and I don't actually have access to such devices, here the question:

Which alternate possibilities do I have to measure this?
We are talking single-phase applications here, and adapters that can expose Live or neutral are no problem to build.

Are there any affordable (sub-1000 Euros) testers available, like Scopemeters? Are there any fast enough power clamps that can trigger on such inrush?
ALternative ideas?

[TimFox]

Most switching power supplies have a high inrush current unrelated to the Y capacitors.
On SMPS supplies rated for use on either 120 or 240 VAC (without switching the range), the inrush is usually higher on 240 V than with 120 V (counter-intuitively).

[EvgenyG]

The design I saw on high power SMPS:
Initially mains current flows (and is limited) through an in rush NTC, and when all big caps are charged, a relay shorts the NTC to bypass it. I believe this is not the case on computer PSU units, I don't believe there's even an NTC there. The big caps with low ESR are most likely the reason why you're getting the circuit breaker tripping. Also, the higher the mains voltage, the higher the is the in rush current (voltage drives the amps).

[EvgenyG]

To answer your question: you need something like this (photo attached) and then put a current clamp and take a single shot on a scope.

[marco58]

If you do not need a certified instrument, take a look at
https://www.sound-au.com/project139a.htm
and some ideas for a possible solution
https://www.sound-au.com/project39.htm

marco

[EvgenyG]

That's a really good website, thanks Marco. Found a nice article about in rush currents there:
https://www.sound-au.com/articles/inrush.htm

[timeandfrequency]

ESP site is top notch information.

Some clamp meters include a 'True INRUSH' current measurement feature (see page 18) which is pretty useful on the field. No oscilloscope or complex setup needed.
Easy peasy operating : just turn on the meter, clamp the jaws on the wire, select the inrunsh function which starts the acquisition mode, switch on your server(s) and the meter displays the inrush current quite immediately.
Max inrush current to be measured is 600 AAC. Should it be more, then choose model 'F406' (up to 1000 AAC) or model 'F606' (up to 2000 AAC).

And the mains adapter shown by EvgenyG is damm useful too. I've got the same with an 'E' socket (for France).
Works like a charm(*) with the mentionned meter (which is also a DC and AC RMS power meter that displays W, VA, var and PF).


(*) Because the adapter features also banana sockets to measure the mains voltage, which is of course needed to calculate power and phase angle for reactive power display.

[bdunham7]

Which alternate possibilities do I have to measure this?
We are talking single-phase applications here, and adapters that can expose Live or neutral are no problem to build.

Are there any affordable (sub-1000 Euros) testers available, like Scopemeters? Are there any fast enough power clamps that can trigger on such inrush?
ALternative ideas?

These two things together will make it simple to measure inrush for individual devices or all of them in aggregate when they are plugged in the power strip.

https://eleshop.eu/cem-ac-010-line-splitter.html

https://www.fluke.com/en/product/electrical-testing/clamp-meters/fluke-374

Alternatively if you have a DSO you can just use a current clamp adapter:

https://www.fluke.com/en-us/product/accessories/current-clamps/fluke-i400s?srsltid=AfmBOooLL4ymstaVs6b_HYV_i31ujEWum8cfwbTV0TdntP1Id46vVu1G

Or some other current clamp.  Note that there are 3 i400-series Fluke clamps and the i400s (shown) and i400e (need banana to BNC adapter) will work with a scope but the i400 (no suffix) will NOT work properly because it is a current transformer.  So shop carefully.  I have the i400e (cheap version) and it works fine with a scope for mains current, BW >> 1kHz.

However, the next question is what do you do with the information after you've measured the inrush currents?  If the sum of your inrush currents isn't 8X the breaker rating or more, then perhaps you have an imbalanced RCD (if you have a combination device) or perhaps you need a different breaker.  If it is high enough to trip a normal breaker, then you need to do something else.  Simply staging the power-up so that you aren't plugging them all in at the same instant would make sense--how often to you need to replug them? 

[asis]

Before diving deep into the measurement process and searching for expensive devices, simply open your PDU (pilot).
If, in addition to the switch with a light bulb, you see there any capacitors and inductors representing (according to the manufacturer) a network filter, then this is the problem that causes the RCD to operate.
-
Remove all this junk except the switch, or replace the PDU.

https://www.pcreee.org/sites/default/files/event/files/presentations/D3%20-%2002%20RCD%20AND%20DISTRIBUTION%20BOARD%20WIRING_Asita%20Langi.pdf

[Coordonnée_chromatique]

Initially mains current flows (and is limited) through an in rush NTC and when all big caps are charged, a relay shorts the NTC to bypass it.

Why a NTC instead of a resistor please ?

[Gyro]

An NTC thermistor allows the supply current to ramp up to near normal value as it heats and reduces in resistance. This means that there is very little additional current step as the relay (if present) contacts close. With a resistor, the value would be more critical, and maybe not possible to achieve an optimal compromise between minimizing initial inrush current surge and the further step change when the relay contacts (essential in this case) close.

The relay (again, if present) is simply there to allow the NTC to cool ready to again limit the surge at the next power-up, if this occurs after a short time. If the relay should fail, an NTC will survive and allow normal operation without damage. If a simple resistor is used, it will dissipate a large amount of heat continuously if the relay fails, causing overheating of the PSU and possible fire hazard.

[pdenisowski]

I did an entire (short) video on this. 



Best way is to use a scope and a current probe.  A few clamping-style multimeters can make accurate measurements of inrush current, but most cannot.

[nightfire]

First, thank you all for tips and tricks regarding measuring devices!

The situation is as follows: Big bunch of notebooks get handed out to students for doing some exams, are later collected and wiped clean. Then they will get hooked up to an appropriate charger to get the battery re-charged.
So now there are numerous manufacturer-provided mains (230V/EU) power supplies on a power bar, that in turn  gets plugged in into the mains- so easily 12 capacitors will really draw some juice.
By my experience (I did similar measurements with a Fluke 435 power quality analyzer at my last employer) in the first 20ms (full wave) the interesting part happens.

Goal: I want to understand how the notebook chargers behave, so I can  draw up a simple chart and be like: On a power bar at a given time only XY of those chargers/wall warts, before they get plugged in into mains.
In the workshop, where some guys will charge them, one could also think about using some different circuit breakers- in germany the B characteristic is quite popular, so this could be changed to the K version.
(C is also ok, but K gives you still more resilience against short spikes, and D is no good in EU systems)

Question here: Are nowadays power clamps or multimeters really fast enough to capture the first 20ms reliably? For understandable reasons I do not want to put in a standard DSO as bench version to that- risk of accident or other hiccup in that environment is not worth it...
(For other spikes and longer periods of drawing power, like a big server spinning up its hard drives, we would talk about seconds where a multimeter easily could do with the min/max autohold function, I know)

[nightfire]

Which alternate possibilities do I have to measure this?
We are talking single-phase applications here, and adapters that can expose Live or neutral are no problem to build.

Are there any affordable (sub-1000 Euros) testers available, like Scopemeters? Are there any fast enough power clamps that can trigger on such inrush?
ALternative ideas?

These two things together will make it simple to measure inrush for individual devices or all of them in aggregate when they are plugged in the power strip.

https://eleshop.eu/cem-ac-010-line-splitter.html

https://www.fluke.com/en/product/electrical-testing/clamp-meters/fluke-374

However, the next question is what do you do with the information after you've measured the inrush currents?  If the sum of your inrush currents isn't 8X the breaker rating or more, then perhaps you have an imbalanced RCD (if you have a combination device) or perhaps you need a different breaker.  If it is high enough to trip a normal breaker, then you need to do something else.  Simply staging the power-up so that you aren't plugging them all in at the same instant would make sense--how often to you need to replug them?

I did some research but in the documents Fluke provided there are no information how the Fluke 374 would handle inrush in which measurement time, so it would be at this point some guesswork- or worth some inquiry to their service dept...

I did find https://www.fluke.com/en/learn/blog/clamps/how-and-why-to-measure-inrush-current  - nice reading, but very broad, and more suitable for motors and similar devices, IMHO.

In the scenario here (charging lots of notebooks in parallel) the measurements could incorporate the following:
- change the circuit breaker to a different characteristic (from B16 to K16 maybe)
- setup a programmable power bar, that does sequencing of the outlets, so that inrush currents are distributed over the course of say, a minute
- invest in (multiple) 16-port USB-C power supply instead of 16 individual chargers that all will have their own power loss etc.
- maybe investigate RCD issues

[bdunham7]

I did some research but in the documents Fluke provided there are no information how the Fluke 374 would handle inrush in which measurement time, so it would be at this point some guesswork- or worth some inquiry to their service dept...

They actually tell you in that link--the 374 and similar current clamps measure the inrush over a 100ms period.  If you want more detailed info, then a current clamp and a DSO is a better option.  However, I can't see why you'd go to all this trouble when the obvious solution is to just plug the power adapters in one at a time.

[nightfire]

Yes, the obvious solution is to do some sequencing. But as there are approx. 1000 Notebooks available that have sometimes (exam times at university) to be charged within a tight timeframe, some deeper understanding of the underlying things is needed, because otherwise some processes do not scale well...

I also have mused about switchable 19" PDU like they are used in data centers, but at some point one has to realize where the overkill begins :-)

[bdunham7]

But as there are approx. 1000 Notebooks available that have sometimes (exam times at university) to be charged within a tight timeframe

Well that certainly changes the picture!  I though you had 16 of them...   If you are looking for a ready-made solution I'd look at something like a Powertronix Isolation Station isolation transformer.  I have two of these for various purposes and they would solve all of your likely problems.  They have NTC inrush devices fitted so no large input surges to trip breakers and they have isolation, so no imbalanced currents (from Y-capacitors) that could trip an RCD.  You'd have to contact Powertronix to obtain suitable models and verify that they have the NTC devices.  Most larger isolation transformers will have these as the toroidal transformers they use also have inrush issues. 

How much power do the individual laptop chargers draw? 

https://store.astrodynetdi.com/media/assets/product/documents/L%20SERIES%20-%20Isolation%20Station.pdf

[asis]

Hi,

Fluke devices (Fluke 199B/C or 190 II family and others) have a "RECORD" function.
Find the current clamps for it, make a series of measurements and process SW FlukeView.
-
But keep in mind that the measurements will be spontaneous, due to the contact bounce of the switching elements and other factors.

[pdenisowski]

Question here: Are nowadays power clamps or multimeters really fast enough to capture the first 20ms reliably?

There are some, but they are several times more expensive than "standard" meters.  Basically you need a very fast measurement / sampling rate - a simple peak hold value on a "standard" meter will often miss the peak.

For understandable reasons I do not want to put in a standard DSO as bench version to that- risk of accident or other hiccup in that environment is not worth it...

If you're using a current probe, there should be no risk at all to the scope - the probe will convert the measured current to a voltage which is well within the safe operating range of any oscilloscope. 

Personally, I find that looking at the inrush waveform (as opposed to just a number) is far more informational / instructive.

[TimFox]

Yes, the obvious solution is to do some sequencing. But as there are approx. 1000 Notebooks available that have sometimes (exam times at university) to be charged within a tight timeframe, some deeper understanding of the underlying things is needed, because otherwise some processes do not scale well...

I also have mused about switchable 19" PDU like they are used in data centers, but at some point one has to realize where the overkill begins :-)

For your purpose, would it be practicable to rewire the feed to the chargers to 120 V (with an appropriate transformer and connectors) to reduce the inrush current per notebook supply?
I have found that inrush current on SMPSs with dual 120/240 V rated inputs is higher at 240 V, although the steady-state input current is half that at 120 V(as expected).

[Atlan]

I used a 1R resistor and a 1013D oscilloscope to measure the current peak when turning on the public lighting. The current spike was tripping the 40A circuit breaker. A current limiter was installed.

[timeandfrequency]

Fluke devices (Fluke 199B/C or 190 II family and others) have a "RECORD" function.
Find the current clamps for it, make a series of measurements and process SW FlukeView.

A well suited current clamp for a Scopemeter is the i30.

[bdunham7]

A well suited current clamp for a Scopemeter is the i30.

The maximum current measurable by that clamp is not high enough for an inrush application.

[UnijunctionTransistor]

Bdunham beat me with a similar reply, which is: make sure the current probe you will be using has a measurement range at least 10X the DUT’s steady state current consumption. Otherwise you run the risk that the probe will clip the inrush waveform.

10X Is an empirical value, sometimes you are lucky and require less headroom, but in other instances it may not be sufficient. YMMV!

[timeandfrequency]

@bdunham7 & @UnijunctionTransistor
For small appliances, of course the i30 is sufficient.
Let's assume that the notebook power supply is 80W which leads to 347 mA @ 230 VAC (as provided in Germany).
The i30 clamp peak measured current is given at 30A so 86.4 X more.

The i30 is actually a LEM PR30 from Switzerland, but rebranded by Fluke (*) and its genuine datasheet plainly mentions the ± 30A instantaneous value measurement span.


(*) I guess that Fluke bought a licence from LEM. The sole change is the case colour.