transcendent
Newbie level 4
I've recently designed and built a multi SMPS board (COTS buck boost which then supplies a 5V buck and a 20V boost rails)
I'm currently doing some testing and have been reading down a rabbit hole of best practice measurement with oscilloscopes, and also on conducted EMI and standards such as CISPR.
So far, I have been doing some rudimentary measurements using a 10:1 passive probe and a spring clip/ground blade to cut out noise from the measurement, limited bandwidth of the oscilloscope to 20 MHz etc, I understand it should be better with 50 ohm coax and a blocking capacitor but I haven't yet got around to that.
My main problem is I'm getting a bit confused with how multiple measurements relate to one another.
a) Is voltage ripple equivalent to conducted EMI?
On the input to my cots converter, i'm measuring about 150mV peak to peak ripple voltage at the 250 kHz switching frequency with a max load. Looking at CISPR 22, it specifies conducted emissions of about 60dBuV at that frequency.
Converting that gives me .... 1mV ?!? How is that possible? Looking at acceptable ripple recommendations gives values of about 1% for output voltages, do these apply for input voltage or output voltages or both?! Am I going to be that reliant on using LC filters on to damp by 20dB +
Also, to what extent does ripple propagate from one switching converter to the next? Is it necessary to EMI filter between stages on the PCB or just at the input and output?
b) I see plenty of mention of using a LISN to the input of the DC converter to isolate the noise from the converter from whatever is powering it. Is measuring using a scope probe giving a roughly representative result for differential noise, say if i used a low noise power source such as a battery with short twisted leads?
I'm currently doing some testing and have been reading down a rabbit hole of best practice measurement with oscilloscopes, and also on conducted EMI and standards such as CISPR.
So far, I have been doing some rudimentary measurements using a 10:1 passive probe and a spring clip/ground blade to cut out noise from the measurement, limited bandwidth of the oscilloscope to 20 MHz etc, I understand it should be better with 50 ohm coax and a blocking capacitor but I haven't yet got around to that.
My main problem is I'm getting a bit confused with how multiple measurements relate to one another.
a) Is voltage ripple equivalent to conducted EMI?
On the input to my cots converter, i'm measuring about 150mV peak to peak ripple voltage at the 250 kHz switching frequency with a max load. Looking at CISPR 22, it specifies conducted emissions of about 60dBuV at that frequency.
Converting that gives me .... 1mV ?!? How is that possible? Looking at acceptable ripple recommendations gives values of about 1% for output voltages, do these apply for input voltage or output voltages or both?! Am I going to be that reliant on using LC filters on to damp by 20dB +
Also, to what extent does ripple propagate from one switching converter to the next? Is it necessary to EMI filter between stages on the PCB or just at the input and output?
b) I see plenty of mention of using a LISN to the input of the DC converter to isolate the noise from the converter from whatever is powering it. Is measuring using a scope probe giving a roughly representative result for differential noise, say if i used a low noise power source such as a battery with short twisted leads?
