Best Output ripple measurment for switcher

hi,

which one is preferred method on measuring output ripple fo switchers, output capacitor probed with a 50ohm coaxial terminated at the scope of 50ohm or using with 10:1 oscilloscope probe with 1M input termination(tip and barrel method)?

I saw one article that measuring outout ripple should be 50 ohm enviroment...Please give the best reason thanks
http://cds.linear.com/docs/Application Note/an70.pdf

this is from Jim Williams

The 50 ohm is better because the scope probe's high impedance makes it an antenna and it picks up radiated noise as well as voltage ripple. If the 50 ohm termination is too much load for power supply then the probe can be decoupled with a large ceramic or mica capacitor in series. However... the "damper" effect of the 50 ohm resistor can increase the power supply's stability, so I actually have both probes on when I do a measurement, so that I can see the effect of the 50 ohm damper. As long as the waveform doesn't change substantially, I assume that the damper is having little effect on the system.

Linear tech definitely knows what they're doing in this regard. I've used their measurement techniques for a while, and have observed vast improvements in measurements when using terminated coaxial cables instead of probes. When "probing" DC signals make sure to have a blocking capacitor as well, and use a 50ohm terminator on the scope side of things.

thanks guys. I also used that coax with 50 ohm input termination at scope end(of course AC coupled using an external DC blocker) and it works well...I see that It doesnt load or affect the switcher load in AC terms since I'm measuring on the output capacitor for the output ripple. because the reactance of the output capacitor is small (uF range and switching freq is above Mhz; Xc = 1/2pifc) compared with the 50 ohm termination. Thats why i think its ok!!Even if you have a large load or small load, the output impedance will be low due to output capacitor. That is considered with higher frequencies,but for Low frequencies or 0 frequency(DC), the DC blocker will eliminate the 50ohm effect specifically in DC load current.So I guess the loading are eliminated and you can use the advantage to minimize any reflections and standing waves at higher frequencies because of a well 50 ohm terminated environment

---------- Post added at 23:26 ---------- Previous post was at 23:22 ----------

please if you there is anyone who does not agree with my reasoning, feel free to correct it...

One comment about the damping effect of the 50 ohm load.

At no load the damper provides a method of slightly discharging the output capacitance (divided by Cdamper//Cout) but at least it knocks down the unidirectional effect of the power supply so the integral term in the control loop doesn't wrap too far.

calling all experts , anymore critics about the 50 ohm? is output ripple more accurate if it was measured at 1:1 attenuation with 50 ohm cable terminated at 50 ohm input scope or standr probe with 10:1 attenuation termonated at 1M input scope???thanks!!!!

The 50 ohm cable and termination will give best results for measuring ripple. You may want to AC couple from the output so that you don't affect the damping of the supply, and so you don't damage the 50 ohm terminator (in case the output DC is high). Just make sure the cutoff frequency of the AC coupling is well below the switching frequency but at least as high as the converter's crossover frequency.

so it is AC coupled at the output of the DUT..Should there be a dummy series resistor such as 50 ohms before the 50 ohm cable for proper termination??

No, it doesn't have to be impedance matched on the DUT side, just on the scope side.

what is the purpose of the 50 ohm? is still there a impedance mismatched if there are no source termination??

Having an impedance match on the load (scope) side of the cable ensures that the source (DUT) side sees a broadband 50ohms at its output, and that there are no reflections off the load. you don't to be impedance matched on the DUT side because there is no signal source on the load side, and thus there should never be any waves traveling towards the DUT from the scope.

thanks..Which is more ideal 1:1 or 10:1 probe?for measuring ripple and noise