Ringing on output

Posted by: thoru on 12/19/2008

Hi,

I've designed a power supply based on the TOP245. I've a problem with some ringing on the output. I first thought that it was the capacitance of the output diode and the leakage inductance of the transformer. I tried to add a snubber, but adding a capacitor didn't change the ringing frequency, so I'm in doubt whether it actually is the output diode.
Where else could the ringing come from? It occurs just before the voltage over the drain of the TOP245 reaches the reflected output voltage (in the downgoing slope of the initial voltage spike). The drain voltage itself shows no sign of ringing though.
Please find attached a schematic of the current design.

Thanks in advance,

Remco Poelstra

Files

PSU_ringing.pdf (78.98 KB)

Comments

Is this being measured after the linear regulator or before it? I would remove the linear regulator and measure the output from the transformer to simplify the circuit and better identify the potential source of the problem. If the ringing persists with the linear regulator removed please please provide a voltage waveform of the ringing you're talking about. What is the frequency of the ringing? The magnitude? Thanks!

Hi, Thanks for your reply. I removed the regulator and associated components and measured at the output of the CLC filter. Attached is a drawing of the waveforms, top shows the source-drain voltage of the TOP245, bottom is the output. At the moment I'm concentrating on the 90MHz ringing, I hope to tackle the rest later. Kind regards, Remco Poelstra

Hi,

I believe that it is the values for the post filter you have chosen are causing the problem. By choosing 10uH inductor and 1000uF capacitor for the LC filter the cut off frequency of the LC filter is very low. This can cause oscillations because you are taking feedback from before the LC filter, which should be a stable system, then summing that with the LC filter which is a stable system. The sum of two stable systems does not necessarily equal a stable system.

Try shorting out the inductor and then measuring the output voltage. If you do not see the ringing with the inductor shorted out, try choosing a smaller value inductor and smaller value of capacitor C4.

Also if you're using a linear regulator an LC filter shouldn't be necessary anyway.

If you use a TL431 voltage reference with 1% tolerance feedback resistors you can get as good as 5% regulation. Perhaps this could also eliminate the need for the linear regulator, improving efficiency, reducing cost and simplifying the circuit.

What is your efficiency and output voltage tolerance specification?

Good luck!

-PI-Wesley

Hi, Unfortunatly, removing the inductor did not help. Placing a snubber over diode D4, reduced the high frequency ringing to within 100mVpp. I think that this is as good as it gets when it comes to the design. So I was wondering whether it's possible to remove the last noise with a better filter? The LCL filter works actually quite nice, but only for the 132kHz rimple, the high frequency noise gets straight through. The same is true for the input filter, where none of the high frequency noise gets reduced. How can I tackle that? I tried a 90Ohm@100MHz bead instead of the output inductor, but it gave similar results to the inductor. As per the specifications, the output voltage should be +-100mV the nominal value. Efficiency is not much a concern.

Hello Remco, as PI-Wesley is on vacation I thought I'd take a look at this. Can you let us know how you are making this measurement. I'm wondering if this is just common mode noise pick up. Most large customers specify a ripple probe that consists of a 0.1 uF ceramic in parallel with a 100 uF tantalum or low ESR electrolytic to measure ripple. Periodic and random deviation (PARD) is usually limited to a 20 MHz bandwidth. Attached is an old document showing the impact of loop area on the measurement and the arrangement we currently use to measure output ripple and noise. Cheers PI-Chekov

Wonderful! Even without the capacitors on the modified probe the output looks clean indeed. But that also suggests that in the final application the surrounding electronics might pick up unnecessary noise also, so shielding the PSU might be a good option. How should I connect the shield to the reference? I mean, I've a primary reference and a secondary one, which one should I use? Since the application is powered by a two-wire 100V line, I do not have an extra earth connection to connect the shield to. Do you have any comments on that? Thanks very much. Remco Poelstra