SEPIC capacitor overvoltage?

[treez]

Hello,

Please assist in measuring short term overvoltages on my SEPIC capacitor......

SEPIC capacitor is 22uF, X7R, 1812, 16V

I have done a SEPIC converter.
Max output power = 5W.
Input voltage = 5 to 8.5V
Output voltage = 5V or 40V (depending on luminaire)
Switching frequency = 40KHz to 110KHz
Constant off time control

The SEPIC coupled inductor is coilcraft MSD1583-473
**broken link removed**

When i looked for overvoltages across the sepic capacitor on the Scope (200MHz bandwidth scope) , i merely placed the scope probe across the SEPIC capacitor.

However, the scope probe ground will have an associated capacitance to the ground, and so i fear that i may have unwittingly filtered any overvoltage spikes away?
-Must i use a differential probe to look for the overvoltages on the sepic capacitor?
I am wondering if there will be overvoltages at start up, due to the ringing with the leakage inductance of the coupled SEPIC inductor?
Also, as the FET switches, i fear that there may be overvoltages on the SEPIC capacitor at these intervals also?

Do i need the differential probe?, and what bandwidth Diff probe must i purchase?

 

[FvM]

The LC series resonance of 22 uF capacitor and 0.46 µH leakage inductance calculates as about 50 kHz. It shouldn't be difficult to observe the peak voltage at this frequency. Ringing at higher frequency can be expected to show measurement artefacts respectively voltage across other resonant structures, but no actual capacitor voltage.

 

[mtwieg]

When i looked for overvoltages across the sepic capacitor on the Scope (200MHz bandwidth scope) , i merely placed the scope probe across the SEPIC capacitor.

However, the scope probe ground will have an associated capacitance to the ground, and so i fear that i may have unwittingly filtered any overvoltage spikes away?

I highly doubt 10pF of stray capacitance is going to distort a waveform on a 22uF capacitor. But I don't recommend placing a grounded scope probe directly across the capacitor. Use two probes on either side of the cap, with their grounds connected to circuit ground, and use a scope's math function to get a differential measurement. Or use a diff probe if you have it. 100MHz should be plenty of bandwidth.

 

[treez]

a 16V cap will not withstand a 40V output off of a 5-8.5V input

...for a sepic with a coupled inductor, the maximum voltage that the sepic capacitor will see is the input voltage.

I don't recommend placing a grounded scope probe directly across the capacitor

....may i ask why?
(i appreciate what you are saying, because i have heard this before, but do you know the reason?)
The trick with two probes , i have found, is very noisy, and doesnt work very well.

 

[mtwieg]

...for a sepic with a coupled inductor, the maximum voltage that the sepic capacitor will see is the input voltage.

Whoops you're right, I was thinking of Cuk topology.

....may i ask why?
(i appreciate what you are saying, because i have heard this before, but do you know the reason?)

Using one probe works in theory when your circuit is perfectly isolated from earth ground, but that's never really the case. If you connect a switching node to earth, you'll usually cause significant CM currents to flow on the shields of the probe cables, which can cause problems. Using large ferrite beads on the cables can help a bit though.

The trick with two probes , i have found, is very noisy, and doesnt work very well.

It should work fine. If it doesn't, then that means that the individual probe measurements are "noisy" and that in itself is a big problem. Make sure the two probes have the exact same ground connection.

 

[FvM]

It should work fine. If it doesn't, then that means that the individual probe measurements are "noisy" and that in itself is a big problem. Make sure the two probes have the exact same ground connection.

Yes. An active differential proble isn't but a dual compensated voltage divider with a differential amplifier. It offers maybe factor 10 higher common mode rejection than individual probes with Y1-Y2 oscilloscope setting, but still picks up some interferences. High voltage differential probes (e.g. 700 or 1400 V CM and DM range) are a standard solution in power electronics, but not necessarily required for low voltage measurements like yours. Of course, if you have it, you'll most likely use it for other measurements, too.

 

[D.A.(Tony)Stewart]

Active FET buffered Diff probes are ideal ( but also very sensitive to ESD > 25V !) , but you can achieve adequate measurement here using two identical probes and identical short ground leads. < 3".

Connecting both probes to the same point of a dynamic waveform should null out on Ch1-2. If not then twist/wrap both probe cables together & repeat.
Calibrate probe with square wave (ac) or input channel gain if neces. (for dc).