Lab Power Supply Turn ON and OFF Characteristics

[Martin72]

Hi,

8V 50 Ohm 40mA

What signal do we see on your images when the channel is set to 500 mV/div?

[BillyO]

Someone asked about that little PS from everyone's most hated company?

Well, here it is.  The FNIRSI DPS-150.  Sorry to disappoint everyone but, it's not truly awful .. although better that supplies shown here that cost 10 times as much (or more?), I'm sure some will find something about it that just makes it the worst power supply in the planet.

[KungFuJosh]

Someone asked about that little PS from everyone's most hated company?

Well, here it is.  The FNIRSI DPS-150.  Sorry to disappoint everyone but, it's not truly awful .. although better that supplies shown here that cost 10 times as much (or more?), I'm sure some will find something about it that just makes it the worst power supply in the planet.

Can you run the test in the lower range? Like a 10mV or 30mV test? And please turn on the grid labels for your screenshot.

[BillyO]

Someone asked about that little PS from everyone's most hated company?

Well, here it is.  The FNIRSI DPS-150.  Sorry to disappoint everyone but, it's not truly awful .. although better that supplies shown here that cost 10 times as much (or more?), I'm sure some will find something about it that just makes it the worst power supply in the planet.

Can you run the test in the lower range? Like a 10mV or 30mV test? And please turn on the grid labels for your screenshot.

I'm not sure what you are asking for Josh.

Do you want me to use 10mV instead of 8V?

I put cursors on the over shoot.  Peak of 0.39V trailing away in 55ms

[Bernd_2]

Hello Martin,

Thank you for pointing out the adjustability of the slew rate on the UDP6932B. The device has three settings: Normal, CV-Slew Rate, max. 124 V/s, and CC-Slew Rate, max. 21 A/s.

The measurements in Post #97 were taken with the CC-Slew Rate setting at 21 A/s.

Today, I repeated the measurements with the Normal and CV-SlewRate 124 V/s settings.

While the result for Normal is the same as for CC-SlewRate 21 A/s, there is a significant overshoot with the CV-SlewRate 124 V/s setting.

Voltage and current rise to approx. 4 Volts and approx. 84 mA within approx. 50 ms and then fall to the set limit value within approx. 450 ms.

Best Regards
Bernd

[KungFuJosh]

I'm not sure what you are asking for Josh.

Do you want me to use 10mV instead of 8V?

I put cursors on the over shoot.  Peak of 0.39V trailing away in 55ms

I'm curious how it looks in the lower range. See my attached SPD3303X as an example.

[BillyO]

I'm still not getting it.  Maybe I banged my head    Your vertical volts is at 400mV/div, mine was at 500mV/div.  Not much difference there.

Also, the scope I'm using is the one in my office.  It's only a SDS1204X-E, so I can't label the grid.

Edit:  What's with those trace labels?

[KungFuJosh]

I think they're self-explanatory.

[Helix70]

Hi,

8V 50 Ohm 40mA

What signal do we see on your images when the channel is set to 500 mV/div?

What do you mean? The Oscilloscope channel is 500mV/division already.

The 8V is the voltage set point on the power supply, 40mA is the current limit setting, and 50 Ohms is the load resistor.

Then I enable the output and you see the voltage output for the off->on transition, and then the voltage fall for the on->off transition.

[Martin72]

In each of your pictures, a maximum of 2V can be seen on channel 1. Is that the output voltage of your power supply?
Or is that the voltage across a (fairly high-impedance) shunt?

[BillyO]

In each of your pictures, a maximum of 2V can be seen on channel 1. Is that the output voltage of your power supply?
Or is that the voltage across a (fairly high-impedance) shunt?

Who are you asking?

[Martin72]

If I don't quote anyone, I usually respond directly to the previous post.

[BillyO]

Oh, okay, understood.

Don't mean to answer for him, but I assume he is using the set-up described in the first post.  Set the supply to 8V & 40ma current limit.  Attach a 50ohm resistor, then put your scope across the resistor.  Turn on (and off) the power supply and record the voltage over time.

I guess the idea is, if the supply is slow in getting the current limit engaged you will get a spike .. up to 8V, then it should settle out to 2V (40ma through 50ohm).

If it does get the current limit in place properly, you will see the voltage rise to 2V and stay there.

And everything in between.  Like the little .39V overshoot of teh FNIRSI.

[Helix70]

Yes, BillyO is correct. The images show the voltage across the 50 ohm shunt, which is being limited to 2V due to the 40mA current limit set on the power supply and 50 Ohm load. No overshoot - good!

Regards,

Josh

[BennoG]

TTi MX100QP should be a reputable brand.
But the overshoot is enormous. It goes to 8V for about 15-20 ms and then drops down.
Also dit a test on lower resistors (0.5 and 0.2 ohm) and on turn on it could go up to 15A for 1 or 2 ms (current limit still 40mA 8V)

This is positive if you are testing devices that have a huge capacitor at the input.
But is negative if you are testing a MCU that has a short somewhere.
Even when testing a LED you can blow up the LED with this.

edit added the turn off

Benno

[jusaca]

I think there was no status report for the R&S case since last october in this thread. Was there any real feedback from them regarding the overshoot?

[Bernd_2]

Hello,

Here is an addition to the measurements on the UDP6932B, posts #97 and #104. To see if the overshoot behavior changes at higher currents, I took three measurements with a current limit of 1 A. All with CV slew rate 124 V/S.

Figures 1 and 2 show the measurement with 8 V voltage and 1 A current limit at 5 ohms. As you can see, the current only overshoots by about 30% and then settles at 1 A after 120 ms.

Figures 3 and 4 show the curve with a voltage limit of 60 V and a current limit of 1 A at 5 ohms. Now the current shoots up by 37% and falls to the set value of 1 A after 110 ms.

Figures 5 and 6 show the measurement with a voltage limit of 60 V and a current limit of 1 A at 50 ohms. Now the current no longer shoots up and only reaches the set value of 1 A after approx. 550 ms.

Best Regards
Bernd

[Mahagam]

Siglent SPD1168X. Not bad.
No OFF plot, because it looks like standard discharge RC = 49.48Ω * 500μF

[David Hess]

I think there was no status report for the R&S case since last october in this thread. Was there any real feedback from them regarding the overshoot?

Do you mean the R&S NGMO?

I wonder if there is something in common for all of these supplies with such poor overshoot.  Do they use a digital feedback loop?  Properly implemented a digital feedback loop should be no worse, and likely better than an analog one.

It would be useful to know how much output capacitance they have.

[Thomas]

TTi EL302RT and GWInstek GPP-4323.
I didn't record channel 2 on the GPP-4323 since it is identical to channel 1.

[d-el]

DIY power supplies are valid?
Quick specs: 1 channel, 37V, 4A, LAN interface, weight less than 4kg.

Precondition:
DSO: LeCroy WR104MXi, BWL 20MHz, Coupling DC50Ω, 500mV/Div
PS: 8V, 40mA

ON:

OFF:


PS:

[BillyO]

This is a cheapie I got for the 20A capability which I rarely need, but it's turned out to be a pretty nice supply.

Bside I84020 300W SMPS.  Got it for $100 on AliExpress.  Could not get exactly 40mA as the adjustment resolution is 10mA.  Looks like it might be a little high.  Over shoot is about .1V

[Martin72]

If this thread is intended to show the settling time of the respective power supplies during loading and unloading, then your last image does not show any regulation, as it is simply a capacitor that is apparently located directly at the output after the (electrical) output switch that is discharging.

[BillyO]

No argument here.  At least it does not spike.

[David Hess]

In a lot of these cases it seems they added enough output capacitance to make up for poor transient response from poor frequency compensation.

Usually only about 50 microfarads per amp is required, but variation from half to twice that is common.  With care and good design, a good low capacitance design is possible with a fraction of a microfarad per amp.