Author Topic: Suggestion for an EEVBlog video. Lab PSU _current_ limit overshoot.  (Read 1047 times)

0 Members and 1 Guest are viewing this topic.

Offline plurn

  • Regular Contributor
  • *
  • Posts: 94
  • Country: au
Suggestion for an EEVBlog video. A "trap for young players" video about lab power supply current limit overshoot - as opposed to voltage limit overshoot that everyone typically complains about.

I don't remember ever noticing current overshoot mentioned in a review so I suspect it is not a well known thing, at least by many people that do reviews. Or perhaps it is so well known among experts that they don't bother mentioning it? I am an electronics beginner and I only found out about it by doing some tests on my first lab power supply and was very surprised by what I found. Example result, set current limit to 20mA, short the output and while measuring peak current, enable the channel - result 16.8A peak. I think it is important that other electronics beginners are made aware of this so they can work around it.

I posted my test results at this thread but will repeat them here. Note that while these tests are done with a Siglent SPD1305X power supply - from what I have read I would expect most or all power supplies to do something similar. I am not taking a dig at the Siglent power supply - in fact I like it a lot.

excerpt from my other post:
- - -


I thought I would try measuring current overshoot. Tests done with a Keysight U1272A multimeter with "Peak detection of up to 250 μs".

Testing done with U1272A set to measure current in the 10A range (which seems to be a 30A range as it is a 30,000 count meter) rather than autorange. Then Peak detect enabled.

Test A SPD1305X settings:
SPD1305X connected direct (pretty much a short) to current measurement on multimeter, then power on the channel with the soft power button.

Result: Usually about 16.8A peak. Yes that is Amps. Sometimes 0.023A which I think means the peak measurement was missed due to current spike being very short, and the limitations on very short duration peak measurement with multimeters. I guess this is a capacitor discharging?

Test B SPD1305X settings:
SPD1305X powered on circuit open, then short the outputs and unshort/open, then connect to current measurement on multimeter. Basically trying to see if connecting after turn on reduces the initial current.

Result: overload - so more than 30A. Probably due to a spark?

Test C SPD1305X settings:
SPD1305X connected to current measurement on multimeter with a green LED inline (in series), then power on the channel with the soft power button. Testing to see how this relates to a more real world scenario.

Result: LED lights up yellow for fraction of a second then changes to green constantly lit. It is a standard green LED which should not be yellow so I assume this is from too much current.
peak current measurement about 0.160A.

Test D SPD1305X settings:
SPD1305X connected to current measurement on multimeter with a green LED inline (in series), then power on the channel with the soft power button.

Result: LED lights up for a fraction of a second then burnt out / open LED. Repeatable result with a second (initially good) LED.
Peak current measurement about 3.145A

Test D SPD1305X settings:
SPD1305X connected direct (pretty much a short) to current measurement on multimeter, then power on the channel with the soft power button.

Result: about 20A peak

If I am testing wrong you are welcome to educate me.

So I am a bit of a electronics beginner and I don't know if it is typical for currently limiting to not work on initial turn on or if this is only a problem with certain power supplies. Maybe a trap for young players? But anyway I don't think this SPD1305X can be relied on to limit initial current with soft power on for delicate loads, or with connecting delicate loads after soft power on. Perhaps delicate loads should have a current limiting resistor installed as an extra precaution. This might be common knowledge - I don't know. At least I did not blow up the expensive 10A fuse in my multimeter.

I was really hoping/assuming that the current limiting would work 100% of the time but I don't know if that is realistic. Perhaps they have a capacitor connected after the current limiting so there is no way to limit that? Is that a typical arrangement? Probably is to ensure low ripple. Does the SPD1305X have too much capacitance after the current limiting? Probably a balancing act.

For now I will assume this is all normal for power supplies unless someone can advise if it is not. Overall I am still happy with this power supply.

Update: current overshoot seems to be normal and is a trap for young players

Update2: It seems that power supplies can be designed to minimise voltage overshoot (voltage-priority mode) or to minimise current overshoot (current-priority mode).

Perhaps with Siglent optimising this power supply to minimise voltage overshoot, that might have compromised current overshoot. It could be argued that this is our (EEVBlog members) fault for complaining about voltage overshoot on power supplies all the time. Siglent is just giving us what we asked for.

- - -
end of excerpt from other post.

So I propose a video something like this - order of the points could be adjusted:

Title: Perhaps "How NOT To Blow Up Your DUT!" (DUT = device under test), but then you would probably want to cover more than just current limit overshoot, such as the more well know voltage limit overshoot. Might be best to restrict this to discussing current limit overshoot. Maybe "Lab Power Supply Current Limit Overshoot!".

- freak us out by setting the power supply with a current limit of 20mA, and destroy a directly connected LED that can handle 20mA. Do the same with another power supply. "Why is it so?".

- measure what is happening to cause this issue. Perhaps with an oscilloscope rather than a multimeter - or show with both so that people without a current probe and oscilloscope can see how to do their own tests.

- testing a few different power supplies to show the current overshoot, or lack of current overshoot.

- explain why this happens, and the compromises power supply designers need to make to choose whether to prioritise minimising voltage or current overshoot. Some info in the URLs below.

- discuss power supplies designed with voltage-priority mode and with current-priority mode (note this is not CV/CC, it is how the power supply is designed to prioritise minimising overshoot of one or the other). Apparently most are designed with voltage-priority mode as that is what engineers expect. Which power supplies have this priority mode as a selectable option? Some info in the URLs below.

- provide some details on how to work around this as a user to protect sensitive loads, such as using current limiting resistors, using a power supply that uses "current-priority mode" rather than "voltage-priority mode" if suitable, using low overall power/current power supplies suited to the load, to minimise overshoots, rather than the biggest power supply you can get (I always wondered why Keysight brings out puny 20 watt power supplies - this might be one reason why). I always thought the bigger the better but this might not be the case.

- mention how this will not matter at all for most loads that will only draw as much current as they need based on their resistance. It will only be a problem for certain loads that can draw more than they can handle - as per the example LED. Mention real world examples - not sure I know any - maybe a device with faulty components that you are troubleshooting?

- If some general designs of power supply are better at minimising various overshoots than others, discuss this. I wonder if a switchmode power supply is better at this than a linear one?

- Optional. Are SMUs(source measure units) free of all voltage and current overshoot? They cost a small fortune so maybe some of that cost goes to limiting all overshoot?

Some info from experts:

Update: I have made a mistake testing the LEDs with too high voltage. I did mention I was an electronics beginner. So that could be the reason why they burned out rather than too much current. Or it could have been both. I don't think this invalidates the concept of the video and the current measurements I took. I think better measurements are needed with a current probe (I don't have one) and oscilloscope to try and determine if the energy contained in the current surge (amps and duration) is significant.
« Last Edit: May 12, 2019, 09:00:49 am by plurn »

Offline rs20

  • Super Contributor
  • ***
  • Posts: 2188
  • Country: au
Too long, didn't read. Where are the oscilloscope traces? But yes, PSUs often have huge outputs capacitors. Connecting a load while the supply is already on will discharge the cap thru your part. Worthwhile warning to have a video on.

Offline plurn

  • Regular Contributor
  • *
  • Posts: 94
  • Country: au
... Where are the oscilloscope traces? ...

So I think I figured out how to get some oscilloscope traces of this without a dedicated current probe. Some details of the set up so people can correct me if I have done it wrong or made bad calculations (quite likely).

Power supply is a single channel linear power supply capable of 30V 5A.

I used the current shunt of an old multimeter that measures 0.011 Ohms at the 10A current terminals (as well as I can measure it - nulling the cables first, multiple measurements, etc). I plugged the power supply into the multimeter current terminals and connected a 1X oscilloscope probe to the multimeter current terminals. It would also be going through a fuse which might matter but choosing to ignore that for now.

With 3V 0.100A set on psu, into 0.011 Ohms I got a peak of about 250mV across the current shunt (not shown). I = V/R

0.25V / 0.011 Ohms = 22.7A

This is in the ballpark of the results I was getting measuring current with the Keysight U1272A.

I have set the oscilloscope to display amps and using the very handy feature of the Siglent SDS1104X-E to be able to set an arbitrary probe multiplication factor, I set it to 90.8X so that the the amps would be displayed approximately correctly. (22.7/0.25 = 90.8)

Some example oscilloscope traces:

set 3.000V and 0.020A on psu. Max 22.43A. Over 10A for 2 milliseconds:

set 5.000V and 0.020A on psu. Max 27.13A. Over 10A for 1.8 milliseconds:

While it is very high Amps, the duration seems pretty short. Don't know if this sort of surge is enough energy to damage sensitive components that would otherwise be ok at the configured power supply settings?

Trying to calculate joules of energy. joules = watts X seconds

Taking just the 10A at 2 ms for the moment for simplicity, 10A / 90.8 = 0.11V. 0.11V x 10A = 1.1 watts. 1.1 watts x 0.002 seconds = 0.0022 joules. That is excluding the rest of the energy from the peak up to 22.52A so that would add a bit more. I don't really know much about joules but 0.0022 joules doesn't sound like much.

For comparison - for a device that would actually draw 0.020A at 3V:
3.000V at 0.020A for 2ms is 0.06 watts x 0.002 seconds = 0.00012 joules.

So the surge of energy (for the 0.011 Ohm resistor - rather than for this device) is over 18 times that. Maybe it is significant?
« Last Edit: May 24, 2019, 08:08:35 am by plurn »

Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo