Power Supply Slow rise time and Ringing

[FreshmanNewbie]

I have this Power Supply EA-PSI 9200-15T.

My input voltage to a board is 28.8V. No Load Current is drawn from the power supply.

1. 

With reference to the above diagram, If I connect the board to the power supply and turn the board ON directly, I get this waveform :



The rise time of the power supply from 0V to 28.8V is around 10ms here.

1. 

But now, If I connect a toggle switch between the power supply and the board. And I keep the switch in OFF position first. Now, I turn ON the power supply and then I turn ON the toggle switch as shown above, I get this waveform.



Ignoring the ringing, the rise time of the power supply from 0V to 28.8V is 20us.

My questions :

1. Why does the rise time vary when I connect and turn ON Power Supply directly and when I connect and turn ON using a toggle switch?
2. Why is there ringing in the supply when I connect it through a toggle switch?

Can someone please help with these questions.

 

[danadakk]

If your supply is running, then you switch on load, you will see inrush
current and any L in traces and supply load changes to its L will show
up. Whereas you turn on supply with load attached the supply control
loop is managing the ramp up and quite a lot of C charging at supply
end is ramping up. Also keep in mind you have a supply with its own
control loop, how was that designed ? Does the control loop, on startup,
go into current limiting, thereby causing a current source ramp up at the
bulk caps in it.

This all assume you are probing with probe ground attached to load side,
and your probe is comped properly, and ground lead short.

Also transient quite large for a power supply in regulation it seems. Do you
have long lead length between load and supply. Lots of L in the path ?


Regards, Dana.

 

[KlausST]

Hi,

in other words:
Go to the kitchen to an eletric oven. OFF, cold.
Touch the plate.
Now switch ON the plate and measure the time how long it takes until it hurts.
put your finger from the plate.
(This is the situation when the power supply is OFF and you switch ON the mains of the supply)


Now go to the electric oven.
Switch it ON and wait 2 minuts.
Then touch the plate and measure the time until it hurts.
(This is the situation when the powersupply was already ON when close the switch to the load)
You get shocked and most quickly you move the finger from the plate.
(This is the the situation of the ringing signal)

Klaus

 

[Easy peasy]

the switch provides a faster rise time - the inductance of the source ( and inside the power supply ) along with capacitance in the power supply forms an LC resonant circuit that is excited by faster rise times - hence the overshoot and ringing - this is a common phenomena - but not commonly known outside of power electronics engineers ...

 

[Akanimo]

I have this Power Supply EA-PSI 9200-15T.

My input voltage to a board is 28.8V. No Load Current is drawn from the power supply.

1. 

With reference to the above diagram, If I connect the board to the power supply and turn the board ON directly, I get this waveform :



The rise time of the power supply from 0V to 28.8V is around 10ms here.

1. 

But now, If I connect a toggle switch between the power supply and the board. And I keep the switch in OFF position first. Now, I turn ON the power supply and then I turn ON the toggle switch as shown above, I get this waveform.



Ignoring the ringing, the rise time of the power supply from 0V to 28.8V is 20us.

My questions :

1. Why does the rise time vary when I connect and turn ON Power Supply directly and when I connect and turn ON using a toggle switch?
2. Why is there ringing in the supply when I connect it through a toggle switch?

Can someone please help with these questions.

It seems like the issue is with the toggle switch. It seems to be an issue of switch bouncing. Try connecting an RC snubber across the switch to damp the oscillations.

 

[FvM]

There's no indication of switch bouncing in the waveform, it shows ordinary ringing of a LC resonant circuit as explained by Easy peasy. The behaviour could be analyzed in detail if we know the amount of board power supply bypass capacitance, cable and possible dedicated filter inductance.

Switching a capacitive load with a mechanical switch involves a risk of fastly wearing the contact.

 

[Akanimo]

Well, we can never be so sure. More than one factor can contribute. It even maybe a factor entirely different from what you think. That's why it is troubleshooting.

So what leads to the firm conclusion that the problem is caused by one factor or another?

If not that which Easy Least has mentioned, what else could it be?

My thoughts:
When the connection is without the switch, at power supply turn on, the output cap of the power supply has to charge from 0V to 28.8V. This takes time. I believe the rise time is a result of soft start (and some additional board input cap, maybe).

When the switch is in place and open, and when the power supply is then turned on, soft-start gets to work and the output gets to 28.8V after some time. The power supply continues to run on no load until the switch is now closed. At this time, the power supply is seeing a step change in load and control is solely the responsibility of the control loop.

Considering the scenario and waveform presented, there is a possibility that the control loop may have a high Q-factor. Also, the switch bouncing that I talked about is not entirely ruled out because as the switch is turned on, the input cap of the board gets charged very quickly. Even if the switch bounces, the voltage may not get to 0V until the load takes out all the energy availble in the input cap of the board. So I wouldn't be expecting the switch bounce waveform that we know because of the input cap.

Thus, one can never be so sure that this is not the problem until that factor is ruled out. So how can this be ruled out?
--Turn of the power supply.
--close the toggle switch.
--turn on the power supply and observe the transient response.

If after these the problem is still there, then it's not the switch. So we can think of what it might be.

At that time we can think of testing the control loop. We can do such testing by connecting a purely resistive load to the power supply through the switch and turning the switch ON and OFF while the power supply is already up and running to observe the transient response.

These are non demanding tests that can be done to keep eliminating possible causes of the problem until the root cause is found.

 

[KlausST]

Hi,

I see no switch bouncing. Switch bouncing may be in the milliseconds range and will cause multiple ringing waveformes in series.

I also don't think it's the SMPS regulation loop. Let's guess there is an output capacitor of just 10uF (usually it's much higher), then the first fall from almost 50V down to almost 15V within about 1us needs a current of 350A (if my mind calculation is correct). I know no SMPS that can actively pull down output voltage (but I'm sure there are some) ...and surely not with that current.

I see a clean, damped LC ringing. Uniform, first peak is about double the DC voltage.

Klaus