Korad KD3005D bench supply power on spike

[Electro Fan]

Just for the heck of it....

Korad connected to Rigol scope as follows:   BNC cable from scope via a BNC to banana adapter to alligator clips on leads connected with normal banana plugs at Korad output terminals.

Scope set for 50 ohms, 1X, 2V, 0A.  Note scope is set for 20mv/Div.

[nctnico]

Yes, there are spikes when you power on/off though a physical button. Which I guess is still bad if you keep the powered device attached when you power on/off the power supply. Not ideal at all.

These spikes are not on the load but induced in the leads of the oscilloscope. If you put a low value resistor across the oscilloscope input (and switch the probe to 1:1 mode) they are gone. As Electro fan just demonstrated 

[radiolistener]

Just for the heck of it....

Korad connected to Rigol scope as follows:   BNC cable from scope via a BNC to banana adapter to alligator clips on leads connected with normal banana plugs at Korad output terminals.

Scope set for 50 ohms, 1X, 2V, 0A.  Note scope is set for 20mv/Div.

Do not touch enable output button. Just press POWER ON/OFF switch.

Set single trigger mode on oscilloscope and press POWER ON button.
All leds and fan on PSU should be off (no mains) before pressing POWER ON switch.

These spikes are not on the load but induced in the leads of the oscilloscope. If you put a low value resistor across the oscilloscope input (and switch the probe to 1:1 mode) they are gone. As Electro fan just demonstrated 

No. There is no oscilloscope probe, I used RG58 cable with BNC connectors and BNC-bannana adapter on PSU side.
When I touch a ground only to oscilloscope BNC, there is no spike.

Here is 50 Ω pass-through dummy load placed on oscilloscope side. This is a pattern at power off.

As you can see, the pulse still present. Just a little smaller - about 2 V peak-to-peak (about 10 mW).

[Electro Fan]

One more....

Korad connected to Rigol scope as follows:   BNC cable from scope via a BNC to banana adapter to alligator clips on leads connected with normal banana plugs at Korad output terminals.

Scope set for 1M ohms, 1X, 2V, 0A.  Note scope is set for 1V/Div.

[ataradov]

Note the time scale. You won't see them at 100 or 200 ms/div.

And also, any tests with soft on/off button are irrelevant. We know the output is clean in this case.

[radiolistener]

One more....

You're doing wrong measurement. Do not touch ON/OFF button. ON/OFF button should be always in OFF state

Use POWER ON/OFF button instead of ON/OFF button!

[tautech]

Do these have a mains X-Y cap input filter network ?

No (actually I didn't verified it inside, just tested it with DMM for capacitance).

Capacitance Neutral-Ground and Ground-Phase both are below 100 pF

With mains switch ON or OFF ?
It needs be ON of course to be able to measure any X-Y network beyond the switch.

[radiolistener]

With mains switch ON or OFF ?
It needs be ON of course to be able to measure any X-Y network beyond the switch.

For both states. There is no difference, just no capacitors.

[tautech]

With mains switch ON or OFF ?
It needs be ON of course to be able to measure any X-Y network beyond the switch.

For both states. There is no difference, just no capacitors.

Well there you go then. If this is proved true who, what corporation gets away without a mains suppression network in their products in this day and age ?

[Electro Fan]

One more....

You're doing wrong measurement. Do not touch ON/OFF button. ON/OFF button should be always in OFF state

Use POWER ON/OFF button instead of ON/OFF button!

I did, in each set of photos you can see what happens when the Ouput button is used AND when the Power button is used.

Obviously, when the Power button is first turned on the Output button is in the off state - so that isn't too exciting.  But you can see in the photos when the Power button was turned off with the Output button in the on state.  Let me know if that isn't clear and we can discuss further or I can take some other photos if you would like.   

fwiw, after using a KA3005P for over 7 years and a KA3005D more recently, I'm not ready to say they are going to replace Keysight et al power supplies in commercial labs but the KA's are reasonably accurate, reliable, and enjoyable to operate.  I think it makes more sense to go for the KA's than the KD's for the better UI including the separate Output button.  I think if we listed all the new test equipment that could be purchased new for about $100 the Korad KA3005D would have to be near the top of the list for overall good value.

Below a repost of one of the images showing the Power Button Turning Off with the Output Button On.

[Electro Fan]

Just to put this to bed, here is an image of the Power Button turning on (with the Output Button off).

[ataradov]

Why are you keep looking at 100 ms time scale? The spikes are a few us long.

You need to try harder, they are there. Your unit is not special.

[DaneLaw]

Tried with my KD3005P with one of these chin crocodile to BNC and just single capture while turning on and off with the KD3005P AC switch.
the start is with 1v out and the latter part is with 0v.

[nctnico]

Why are you keep looking at 100 ms time scale? The spikes are a few us long.

And a tell-tale sign that you are looking at an 'EMC' effect. The control loop in a PSU isn't fast enough to produce spikes in the us region (and if you look closer at the screen you'll see the actual pulses are in nanosecond territory). A power supply misbehaving is in the milli-second time scale.

[Electro Fan]

Why are you keep looking at 100 ms time scale? The spikes are a few us long.

ataradov, I chose the timebase setting to show the knees of the curves.  And when I look around at 5us I don't see any spikes.  Not trying to hide anything or prove anything - just showing what I see.  So, I'm not sure what you are looking/asking for.  Please let me know what power supply, load, and scope setup and capture parameters you are looking for and I'm happy to give those a try.

Why are you keep looking at 100 ms time scale? The spikes are a few us long.

And a tell-tale sign that you are looking at an 'EMC' effect. The control loop in a PSU isn't fast enough to produce spikes in the us region (and if you look closer at the screen you'll see the actual pulses are in nanosecond territory). A power supply misbehaving is in the milli-second time scale.

I think this (what nctnico is saying, if I understand correctly), is the case.  We can look at the power supply or we can look at other environmental effects.  Seems like the power supply is performing properly but if I'm not understanding and missing something please let me know.

Edit: ataradov was right, I needed to try harder.  Events found and captured... will post shortly.

[Electro Fan]

Ok, found a couple events when looking at 1us/Div.  This was with the 10x probe at 1Mohm connected to the alligator clips of the banana leads with the KA3005P set for 1V, 1mA.  These were using the Power Button to turn the power supply on, and the Power Button to turn the power supply off (when the Output Button was still on).  Happy to try other combinations now that I'm looking in the right neighborhood. 

I now have various other combinations to try......

[radiolistener]

I chose the timebase setting to show the knees of the curves.  And when I look around at 5us I don't see any spikes.  Not trying to hide anything or prove anything - just showing what I see.

Can you show 1 uS/div scale for POWER-ON sequence?

Also, could you please check if your KORAD has Y-capacitors on the mains?

And a tell-tale sign that you are looking at an 'EMC' effect. The control loop in a PSU isn't fast enough to produce spikes in the us region (and if you look closer at the screen you'll see the actual pulses are in nanosecond territory). A power supply misbehaving is in the milli-second time scale.

PSU noise from a dynamic LED indication can be found in the nanosecond time scale

Well there you go then. If this is proved true who, what corporation gets away without a mains suppression network in their products in this day and age ?

Just inspected inside, there is no Y-capacitors. The mains connector is connected to the transformer through the power switch. Ground is connected to the shell.

I found there are two blue capacitors between ground (shell) and + and - output terminals.

[Electro Fan]

Ok, the plot thickens.

I remembered Dave did a video with a BNC to probe connector so I gave that a try with the same 1V, 1mA settings I had in the post above.

In general, I think the better connection minimized the severity of the spikes.  I ran about a dozen power cycle tests.  Some of the results might vary based on how long I left the power on or off - not sure.  Perhaps most interestingly, when I set the trigger to just above the 1V signal level there were no spikes to trigger the 1V signal. On the other hand, there are still spikes when I turn the Power Button off.  You can see they vary across the three samples attached below - might be random or it might be related to how long (several seconds here or there) between power ons and power offs.

[Electro Fan]

Can you show 1 uS/div scale for POWER-ON sequence?

Also, could you please check if your KORAD has Y-capacitors on the mains?

Looks like you found what you were looking for on the inside... thanks for sharing the findings.

I ran the 1us/Div Power-On sequence.  Nothing happens at Power Button Power On and with the Output Button Power On the signal triggers cleanly (no spikes).

[Electro Fan]

I'd like to think it's persistence more than OCD.

I have a theory that the turn off spikes might be influenced by how the Power Button operates during Power Off (ie, how fast it is making/losing contact inside the switch).  I guess the good news is that if the spikes don't occur at 1) Power Button On, 2) Output Button On, and 3) Output Button Off, then the only trick is to get a DUT disconnected after the Output Button is off and before the Power Button goes off.

[vikisonline]

Well sent back the KD3005D. Got the KA3005D...

Well the KD had about a +/-20mV ripple/noise on the line from the set voltage.

Well looks like the KA I got has +/-100mV from the set voltage.

GAH!

Im tempted to send this back too and get a sigilent. I was really hoping this would behave exactly like the KD but with the soft power button to avoid the crazy spikes.

Though to be honest Im also at the point where Im willing to break down and attempt to live with it. I don't think this is high enough to cause issues with anything I will be doing. Though a far cry from the KD unit I had,
and a very far cry from the manufacturer claimed 2mV....
this was at 2v and 0.143 amps drawn

As expected no spikes with the soft button.

[radiolistener]

Well looks like the KA I got has +/-100mV from the set voltage.

this is strange. Most of all this is error of your DMM.

My KORAD KA3005D has error about 0.01 V or even better, depends on the voltage.

KORAD KA3005D has a calibration mode, so you can calibrate it, if you have precise DMM.

[ataradov]

100 mV is a peak to peak noise. I checked my unit and there is a periodic ringing as shown on the second capture. It is about the same amplitude on open terminals, and goes down a bit with load, but still present at loads of ~200 mA. I don't have anything to quickly create higher load.

[radiolistener]

my KA3005D also have some noise pulses, most of all it come from a dynamic LED indication or a fan PWM. But their average power is very low. And actually KA3005D noise power below 1 MHz is even better than complete linear PSU HY1505D.

Since KA3005D noise power is very low, I don't catch any issue with it. It's even hard to catch this noise with a receiver.


For example here is comparison of the noise between KA3005D and HY1505D taken with a radio receiver.
You can see PSU power-on sequence on the screenshots.

As you can see, KA3005D has some noise peak at about 600 Hz, but the power of this noise is below -80 dB and even better than 50 Hz leak (mains). Which is better than fully linear PSU HY1505D. Also overal noise floor for KA3005D is much better.

If you're not obsessed with extreme ultra low noise, it doesn't affect you. For ultra low noise circuit you can buy another fully linear PSU without digital circuits inside (such as digital control or digital display).

[Electro Fan]

Maybe try with 20M bandwidth limit - I think ripple specs generally assume a bandwidth limit.

If I set the Korad for 2V and up to 4mA but with no load I get about about 2mVRMS.

I think to measure ripple you have to have a pretty good probe technique/connection and you you need to be sure you aren't in a noisy environment.